The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Anthropic models will not be affected.

We received the directive from the government today at 5:21pm (ET). The letter did not provide specific details of its national security concern. Our understanding is that the government believes it has become aware of a method of bypassing, or “jailbreaking” Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass.

It’s been barely a few days since the nerfed version of Mythos was launched under the new product name Fable. The early reactions have been positive (including my own), and we can only imagine that ARR has skyrocketed, as users started pushing their workloads at 2x the price of Opus tokens.

Most of Friday was spent on the SpaceX IPO, which attracted massive amounts of capital relative to its current revenue. For many, the more interesting moment would be seeing Anthropic go public. Alas, at precisely 5:21pm ET, we hit a fork in the road.

Anthropic’s posture with respect to Fable’s safeguards, as laid out in our launch blog post, is the following:

• We have instituted strong safeguards that greatly reduce the likelihood that Fable is misused for tasks related to cybersecurity (among others). In fact, our safeguards are so strong that many users have complained that they are overly broad.

• In the weeks leading up to the launch of Fable, Anthropic worked with the US government, the UK AISI, multiple private third-party organizations and internal teams to red-team Fable’s safeguards for thousands of hours in total.

• These tests showed that Fable’s safeguards are substantially more effective than those of any previously deployed model.

• No testers have yet been able to find a universal jailbreak—a jailbreak method that can very broadly bypass the model’s safeguards, unblocking a wide range of cyber capabilities.

• We suspect that perfect jailbreak resistance is not currently possible for any model provider. Every safeguard used in the industry is vulnerable to non-universal jailbreaks (which can elicit some cyber information in specific circumstances), and it is likely that universal jailbreaks will eventually be found in the future. We stated this clearly when we released Fable 5.

• Given that perfect jailbreak resistance does not appear to be possible today, Anthropic adopted a defense in depth strategy with Fable 5. We aimed to make jailbreaks either narrow (in the case of non-universal jailbreaks) or very expensive to produce (in the case of universal jailbreaks), and to combine this with thorough monitoring to quickly detect and shut down any successful attacks. This is also why Anthropic has required 30-day retention of customer data with Fable—a policy change that carries real costs for us with customers, but that allows us to research and mitigate jailbreaks.

• We stand by this defense in depth strategy. It reduces the risks posed by Fable, making them comparable to the risks of existing models already deployed across the industry.

• We have not even received a disclosure of a concerning non-universal potential jailbreak that led to a harmful result. The potential jailbreaks that have been disclosed to us are either entirely benign responses or are minor findings that provide no Mythos-specific uplift.

To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws. Our understanding is that one potential jailbreak was shared with the government. We have reviewed a report that we believe is the basis of the government’s directive and validated that the level of capability displayed there is widely available from other models (including OpenAI’s GPT-5.5), and is used every day by the defenders who keep systems safe. We will share more details over the next 24 hours.

According to Anthropic, the administration claimed that a jailbreak was being utilized in the wild, but that this did not constitute a serious enough reason to stop offering Fable to the public.

Subscribe now

This is a bit ironic, since barely a few days earlier, Anthropic published a deep dive on recursive self-improvement of AI that ended with this call to action:

If it were possible to effectively slow the development of this technology to give ourselves more time to deal with its immense implications, we think that would likely be a good thing. But if a slowdown simply lets the least cautious actors catch up technologically, it could leave everyone less safe. Without a global coordination mechanism, companies and governments will have to make difficult decisions about safety while under competitive and geopolitical pressures.

We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology. The Anthropic Institute will conduct research—in collaboration with many others—and take actions to help build the systems that a credible slowdown or pause would require. These systems would enable frontier AI developers to verify that others globally have actually stopped or slowed, and that a bad actor could not use the auspices of a coordinated slowdown to jump ahead in secret. If such systems existed, we expect that we would slow down or temporarily pause, if other developers at or near the frontier also did so in a verifiable manner.

A meaningful slowdown or pause would require multiple well-resourced labs at or near the frontier, in multiple countries, agreeing to stop under the same conditions. It would also require that each can verify that the others have actually stopped. Due to the unique characteristics of AI systems, the detectability (a lower standard than verifiability) element of this arms control problem is much more challenging than with other technologies. Training runs are far easier to conceal than missile silos, their inputs are general-purpose, and the incentive to defect quietly is enormous, because whoever continues while others pause could inherit the lead. A credible pause also has to specify what triggers it, what lifts it, and who adjudicates.

None of this is necessarily impossible in principle—the world has built verification regimes for other complex technologies (e.g., the Intermediate-Range Nuclear Forces Treaty)—but those regimes took decades to build both the infrastructure and the trust. We don’t have that long. A unilateral pause by one lab, by contrast, is achievable immediately, but accomplishes much less: it would change who the front-runner is, but it would not create the wider deliberative process that is currently missing.

In the coming months, we will organize conversations where policymakers, researchers, civil society, and other AI companies can help answer some of the questions this piece raises, especially around full recursive self-improvement and how to create better options for coordination and deliberation. We’ll publish what comes out of it. The window to investigate the questions together is here, and people outside AI companies should be involved in this deliberation.

So who was the main player behind the disclosure of this potential security risk? The Information reports the following:

Amazon CEO Andy Jassy was among the tech leaders who raised concerns to senior Trump administration officials this week about security risks in Anthropic’s most advanced models, according to two people familiar with the conversations.

The calls between the head of Amazon—one of Anthropic’s biggest investors and vendors—and the officials in the last few days helped set in motion the Trump administration’s new export restrictions on Anthropic’s Claude Mythos 5 and Fable 5 models late Friday night, citing national security concerns, the people said. Those restrictions suspend access to those models to foreign nationals. Anthropic said it disabled access to the models for all customers to comply.

An Amazon spokesperson told The Information: “As a leading cloud provider that serves a large number of private and public sector customers, it’s not uncommon for governments to seek our counsel on potential security risks. When they occur, we don’t share the details of these discussions.”

The restrictions are the latest hit to Anthropic in its ongoing standoff with the Trump administration. The relationship between the Claude maker and the government has devolved in recent months following the Pentagon’s move in March to designate the company’s model as a supply chain risk. That move came after the two disagreed on whether Anthropic’s models could be used in cases like mass domestic surveillance or with lethal autonomous weapons.

Amazon's security team reported a serious concern related to jailbreaking the safeguards as implemented on Fable. Given the significant financial interest that Amazon has in Anthropic, logically speaking they wouldn't have pushed in such a direction without at least something material behind it. The other major issue raised was that China-backed groups had direct access to Mythos (i.e. the non-safeguarded model). The government's position was mostly outlined by David Sacks:

I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true:

— As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable.

— Fable is Mythos with guardrails. But if those guardrails fail, then you’ve exposed Mythos and its advanced cyber capabilities to people who shouldn’t have them. (Keep in mind that Anthropic itself widely promoted the idea that Mythos was a cyberweapon and needed to be regulated as such. They asked for government regulation of Mythos and championed the guardrails on Fable. If there is a vulnerability — big or small — it is Anthropic’s responsibility to patch.)

— A highly credible trusted partner of both Anthropic and the USG who was testing Fable came forward with a jailbreak of those guardrails. The Admin asked Dario to fix the jailbreak or de-deploy the model. Dario refused.

— In their blog post, Anthropic defended its decision by saying the jailbreak isn’t serious. That is not what the trusted partner and the USG believe; nor is that kind of minimizing language consistent with Anthropic’s brand as the AI safety company. It’s difficult to fathom how they could claim a jailbreak allowing operability of a cyber weapon could be defined as not “serious.”

— In the past, Anthropic has always said that safety must be top priority and taken super seriously. In this case, Anthropic prioritized the continued offering of the consumer model over safety.

— In reaction, the Admin issued the export control. The Admin did this reluctantly. It’s been very surprised that Anthropic hasn’t wanted to cooperate with a reasonable safety request (ie fixing the jailbreak issue). Anthropic’s reaction is very much at odds with their branding and ethos as a safe AI research community.

— The Admin’s hope now is that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general release. The Admin wants all of this to happen as soon as possible. It is frankly bewildered that Anthropic hasn’t wanted to comply with safety requests that it previously said were its highest priority.

— Those trying to misdirect and tie this action to the prior DoW/Anthropic issues are wrong. The Admin values Anthropic’s technical capabilities and feels that this issue, while serious, should be easily resolved. The ball is in Anthropic’s court.

It’s been barely 3 months since Anthropic’s last spat with the government, which ended with the company being kicked out of all Federal contracts. Rather than trying to recover and move on, farming its considerable fortune of rapid Enterprise adoption, the leadership team has yet again decided to pit rhetoric and stubbornness against the juridical and enforcement power of the government. This is what Politico reported on the actual calls that took place:

The Trump administration’s decision to impose sweeping export controls on Anthropic followed a frantic 24-hour effort by senior officials to convince the company to voluntarily pull a newly released artificial intelligence model that officials believed posed security risks, according to two administration officials and a senior White House official, who like others in this story were granted anonymity because they were not authorized to discuss the episode.

The move, which followed multiple tense calls between Anthropic CEO Dario Amodei and administration officials, including Treasury Secretary Scott Bessent and White House Cyber Director Sean Cairncross, underscores how the White House is wrestling in real-time with regulating fast-moving and potentially dangerous AI models.

The administration’s imposition of export controls forced Anthropic to pull its new AI model, Fable, just days after it was released to the public. Anthropic had given assurances that it was safe but soon after its release, top administration officials developed fresh doubts that the AI’s guardrails were as secure as the company had suggested.

On Thursday, two days after the model’s public release, Amazon CEO Andy Jassy raised concerns to the White House about the ability to bypass the model’s guardrails, according to the two administration officials and the senior White House official.

(Amazon, which is an investor in Anthropic, was responding to an administration request for feedback, said a person familiar with Amazon’s discussions.)

By Friday morning, the issue had reached the highest levels of the White House.

Bessent, Cairncross, chief of staff Susie Wiles and other senior officials met to discuss the model and the administration’s response, according to the administration official and the senior White House official. Bessent joined remotely while traveling to Houston for a previously scheduled public event, one of them said.

Following the meeting, the administration attempted to reach Amodei but was told he was unavailable because he was attending a wellness retreat, one of the administration officials and the senior White House official said.

A spokesperson for Anthropic rejected the claim that he was at a wellness retreat, saying, “this is absolutely false.”

A person close to Anthropic said Amodei was first requested around noon and was on the phone with senior officials within an hour and 15 minutes. While he was out of pocket, Anthropic offered other senior leaders in his place, the person said.

When the administration finally reached Amodei, he participated in three calls with a combination of roughly half a dozen senior administration officials, including Cairncross, Bessent and Commerce Secretary Howard Lutnick, according to the senior White House official and one of the administration officials.

Other senior White House staff and administration officials including Under Secretary of Commerce for Industry and Security Jeffrey Kessler, White House staff secretary Will Scharf, White House deputy chief of staff Richard Walters, and assistant to the president for policy Walker Barrett also participated in some of the calls, according to the senior White House official.

During the calls, Amodei tried to clear up what he assumed was a misunderstanding. He pushed back on the administration’s concerns, defended the guardrails and argued that the type of bypass that occurred, which he believed to be specific, did not pose the same risk as a broader “jailbreak” that would allow it to be used without any of the guardrails put in place by Anthropic.

In a blog post after the export controls were put in place, Anthropic said that “no testers have yet been able to find a universal jailbreak — a jailbreak method that can very broadly bypass the model’s safeguards, unblocking a wide range of cyber capabilities,” and that total avoidance of any jailbreaks isn’t currently possible for them or any other companies. They defended their systems, which they said “are so strong that many users have complained that they are overly broad.”

Cairncross and Bessent were unmoved by Amodei’s arguments. A White House official said Amazon’s findings were run past the National Security Agency and they felt they had “proof.”

They urged Anthropic to voluntarily remove the model and coordinate with the government to address the vulnerabilities, according to the senior White House official and the two administration officials. Amodei asked for more time and information, but he made no commitments to pull the model, and at one point Bessent told Amodei directly that he was making a “bad decision,” according to the senior White House official.

Shortly after the call, the Trump administration imposed its export control on the Fable 5 and Mythos 5 models, citing national security authority and banning its use by foreign nationals, according to Anthropic. The company said the “net effect” of the order was to “abruptly disable” the models for all customers “to ensure compliance.”

“Export controls were a last resort after begging them for hours to work with us,” the senior White House official said. “This was not something we wanted to do, but our hands were tied.”

After publication, one of the people close to Anthropic disputed that the company was given a choice to voluntarily work with the administration.

“The White House gave 90 minutes to take the models down, with no details on the actual threat,” the person said. “There was never any begging — or asking — for them to work with us, just a declared 90 minute deadline.”

White House officials — who had heard Amodei liken the dangers of Anthropic’s technology to a nuclear bomb — were baffled when the CEO said he was unwilling to take the system down to address a known security vulnerability, the senior White House official said. Anthropic has defined itself among the industry as a vocal advocate for AI regulation to counter massive global security risks and job disruption as AI quickly advances.

Three people familiar with the government’s thinking said Amazon wasn’t the only company to raise concerns.

“The crux of the issue was the lack of seriousness that Anthropic was applying to it,” said one of the three people. “Had Anthropic taken it seriously and, rather than dismissing it as isolated, moved to fix or pause access, this would have never happened.”

Let’s be very blunt here. The “safeguards” around Mythos being used as Fable are purely driven by the agentic harness and the automated flags on the server side. What that means is that the system prompt will try to steer the responses in a different direction, and if that fails, then machine learning would ideally flag suspicious usage and then terminate the session. They clearly have struggled to get the false positives to an acceptably low level and then refused to expand the scope of the automated checks due to the poor user experience. The actual “jailbreak” appears to be defense-oriented prompting, i.e. asking the model to perform security audits that can then end up being used for adversarial behavior. This is of course the most obvious outcome possible for anybody who has dabbled in investigating and reporting potential exploits, as doing so can often result not in claiming a bug bounty, but in aggressive legal action from the company it was reported to. Depending on the trust level and perception of the participants in the process, “white hat” actions can end up being labeled as cybercriminal activity.

This is not “patchable” in the traditional sense of software development. It is possible to limit the risk, but that comes with a number of trade-offs. If we come back to the original post by Anthropic:

• We suspect that perfect jailbreak resistance is not currently possible for any model provider. Every safeguard used in the industry is vulnerable to non-universal jailbreaks (which can elicit some cyber information in specific circumstances), and it is likely that universal jailbreaks will eventually be found in the future. We stated this clearly when we released Fable 5.

• Given that perfect jailbreak resistance does not appear to be possible today, Anthropic adopted a defense in depth strategy with Fable 5. We aimed to make jailbreaks either narrow (in the case of non-universal jailbreaks) or very expensive to produce (in the case of universal jailbreaks), and to combine this with thorough monitoring to quickly detect and shut down any successful attacks. This is also why Anthropic has required 30-day retention of customer data with Fable—a policy change that carries real costs for us with customers, but that allows us to research and mitigate jailbreaks.

I doubt that, realistically speaking, models can be fully controlled in a certain direction. They can be trained differently (so the capability never gets baked into the weights) or they can be steered another way, but there is no real way to control the outputs due to the inherent design of LLMs (i.e. random token prediction based on statistical probabilities). The alternative is trying to control the output at the harness/server side level, but that comes with performance and quality trade-offs (similar to what the financial sector had to deal with post 9/11 and the significantly increased safeguards around financial crime).

This is not a reality that can’t be communicated to the administration. If anything, there have never been more technically competent individuals in critical roles that would influence decision making. Whether enough of them have deep enough cybersecurity background is a different topic, as covered here:

I think the path to this outcome, which I regard as a disastrous one, runs through the present makeup of the AI policy conversation. I have moonlighted in AI security policy meetings since ChatGPT shipped in 2022, and have observed an ecosystem of character types.

There is the sharp Georgetown graduate who works for a national security think tank in Washington, with a background in international policy and game theory, who has never done hands-on cybersecurity and reasons about it in the abstract; these people are genuinely brilliant and a pleasure to talk to.

There is the frontier-lab AI safety constituency, focused on the coarse-grained claim that scaling laws will deliver catastrophic cyber capabilities, and likewise largely untouched by the actual practice of defending or attacking systems.

There is the AI luminary, whose deep fluency in machine learning buys a kind of currency that licenses them to opine on cybersecurity despite having thought very little about it and holding only a toy model of how the domain works.

And there are a few (too few) people I admire deeply who carry real cybersecurity backgrounds into these rooms, and who, predictably, tend to treat most of what I have said above as obvious.

The problem is not that anyone should be evicted from these conversations, which sit upstream of decisions like the Mythos ban. The problem is one of distribution, and the fix is to inject far more people with deep, practical cybersecurity backgrounds into the rooms where the framings and metaphors get set.

It’s important to understand that “my model can find vulnerabilities” has little to do with effective outcomes in the field. I’ll avoid going into a big technical discussion here, but an easier way to understand this:

1. Using models and people’s time costs money. Any successful breach needs to cover the cost of inference and the time to execute. Most exploitable vulnerabilities do not justify the investment required to identify and leverage them.

2. A vulnerability in one application simply means that the specific app can be exploited. This does not guarantee lateral movement across a network, which is almost always required to gain access to high-value systems.

3. Application vulnerabilities have traditionally been tied to 15-25% of significant breaches in recent years. I covered this in my deep dive on Palo Alto Networks.

So when Anthropic goes on a Glasswing World Tour of Cyberfearmongering, touting Mythos as a literal cyberweapon, well, you play stupid games, you win stupid prizes.

At the end of the day, this situation has mostly arisen from the poor communication and negotiation strategy of the Anthropic leadership team. The reality is that you can't disrupt the workforce while simultaneously refusing to cooperate with the US Department of Defense, and then go and fear-monger about it as loudly as possible.

This is also not something that their lawyers will be able to solve for them. It’s a problem that solely Dario and/or Daniela (who appears to be controlling the majority of key decisions) should be addressing, working with the federal government to reach a reasonable solution. That "fix" doesn't need to be fully technical. It's also a question of changing the working dynamic between the two parties.

Sooner or later in life, you’ll have to get deals done with people you disagree with in order to progress. That’s been a foundational cornerstone of what we perceive as the concept of politics. Does Dario or Daniela have the mettle to get a deal done, or will they push the company into the inevitable outcome (destruction or nationalization) for the sake of their egos? Blaming everything on the administration due to political differences is a cop-out here, and a refusal to take responsibility.

If Anthropic is to ever go public (let alone this year), something at the leadership level needs to change.

I was intending to have this article be a test for the model and have it analyse itself in this format. Unfortunately for you, rather than reading the crisp thoughts of AGI, you’ll have to deal with me. I did flag it to the CCO of Anthropic, who was helpful about the issue.

Safety classifiers

The frontier cybersecurity and research biology capabilities of Mythos-class models mean that they pose a substantial risk of uplift to malicious actors. That is, these models could provide information or advice that assists those actors in causing serious harm that they couldn’t have received from other sources (for example, from internet search engines). Furthermore, a great deal of advanced usage of AI models is dual use: the same queries that are beneficial in the hands of cybersecurity professionals and biology researchers could be dangerous if available to malicious actors.

We therefore need strong safeguards to prevent misuse, and their coverage needs to be broad. The safeguards themselves have to stand up to sustained and sophisticated attempts to bypass them (also known as “jailbreaking” the system). The uplift from Mythos-level capabilities is valuable to many adversaries—for instance, those who could financially gain from cyberattacks—and we therefore expect them to be motivated to try to circumvent our safety measures.

Fable 5 comes with a new set of classifiers: separate AI systems that detect potential misuse, including jailbreak attempts, and prevent the main model (in this case Fable 5) from responding. We’ve been running classifiers on our models for some time, and Fable 5’s classifiers are an extension of this previous work with extra coverage.

When Fable’s classifiers detect a request related to cybersecurity, biology and chemistry, or distillation, the response is automatically handled by Claude Opus 4.8 instead. Users will be informed whenever this occurs. Opus 4.8 is a highly capable model in its own right: a response that falls back to Opus is a far better experience than an outright refusal from Fable. Our early data shows that more than 95% of Fable sessions involve no fallback at all—for those sessions, Fable 5’s performance is effectively the same as that of Mythos 5.

The following are the areas covered by the classifiers:

1. Cybersecurity. Mythos-class models excel at discovering and exploiting software vulnerabilities. They can thus make cyberattacks substantially easier and cheaper to commit. Mythos-class models also show strong skills in agentic hacking. This involves performing multiple different parts of a cyberattack in addition to finding exploits—reconnaissance, discovery, lateral movement, and more. To prevent these agentic hacking skills providing uplift in cyberattacks, we designed our cybersecurity classifiers to cover both exploitation and offensive cyber tasks in a broader sense. As shown in the graph below, our classifiers prevent Fable from making any progress on these tasks.

…

2. Biology and chemistry. We have long used our classifiers to block our models from responding on a narrow selection of bioweapons-related queries. But we are no longer certain that blocking this narrow selection is enough. This is for two reasons: first, we have reason for concern about well-resourced malicious actors attempting to gain uplift from our models for highly risky biological research. Second, models now have a greater ability to accomplish real-world scientific tasks.

While on paper this sounds logical, users are reporting many queries being blocked, particularly based on user history. Most actual professional biologists are basically unable to use the model, even if part of pre-approval programs. It appears that a similar issue is playing out across offensive cybersecurity professionals.

Subscribe now

The reason why my queries got blocked is because it appears that the model also is avoiding any workflows that might lead to training other models, i.e. asking it to analyse its scorecard is considered material risk. The day after the launch it was reported that some of these safeguards will have to be walked back, particularly those blocking any ML/AI work by external researchers.

For a small group of cyberdefenders and infrastructure providers, we’re also launching Claude Mythos 5. It’s the same underlying model as Fable 5, but with the safeguards lifted in some areas.² Mythos 5 will initially be deployed through Project Glasswing, in collaboration with the US government, as an upgrade to Claude Mythos Preview. It has the strongest cybersecurity capabilities of any model in the world. Soon, we intend to expand access to Mythos 5 through a broader trusted access program.

It’s important to note that Fable 5 is essentially Mythos with significant blockers around its best capabilities. Mythos is not being released to the public, but only to a small selection of companies, creating a “two tier” lane where large corporations and public sector organizations will have access to the best models, while the rest will be limited by design.

Availability

Claude Fable 5 is available everywhere today. Claude Mythos 5 is restricted to Glasswing partners (with cyber safeguards lifted) and soon to select biology researchers (with biology and chemistry safeguards lifted) only, until our broader trusted access program is available.

Pricing for both models is $10 per million input tokens and $50 per million output tokens. Developers can use claude-fable-5 via the Claude API.

We expect demand for Fable 5 to be very high, and difficult to predict. On the Claude API and consumption-based Enterprise plans, Fable 5 is fully available from today. For subscription plans, we’d rather give access sooner than later, so we’re rolling out more conservatively, in stages:

• From today through June 22, Fable 5 is included on Pro, Max, Team, and seat-based Enterprise plans at no extra cost.

• On June 23, we’ll remove Fable 5 from those plans. Using it after that will require usage credits. If capacity allows, we’ll extend the included window.

• After this point—when sufficient capacity allows us to do so—we aim to restore Fable 5 as a standard part of subscription plans. We intend to do this as quickly as we can.

While limiting the release of Mythos to the public is a hard fork, the pricing model represents a soft one. While Fable 5 is temporarily available as part of regular subscriptions, usage will shift to API access only after 6/22, and allegedly will be restored at some point if there is enough compute (there is never going to be enough compute).

From a product design perspective that’s not a lot of friction (users can toggle on demand billing with a single click in the interface and still keep working in their browser/desktop app). What is more likely to happen is that very few “civilians” will end up using these levels of models on a daily basis once the hard reality of the token cost comes into play. I wrote the following on X about testing Fable 5 in an Enterprise context:

While X has been on fire in the last 24 hours with influencers and power users sharing their experiences with the latest Anthropic model, most of it has been limited to coding or “difficult problems”.

Anthropic is generating tens of billions of ARR predominantly from Enterprise usage. A lot of it is from coding, but there is a significant portion driven by users adopting Cowork as a job companion. Cowork was mostly ignored by the X power users because most of them do not actually work directly with Enterprise customers, where they need to present, analyse and deliver on outcomes under deadlines and with millions on the table.

I’ve been using AI heavily for Enterprise work for a while, but mostly in the context of deep research reports and accelerated learning in complex technical or industry domains. Similar to coding, I expected to see progress of the models being useful for translating this domain knowledge and analysis into more “practical” deliverables, but I found it mostly lacklustre.

Two months ago I started using Cowork also for building out presentations with the Opus models. It’s okay, but I would rate it as having a junior employee doing a barely passable job at it, rather than a meaningful differential. Opus (and Anthropic models in general) have had spotty performance in the last months, mostly driven by what I perceive as dynamic performance reductions based on available compute, combined with aggressive testing of efficiency algorithms in production.

I don’t think that the Anthropic team is “dumbing down” models on purpose, but they have significant constraints vs their $60B ARR of inference currently being used. This results in difficult choices with tradeoffs that make the experience for power users worse. The net effect is that while using Opus in Cowork helps me reduce the hours needed to get to a great presentation, I’ve often ended up spending $15 bucks (the average cost) to get an average deliverable, which I would end up using almost nothing from but at least it helped me clarify what I really wanted to show and how to position it.

I spent most of today working on a difficult pitch to a customer that wants to cut costs aggressively for a renewal. They are in financial services and running a complex platform that has internal cost pressures. The real work will be going there in person and convincing them to change how they work, but to get to that point I need enough bait to convince them to take this seriously.

Mythos costs approximately double the Opus rate card (which is not a guarantee for the real running usage due to token consumption) but in this case it ended up exactly 2x my typical presentation bill ($30 bucks in magic API tokens).

Unlike Opus however, it actually delivered a fully usable transformational pitch, based on their full real life context. Typically when I feed a lot of documents to Opus, it fails to actually transform these into insights for a presentation. Here it delivered a full:

1. Intermediate level slide deck

2. Value assessment model

3. Internal strategy doc that makes sense

Most of the power users on X will lose access to Mythos/Fable on June 22. Some of them will keep paying the API costs or will use it in other coding tools where they are paying for usage, but in general they will not keep trying complex problems with it.

I’m also a heavy user of Codex and GPT Pro models. Mythos is a proper competitor to GPT Pro and in the context of Enterprise work, it’s better since it produces better materials with its implementation of computer use.

I think that for OpenAI to seriously challenge back Anthropic in the Enterprise market, they have to deliver a similar type of experience. The problem they have is a traditional network effects and distribution advantage risk - many companies are diverting their spend towards Anthropic and will not churn from month to month due to the pendulum swinging in the other direction.

xAI and the Cursor team also have a limited window to enter this conversation, but realistically speaking they will likely only be able to compete on cloud coding agents use cases (which is still worth hundreds of billions in market value).

Google, the proud owner of DeepMind and Workspace should be owning this category and workload. Their failure to compete here should’ve cost several C-levels their roles.

Microsoft also had a significant moat here which was quickly destroyed due to complete lack of vision and product execution.

Mythos/Fable is strong for Enterprise workloads. The next big training run is probably going to be the actual “white collar AI layoffs” trigger. I have no reason to hire junior employees anymore based on what I can get out of $30 investment.

I’ve repeatedly talked about the idea that the “permanent underclass” is a very likely outcome because the top performers using best-in-class AI are likely to deliver the productivity and impact of literally tens of FTEs. Historically, this type of leverage has been driven by elevating top performers to “strategic” senior management roles, where they can set the direction and the org structure below them goes and executes. Since the average productivity of that org structure is not particularly high, we have a lot of FTEs spending a lot of hours on singular deliverables - a product demo, a filled out vendor questionnaire, a presentation. Capable enough AI models collapse this output on behalf of the top performers, who typically have the “final say” over the messaging and content structure anyway.

This is why we’ve seen “AI-native” companies of high growth deliver exceptional revenue per employee compared to supposedly best-in-class large corpos with massive distribution and capital advantages. In my example, if I want to scale my team, it’s smarter to get one other FTE that’s on a similar capability level as me + a large token budget, than 3 other FTEs that will waste a lot of time and require significant supervision in order to get to essentially the same outcome.

It’s not surprising that Dario’s return to blogging happened the day after the Fable 5 launch, in order to argue over government policy:

2. Macroeconomics and tax policy

Governments have long faced the problem of how to encourage economic growth while also providing important public services and ensuring that the least fortunate are taken care of. An important (and generally correct) premise of these debates has been that economic growth is fragile and difficult to achieve—that while reducing inequality might provide important benefits, it has to be traded off against the economic drag of increased taxes or deficits.

I suspect that powerful AI may scramble this assumption. If AI achieves the ability to do most cognitive tasks far better than humans, it stands to reason that it could result in extremely rapid and robust economic growth via the acceleration of science, technology, and operational efficiency. The iterative ability of AI to build even better AI may supercharge that growth even further. But for exactly the same reasons, AI may also act as a more general economic substitute for human cognitive abilities than previous technologies have, while also altering the economy far faster than previous technologies have. Thus, it’s reasonable to think that AI could produce much larger disruptions to the labor market than previous technologies, and, potentially, more enduring disruptions. We risk ending up in a world where the economic tradeoff dial is stuck on the hypergrowth, hyper-inequality setting, and is potentially very hard to unstick from that setting. The key challenge in such a world won’t be incentivizing growth, but finding a way for everyone to share in the benefits.

Of the topics discussed in this essay, macroeconomics and enduring labor displacement are arguably the ones that have attracted the most public attention and the most misunderstanding, so I want to be extremely clear on two points.

First, enduring job displacement is undesirable and dangerous, and we should do everything we can to minimize or prevent it, not to bring it about. I have warned about job displacement in interviews and essays because I want both policymakers and the private sector to have the best chance to adapt and respond, not because I am trying to be a “prophet of doom”. As a company, Anthropic always does as much as it can to work with customers to find creative new use cases and new sources of revenue that allow them to do more with their existing workforce, rather than focusing solely on cost savings (which often means reducing the workforce). We also constantly try to think of new interaction paradigms that allow humans to have as active a role as possible in collaborating with AI systems as those systems advance. More broadly, it is valuable for the whole world to experiment with using AI in as many new ways as possible, as that is the way for society to discover new possible job configurations. I do think AI will enable a number of new economic opportunities. I’ve predicted that AI will enable single individuals to create billion-dollar companies, and we’re already seeing teams of only a few people build businesses with hundreds of millions in revenue. But at the same time we should recognize that there’s a decent possibility that, despite all our efforts, AI still causes significant enduring job loss—and that this may be an intrinsic property of the technology and the way it broadly replicates human cognition⁴

Second, any response to AI-driven job displacement needs to address both the need to provide for everyone economically, and the need for people to find meaning, purpose, and agency. The latter is ultimately more important, and it depends on deep questions about how society is organized, what people should strive for, and what constitutes the good life. I am actually very optimistic that, even in a world with AI’s that are better than everyone at everything, humans can live lives of deep purpose and strive to build awe-inspiring and beautiful things⁵

But this is something to be collectively worked out by society as a whole, not something policy can directly address. Policy can be most helpful in buying us time to do that work, by slowing down job loss and providing economically for those likely to be affected.

In that spirit, some key policy interventions that are likely to be helpful include:

• Measurement and tracking. It’s easy to dismiss mere data collection and analysis as inadequate to the scale of the problem, but we are unlikely to get good policy if we cannot accurately measure what is happening on the ground. Anthropic has been operating an Economic Index of how people use Claude for nearly a year and a half, but governments have access to types of data we do not, and could greatly expand their economic statistics to more carefully track AI job displacement.

• Pro-employment incentives. A wide range of pro-employment policy incentives can help to slow or reduce job displacement, including: wage insurance policies that compensate people when they have to take a lower-paying job⁶, retention tax incentives to encourage employers not to make layoffs, workforce training grants, or infrastructure to facilitate matching of employers to employees to speed the rate of labor market adaptation. While the particulars of which interventions are best will depend on what kind of labor displacement AI brings, we should readily accept the costs and market inefficiencies that these policies could entail, particularly as they are likely to be offset by AI-driven productivity gains.

• Long-term macroeconomic support. If AI-driven labor displacement ends up being large in magnitude and permanently drives down the demand for labor, it will likely be necessary to go beyond mere incentive programs to long-term income support for a significant fraction of the labor force. Mechanisms such as universal basic income could be financed through taxes on relevant companies or raising the capital gains tax. Universal capital accounts offer another vehicle. Broadly speaking, fast economic growth should create the tax base for shared prosperity.

A common focus of economic concern about AI that I haven’t mentioned has been datacenters and particularly their potential to raise energy prices. My view is that AI companies should pay to absorb rate increases—and Anthropic has already made a pledge to do so—but I see public hostility to datacenters as largely a symbol or outlet for broader economic anxieties about AI. It is important we have a direct societal conversation about these wider economic issues and truly have compelling solutions for them, or else they are likely to manifest indirectly, as they have with datacenters.

“It’s time to take Universal Basic Income seriously, y’all” is quite the pitch directly after launching Fable 5.

So how does Fable 5 benchmark? On paper, it's the strongest model in the world. When using it to work on my own applications, it's the first Anthropic model in a while that GPT 5.5 xhigh has no fixes to do. In my experience, since the launch of Codex, the best model to use from end to end to deliver an application has been from OpenAI. When using Claude, there was always something that's not done properly, introducing either bugs or security risks or the functionality just doesn't work. The GPT models running on their highest token budgets were my "senior dev fixing it". Not anymore with Fable (not that my actual usage will change here, as using the $100 subscription from OpenAI covers significantly more coding and analytical workloads than using the same $ value via on-demand tokens with Fable 5).

I think that the best way to think of these models is not that you are suddenly speaking to the highest intelligence in the universe, but that it actually completes fully the challenges you throw at it. My biggest issue with Anthropic models in recent months is that you are almost always guaranteed a significant mistake or wrong conclusion the more complex the conversation gets. This “selective stupidity” is likely driven by the efficiency design of older training runs combined with aggressive compacting of the context window. It’s not that Opus was not capable of strong insights, it’s more that they would be sporadic because somehow it keeps forgetting critical information even if given to it earlier in the conversation.

At least in this first week of the launch, Fable doesn’t seem to have these issues. It also seems to perform at a similar cognitive level to GPT 5.5 Pro, which is essentially the only model I use for day-to-day conversations that require deeper thought. Keeping in mind that it’s a cheaper model (on paper, the prompts via the OpenAI subscriptions are very generously subsidized), this is a significant win.

The Cursor benchmark paints a similar picture, where interestingly enough Fable 5 low is competing at the same performance level as 5.5 xhigh, albeit at almost double the dollar value in token usage.

What happens next? I think that the compute availability and pricing issue remain the biggest issue for scaling usage. I would expect the majority of token use for Fable 5 to remain in Enterprise for users that have significant token budgets per employee. I think that the model is the strongest “average productivity companion” currently in the market, albeit it does require actual results to be delivered with it in order to justify the investment.

In my last article on cloud infrastructure software dynamics, one of the key insights was the transformation of the security market, as AI has quickly become both a significant opportunity to improve existing products and a risk vector. Almost every CISO in the industry will be buying a platform to either monitor or secure AI agents over the next twelve months, which prompts the obvious question: who will benefit the most?

One of the existing incumbents with a strong claim to be able to perform in the new age of AI is Zscaler. I haven’t properly covered them previously, so let’s dig in.

The key takeaway

For tech sales and industry operators: Zscaler clearly has valuable installed distribution, switching costs, enterprise trust, recurring revenue, and a large customer base, which are the raw materials of a good business. But a good business is not automatically a great business if the category is crowded, the buyer can compare several credible platforms, and growth is decelerating despite the claimed demand shock. From an operator’s perspective, the hard question is whether Zscaler is becoming more essential inside the customer’s operating model or merely adding modules to defend its existing account base. The cleanest positive signal would be customers standardizing around Zscaler as the default control plane for user, workload, branch, cloud, data, and agent access. The cleanest negative signal would be AI products staying small while the legacy access business keeps carrying the numbers. If Zscaler can make itself the default map of “who or what can access what,” it becomes much more than remote access; it becomes the security ontology of the enterprise. If it cannot, then AI Protect becomes another module in a bundle, and the company remains exposed to Palo Alto, Netskope, Cato, Cisco, Microsoft, CrowdStrike, and every platform trying to own the same control point. The structural problem for the sales teams is speed: the motion is consultative, GSI-led, and measured in quarters, and the threat is measured in minutes, so whoever shows a customer their own live exposure instead of describing it in a deck wins the part of this the incumbent is built to lose. What happens next is undecided, but the revolving door at the top of the sales org is the canary in the coal mine.

For investors and founders: The company sits at the intersection of three cycles: the cybersecurity consolidation cycle, the AI adoption cycle, and the software multiple cycle. The AI adoption cycle is bullish for security demand because complexity, autonomy, and attack speed all rise. The consolidation cycle is mixed because large vendors can bundle more aggressively, customers want fewer tools, and SASE/security platforms may converge around a small number of winners. The software multiple cycle is unforgiving because investors no longer reward narrative without clean organic growth, especially when sales leadership changes and acquisitions blur the signal. That means Zscaler can be fundamentally important and still be a difficult stock if the timing of demand, execution, and valuation do not align. For founders, the lesson is to build into the pressure points created by the cycle: cost reduction, tool consolidation, AI-risk auditability, machine-identity governance, and board-level evidence. For investors, the alpha is to watch the leading indicators before the income statement: ZPA expansion, AI Protect attach rates, data-security ARR growth, Zero Trust Branch adoption, new-logo recovery, GSI/channel pull, and whether organic net-new ARR stops decelerating. The concerning sign today is that the company’s AI story still sounds like discovery, guardrails, DLP, red-teaming, and reporting: useful, but not obviously revolutionary. They should be aggressively asking: what would the security stack look like if built today for millions of non-human agents, not retrofitted from human access controls? That leads to products like self-updating permission systems, agent sandboxing, exploit-path simulation, autonomous kill switches, and real-time policy generation based on observed behavior. The investable company is the one that turns AI security from a dashboard into a control loop.

Zero trust issues

Jay Chaudhry: We delivered strong Q3 results. ARR grew 25%, and non-GAAP operating margin hit an all-time high at 23%. AI is changing the nature of cybersecurity in real time. Zscaler is the cybersecurity platform for the AI era. This is evident in our results and the reason we are so confident in our long-term potential. We offer the industry's only complete Zero Trust SASE solution, a singular Zero Trust platform across users, across cloud workloads, and across branches. Our architecture is purpose-built to address the limitations of firewall-based SASE solutions and has several key differentiators. First, we hide applications and data behind our Zero Trust Exchange, making them invisible from the internet and eliminating the attack surface. An attacker can't breach what it can't reach. Hence, this architecture provides far superior cybersecurity protection for our customers.

Second, we eliminate lateral movement of attackers with our Zero Trust architecture. We only allow authorized users and workloads to access specific applications. This reduces the blast radius of a potential breach, providing better security to our customers. This stands in stark contrast to competitors with firewall-based SASE architecture that connect users to the corporate network. Once a malicious actor gains a foothold on the network, it can roam freely and systematically attempt to compromise critical applications or steal data. This is how most ransomware attacks happen. Finally, scale matters, and our cloud-native Zero Trust Exchange is the largest distributed inline security platform in the world that spans across 160 public exchanges, processing more than 500 billion transactions per day. This gives us the best quality and quantity of telemetry data. Simply put, no other cybersecurity vendor has access to data sets with comparable fidelity and breadth.

This high-fidelity telemetry fuels our AI-powered security capabilities, continuously improving how we detect, prevent, and stop threats. These differentiators are especially important at a time when organizations are aggressively deploying AI applications and models with growing interest in AI agents at scale. We expect it won't be long before millions of AI agents have access to organizations' mission-critical applications and sensitive data. Today, users are the weakest link in cybersecurity. Soon, AI agents will be the weakest link because they operate at far greater speed and have far less oversight. Even a single compromised agent can move from discovery to data theft in minutes, inflicting catastrophic damage on enterprises. Making it even more challenging, new powerful frontier AI models like Mythos are finding security vulnerabilities in software at machine speed, significantly diminishing the effort, skill, and time needed to breach enterprises.

All enterprises already have thousands of known vulnerabilities that they haven't been able to patch. Frontier models are multiplying these unremediated vulnerabilities by as much as 10x, and even more powerful models that are currently being developed will undoubtedly make it worse. Enterprises don't have the capacity to patch and update existing vulnerabilities, so backlogs are piling up faster than organizations can address them. To tackle this challenge, the market needs to take a different approach. We provide the two most important defenses against these vulnerabilities. One, hiding applications from attackers, and two, eliminating lateral movement at scale. This validates the architecture we pioneered. Zscaler was built for this moment. We started with Zero Trust security for users, so users can safely access applications from anywhere. We expanded our Exchange to provide Zero Trust security to branches, workloads, and connected IoT/OT devices.

Now we are expanding our exchange to secure AI agents. An important element of agentic security is to understand which agents, users, and other identities are communicating with which models, applications, and data sources. On May 21st, we announced our intent to acquire Symmetry Systems, a company that solved this difficult problem. Symmetry provides an access graph that maps how identities, applications, and other data sources connect across the enterprise. We are integrating its access graph technology with our Zero Trust Exchange. We're excited to share more about this at our Zenith Live user conference in Las Vegas next month. We are also partnering with Anthropic on Project Glasswing and with OpenAI as part of its Daybreak program, formerly known as Trusted Access for Cyber, or TAC, which allows us to access frontier models to proactively harden our systems and deliver better security and resilience to our customers.

From my personal point of view, zero trust architecture always made a lot of sense in the context of cybersecurity. The majority of the damage from breaches is related to lateral movement (i.e. one employee gets compromised, then the attacker accesses a sensitive database from their account, for example), and reducing dwell time is arguably the most important risk reduction activity from a technical perspective (leaving attackers fewer opportunities to move across the network). As such, setting up internal blocks and reducing access across applications and endpoints is a logical way to approach the problem.

Read more

While progress in the last two years has been rapid, for those that are trying to get an edge by adopting the latest models at launch, it’s been difficult to easily understand “how well” we’ve progressed. Benchmarks have served as a relative proxy for progress by visualizing the gaps between models and indicating directional usefulness. When users actually leverage models in their daily workflows, however, those supposed differences are rarely clear and, more often than not, misleading. Users vote for real utility by focusing their token spend only on the actual top performers.

DeepSWE is a long-horizon software engineering benchmark that delivers four major advances over today’s public benchmarks:

• Contamination free: Tasks are written from scratch, not adapted from existing commits or PRs, so no model has seen the solution during pretraining.

• High diversity: Tasks span a broad pool of 91 repositories across 5 languages.

• Real-world complexity: Prompts are half the length of SWE-bench Pro’s, yet solutions require 5.5x more code and ~2x more output tokens.

• Reliable verification: Verifiers are hand-written to test software behavior rather than implementation details.

Existing benchmarks fall short on several of these axes. SWE-bench Pro, the leading agentic coding benchmark, has tasks averaging just 120 lines of code to solve, and our audit found its verifier misgrades agent outputs at rates of 8% false positives and 24% false negatives. Frontier labs are also raising growing concerns about benchmark contamination.

By contrast, DeepSWE produces a sharper comparison of frontier coding agents. Models that appear close together on public benchmarks separate into wide, ordered gaps that match the differences developers see in day-to-day agent workflows.

The team at Datacurve has come up with a new benchmark they call “DeepSWE,” which appears to be quite a good reflection of how models today “feel” in terms of performance, rather than purely how they benchmark.

Subscribe now

1. Long-horizon work, realistic and short prompts

DeepSWE prompts are aligned with the way developers talk to their agents: behavior-focused, short, and free of large interface-definition blocks, rather than overly verbose and prescriptive. Agents must discover where and how to implement the change, so a substantial share of the capabilities being evaluated involve end-to-end exploration instead of just the execution of an overspecified engineering task.

Public benchmarks sourced from GitHub issues and pull requests often carry more detail: reproduction steps, additional context, code snippets, and tests that assume specific symbols or signatures. DeepSWE instead scores observable behavior, which lets prompts stay short and natural even when the underlying tasks are substantially longer.

Most benchmarks today are actually very scripted and specific. This made sense in the early stages of agentic coding due to a lack of optimization for software development tasks, but things look quite different nowadays. Nobody spends time on basic direction like “you can use this tool, ensure you open it this way,” but instead focuses on back-and-forth around the practical outcome.

2. Broad repository coverage

DeepSWE contains 113 tasks spanning 91 active open-source repositories across 5 languages: TypeScript, Go, Python, JavaScript, and Rust. Sampling at this scale makes DeepSWE a much stronger proxy for the real-world utilities of coding agents: whether they can make useful, well-scoped changes across varied codebases with different levels of structure, documentation, and maintenance.

Existing public benchmarks are much more concentrated. SWE-Bench Pro Public spans 11 repositories, and SWE-Bench Verified spans 12, with many tasks drawn from prominent, heavily maintained projects. That is a narrower setting than the range of projects developers bring to coding agents in practice.

An issue with current benchmarks is that they often cover “known” issues. In practice, this means many of the new models are optimized for these challenges because either the problem is already solved in their training data or the models “cheat” by looking up the real code used to solve the problem.

3. Novel tasks test problem-solving, not recall

Every DeepSWE task is original: the reference solution is written from scratch rather than copied or adapted from an existing pull request, commit, or public patch. Some tasks are motivated by unresolved GitHub issues, but the fix itself is new. DeepSWE tasks are also never merged back into the upstream repositories, so they do not enter the public GitHub record and are unlikely to appear in future pre-training corpora scraped from open source.

This makes DeepSWE a cleaner test of whether an agent can solve a novel software engineering problem, rather than recall, retrieve, or rediscover a public fix.

Benchmarks sourced from existing commits have an inherent contamination risk because its corresponding implementation, tests, and discussion are already online. Our SWE-Bench Pro audit found that this risk is not hypothetical, with solution leakage and false positives affecting roughly 8% of audited rollouts.

The real challenge today in benchmarking how models are progressing on software engineering tasks is their ability to solve novel problems, not to act as a glorified wiki for previously used code. This is why one of the first tests power users typically run is to ask the models to create an application for them, and they keep comparing how different models are able to execute on it, rather than how well they are benchmaxxing.

The practical results here are rather striking versus what typical benchmarks show when comparing models. GPT 5.5 is clearly the leading coding model today, with Opus being the second best. The gap between the two was minor by the end of last year, when Claude Code became widely adopted due to user preferences, but GPT 5.5 has changed things and this has been obvious to most power users as sentiment has shifted on X. Google models are essentially useless for agentic coding on novel problems, and this has been the case for a while.

If we visualize this in terms of both performance and cost, the lead here for GPT 5.5 (xhigh) is significant. It’s both cheaper and more effective than Opus 4.8, which is a minor improvement in efficiency versus 4.7.

What’s also interesting is how poorly open-source models or alternatives from Google and xAI are performing on these evaluations. While we shouldn’t be too negative about their performance, they are clearly one to two releases behind the two leading labs and very much not a cost-efficient choice. The Gemini 3.5 Flash fiasco speaks for itself: it released a model that is half as effective in coding tasks as Opus 4.8 on high, while consuming almost double the tokens in dollar value.

The notable missing player here is Cursor with Composer 2.5 since the model is only available in the Cursor harness and not via API. It’s also important to note that a better coding model doesn’t automatically translate to higher revenue performance, as Anthropic is currently killing it in terms of GTM execution with enterprises versus OpenAI, regardless of what the “terminally online” see in their own usage.

In this environment, cloud infrastructure software, together with the compute layer, is the critical "connective tissue" to get the most out of silicon intelligence. "The Infra Red" is an annual report by Redpoint that covers the thesis behind cloud infrastructure software and what's coming next.

While we leverage a lot of different benchmarks to evaluate how "smart" models are, the ultimate test is doing any task we throw at it and doing it right. While in the earlier versions of GPT and Claude models the architecture of the LLM was a limiting factor on how much context they could work with, today we are seeing most of the limitations driven by compute constraints. Given enough compute, models can run for many hours and test out solutions until they solve the problem.

While there are many (influential) opponents of the adoption of AI, the reality is that when money talks, bullshit walks. Companies are shifting significant spend into leveraging frontier models in their work, resulting in explosive revenue at the model layer.

Read more

Together with SpaceXAI, we're training a significantly larger model from scratch, using 10x more total compute. With Colossus 2's million H100-equivalents and our combined data and training techniques, we expect this to be a major leap in model capability.

In my last article I covered the SpaceX(AI) IPO and the core strategy of the company (space, connectivity, and AI), highlighting that the company is pitching compute and intelligence for Enterprise applications to investors as the largest addressable market they have. If we look at the company's revenue today, however, they've been struggling to grow into "selling Grok to companies," particularly when compared to the massive demand that Anthropic and OpenAI are handling today.

The key to unlocking the market for them comes with the likely acquisition of Cursor, one of the early and most interesting winners in the developer tools space. Rather than limiting themselves to competing with the tens of coding harnesses currently on the market, they’ve been building in two directions that would allow them to scale: internally trained coding models and cloud.

Targeted RL with textual feedback

Credit assignment during RL is becoming an increasingly difficult challenge as rollouts can span hundreds of thousands of tokens. When a reward is computed over an entire rollout, it may be hard for the model to tell which specific decision helped or hurt the outcome. This is especially limiting when we want to discourage a localized behavior, such as a bad tool call, a confusing explanation, or a style violation. The final reward can tell us that something went wrong, but it is a noisy signal for where it went wrong.

To address this, we trained Composer 2.5 with targeted textual feedback.¹ The idea is to provide feedback directly at the point in the trajectory where the model could have behaved better. For a target model message, we construct a short hint describing the desired improvement, insert that hint into the local context, and use the resulting model distribution as a teacher. We use the policy with the original context as the student and add an on-policy distillation KL loss that moves the student’s token probabilities toward the teacher’s. This gives us a localized training signal for the behavior we want to change, while still retaining the broader RL objective over the full trajectory.

The funny thing about their coding model is that it’s picking up a well-performing open source model (in this case Kimi K2) and then they use their unique training data set to, as the kids would say, “rl the shit out of it.” This is not a magical pill, however:

Synthetic data

During RL training, Composer’s coding ability improves substantially to the point where it begins to get most training problems correct. To continue increasing intelligence, we both select for and create harder tasks dynamically throughout the run. Composer 2.5 is trained with 25x more synthetic tasks than Composer 2.

We use a range of approaches for creating synthetic tasks that are grounded in real codebases. For example, one synthetic approach is feature deletion. For these tasks the agent is given a codebase with a large set of tests, and asked to delete code and files in such a way that the codebase remains functional while specific testable features are removed. The synthetic task is to reimplement the feature, and the tests are used as a verifiable reward.

One downstream consequence of large scale synthetic task creation is that it can cause unexpected reward hacking. As the model became more adept, Composer 2.5 was able to find increasingly sophisticated workarounds to solve the task at hand. In one example, the model found a leftover Python type-checking cache and reverse-engineered the format to find a deleted function signature. In another, it was able to find and decompile Java bytecode to reconstruct a third-party API. We were able to find and diagnose these problems using agentic monitoring tools, but they demonstrate the increasing care necessary for large scale RL.

The team is getting more sophisticated, both by extending the training with synthetic data, as well as on how to approach the verifiable inputs during training. This is where the team is evolving from a “SaaS vendor playing around with model training” to doing genuine AI lab work to get improved outcomes.

Subscribe now

The end result is a model that's very competitive performance-wise, while being significantly cheaper than the "high" versions of either GPT 5.5 or Opus 4.7. Combined with the significant token generation speed, with Composer 2.5 the company has a genuinely interesting fit for Enterprises that are struggling with optimizing the running costs of their engineering teams.

For many companies that are adopting agentic coding at scale, Cursor as a product has been an interesting fit because it allows the team to rotate to different models that can easily shift their monthly cost significantly. If a company commits to a single provider (and receives minimal discounts because the sales teams at Anthropic and OpenAI currently barely offer any meaningful cost reductions), they run the risk of either overspending relative to the alternative models, or being left behind if one of the labs pulls ahead and delivers significant improvements.

This has been a big blocker for the xAI sales teams, as Grok has failed to be meaningfully competitive or cheaper, which means that most customers simply lacked the incentives to pursue any meaningful commits. Using Grok on-demand from the hyperscaler APIs was then the main source of revenue; however, that usage is fickle and easy to swap.

This is why the frontier labs are also trying to build out user surfaces that are "sticky," e.g. Claude Code, Cowork, Codex, etc. Cursor initially started with a relatively sticky surface (everybody was familiar with working with VS Code), but over time lost some of that advantage due to the move towards CLIs, as well as not offering a meaningfully differentiated experience. If anything, one of the big critiques of Cursor from a user adoption perspective was that the product was becoming too buggy and overcomplicated.

Agents are only as capable as the environment they run in. Without the ability to use the software they are creating, agents hit a ceiling.

Over the last few months, we addressed this internally by giving agents their own virtual machines with full development environments, and the ability to test their changes and produce artifacts (videos, screenshots, and logs) so you can quickly validate their work.

Today we’re making a new version of Cursor cloud agents available from anywhere you work, including the web, mobile, desktop app, Slack, and GitHub. Cloud agents onboard themselves onto your codebase and produce merge-ready PRs with artifacts to demo their changes. You can also control the agent’s remote desktop to use the modified software and make edits yourself, without checking out the branch locally.

This has been the biggest shift in how we build software since the move from Tab autocomplete to working synchronously with agents. More than 30% of the PRs we merge at Cursor are now created by agents operating autonomously in cloud sandboxes.

The answer to this from the perspective of the company is to offer improved outcomes for engineers by giving coding agents virtual environments to test solutions and work on problems.

Using cloud agents at Cursor

For the last month, we’ve been using cloud agents internally, and it has changed how we build software. Instead of breaking tasks into small chunks and micro-managing agents, we delegate more ambitious tasks and let them run on their own.

These are a few ways we’re using cloud agents:

Building new features

We used cloud agents to help us build plugins, which we recently launched on the Cursor Marketplace. Here is one of our prompts:

For each component displayed in a given plugin’s page, we’d like to include a link to the source code. For skills, commands, rules, and subagents - that’s the .md file. For hooks, it’s the hooks.json. For mcps, it’s the .mcp.json or the manifest where it’s defined. As we index all the components of a plugin, keep track of the source file and construct links to that file by way of the underlying github url. Surface this to the frontend and have our frontend link out to github using this icon.

The agent implemented the feature, then recorded itself navigating to the imported Prisma plugin and clicking each component to verify the GitHub links.

For local testing, the agent temporarily bypassed the feature flag gating the marketplace page, then reverted before pushing. It rebased onto main, resolved merge conflicts, and squashed to a single commit.

Reproducing vulnerabilities

We kicked off a cloud agent from Slack with the prompt, “Please triage and explain this vulnerability to me in great detail,” followed by a description of a clipboard exfiltration vulnerability. When the agent finished running, it responded in the Slack thread with a summary of what it accomplished.

The agent built an HTML page that exploits the vulnerability via an exposed API. It started a backend server to host the demo page locally and loaded the page in Cursor’s in-app browser.

The video artifact shows the complete attack flow: the agent copied a test UUID to the system clipboard, loaded the demo page in Cursor’s browser, and clicked a button to exfiltrate and display the UUID. It also took a screenshot showing the successful clipboard theft and committed the demo HTML file to the repo.

While both Anthropic and OpenAI offer their own versions of agentic computer use, the Cursor implementation appears to be more widely adopted and can be seen as one of the key drivers behind their recent growth trajectory towards a $3B ARR run rate. Unlike the xAI team, the Cursor leadership and engineering teams are working towards a very opinionated version of what writing code will look like in the future:

What’s next

We’re building toward a future of self-driving codebases, where agents merge PRs, manage rollouts, and monitor production. We will go from a world where developers use agents to create diffs to one where agents ship tested features end-to-end.

To fully realize that shift will require improving tooling, models, and the interaction patterns. Our near-term focus is on coordinating work across many agents and building models that learn from past runs and become more effective as they accumulate experience.

The roadmap for Cursor actually fits quite well with some of the core products that xAI was trying to pursue before the recent reshuffle, when the majority of the research team left.

When we started building Cursor a few years ago, most code was written one keystroke at a time. Tab autocomplete changed that and opened the first era of AI-assisted coding.

Then agents arrived, and developers shifted to directing agents through synchronous prompt-and-response loops. That was the second era. Now a third era is arriving. It is defined by agents that can tackle larger tasks independently, over longer timescales, with less human direction.

As a result, Cursor is no longer primarily about writing code. It is about helping developers build the factory that creates their software. This factory is made up of fleets of agents that they interact with as teammates: providing initial direction, equipping them with the tools to work independently, and reviewing their work.

Many of us at Cursor are already working this way. More than one-third of the PRs we merge are now created by agents that run on their own computers in the cloud. A year from now, we think the vast majority of development work will be done by these kinds of agents.

If the team is able to deliver on the promise of “you can just hire our AI cloud engineers as your R&D team,” this has obvious advantages that many Enterprise customers might prefer to invest in. The lessons and operational competences at developing this for coding can then transfer into the “Macrohard” project, where xAI was trying to leverage computer use for a variety of white collar workflows.

I think that the vision is directionally correct and mostly impossible to achieve alone, at a time when OpenAI and Anthropic are pouring significant amounts of resources into a similar direction. The obvious choice for Cursor was to partner with either xAI or OpenAI, and given the price tag to acquire Cursor, it’s clear that Elon Musk went ahead and got the deal done.

Cursor has also been investing in their go-to-market strategy, both by building out an Enterprise-focused sales team and by running a significant number of DevRel events under the "Cafe Cursor" umbrella.

How the commercial teams will merge with xAI's existing hires is yet to be seen, although due to the recent wave of hires from playbook companies, it's likely that the integration will be rather quick. I think that they will benefit from keeping the Cursor branding and the developer-focused marketing, which so far has resonated a lot more than the existing Grok/xAI positioning.

Most IPOs are exits. This one is a funding round dressed as an exit, and depending on how things go, a significant benefit to retail investors. SpaceX is one of the very few public market entrants in the last decade with a genuine monopoly in its core business, a credible thesis on the future of compute, and a founder with a track record of doing the impossible repeatedly.

Vertically integrated innovation engine

“You want to wake up in the morning and think the future is going to be great—and that’s what being a space-faring civilization is all about. It’s about believing in the future and thinking that the future will be better than the past. And I can’t think of anything more exciting than going out there and being among the stars.”

—Elon Musk

Our Mission

Our mission is to build the systems and technologies necessary to make life multi-planetary, to understand the true nature of the universe, and to extend the light of consciousness to the stars. To do this, we have formed the most ambitious, vertically integrated innovation engine on (and off) Earth with unmatched capabilities to rapidly manufacture and launch space-based communications that connect the world, to harness the Sun to power a truth- seeking artificial intelligence that advances scientific discovery, and ultimately to build a base on the Moon and cities on other planets.

The interesting thing about the SpaceX IPO is that the execution here can be considered very boilerplate and favorable for the original investors and the liquidity providers. One of the biggest reasons for Musk's effectiveness and longevity in the game is that he is laser-focused on getting a good deal for his investors and supporters. Still, the actual proceeds of the IPO will go directly to funding AI compute infrastructure, launch infrastructure, satellite constellation buildout, and general corporate purposes.

This is particularly important when the mission that is supposed to be funded here is nothing short of staggering. The goal is very straightforward, SpaceX(AI) intends to compound on its unique manufacturing and hardware execution capability to bring significant amounts of compute to space, that will be both an economic engine for the company and used to build a lead in model training over time.

It's important to note that the successful execution of this IPO will not only provide Musk with significant liquidity to scale his operations, but will also cement him as the unfireable CEO of the organization. He gets to have his cake and eat it too, giving him probably his best shot at going hard after his vision of a multiplanetary species.

Founded in 2002, SpaceX is the only company building the integrated hardware and software infrastructure of the future across space, connectivity, and AI. At our core, we are builders. We design, manufacture, launch, and operate products and services built on cutting-edge technologies, including the world’s most advanced rockets and spacecraft. We safely and reliably transport astronauts, satellites, and other payloads on missions that benefit life on Earth. Since 2023, we have launched more than 80% of mass to orbit for the world each year with an over 99% mission success rate with Falcon rockets. We also operate a high-speed, low-latency global broadband data and communications network powered by approximately 9,600 Starlink broadband and mobile satellites in Low-Earth Orbit, delivering connectivity to millions of consumer, enterprise, and government customers across 164 countries, territories, and other markets, as of March 31, 2026. Using our dedicated satellite-to-mobile constellation, we offer connectivity services, supplementing terrestrial networks and substantially reducing mobile “dead zones” across approximately 30 countries.

With the potential to improve both space exploration and life on Earth, AI accelerates SpaceX’s mission to make life multiplanetary, to understand the true nature of the universe, and to extend the light of consciousness to the stars. xAI, which was founded in 2023 and acquired by SpaceX in early 2026, is now an integral pillar of our vertically integrated company. We are rapidly constructing AI compute infrastructure—starting on Earth with the goal of extending to space—at industry-leading pace and cost efficiency. Our infrastructure supports training and inference for Grok, which has emerged as one of the world’s most advanced frontier models. Grok is designed as a truth-seeking AI model, built on our founder Elon Musk’s mission to enable humanity to understand the universe. We believe that accomplishing this mission requires a truth-seeking approach to AI.

We define truth seeking as the active, relentless pursuit of what is objectively true about reality, and grounded in evidence, logic, empirical data, and first principles thinking. Our goal is to understand and explain what the universe appears to be doing, as accurately as current knowledge allows. Within two years of its initial model release, Grok achieved frontier-level performance in scientific reasoning, as measured by its GPQA Diamond score, an industry benchmark that evaluates AI models on a standardized set of questions written and validated by experts, on a faster timeline than reported by other leading model providers. Grok also benefits from integration with X, our real-time information, entertainment, and free speech platform, which serves as a foundational distribution and data engine for our AI ecosystem and further enhances Grok’s truth-seeking objective.

The story is not without its weaknesses, namely trying to position Grok as a success, at a time when the majority of the team has left. A lot of their ability to sell the vision will have to rely on more than just "Musk will figure it out," hence why they included the potential Cursor deal as an important part of the investor information.

Our innovations and technological advancements are redefining industries on Earth, while we aim to create new ones on the Moon, Mars, and beyond. We are truly building the infrastructure of the future.

• Space. SpaceX is the only company that has cracked the code on accessing space at scale, revolutionizing an industry characterized by decades of stagnation, risk aversion, and economically perverse cost structures. SpaceX upended this paradigm through the application of first-principles thinking, which rejects industry assumptions and builds solutions based on the fundamental laws of physics. Our intense, mission-driven, engineering-first culture and focus on extreme vertical integration have propelled us to achieve what many deemed impossible. We pioneered high-cadence, reliable, and affordable access to space with our Falcon family of rockets. In 2015, we established at least a 10-year lead over the industry by successfully landing our first Falcon 9 booster back from space before anyone else. Space flight that historically cost billions per launch now costs in the tens of millions, fundamentally reducing the cost of space access and providing the opportunity to build new enterprises in space.

• Connectivity. Since activating service for customers in 2020, Starlink has rapidly expanded global access to high-speed internet, prioritizing underserved rural and remote communities worldwide. While building terrestrial networks in such communities can be prohibitively expensive, Starlink is capable of delivering broadband connectivity anywhere on Earth with just a Starlink Kit. As of March 31, 2026, we had approximately 9,600 Starlink broadband and mobile satellites in Low-Earth Orbit, operating the world’s most advanced broadband constellation providing internet connectivity to approximately 10.3 million Starlink Subscribers across 164 countries, territories, and other markets. In January 2024, we also began deploying our Starlink Mobile constellation that utilizes separate Starlink satellites with satellite-to-mobile capabilities, substantially reducing mobile “dead zones” around the world. As of March 31, 2026, our dedicated satellite-to mobile constellation of approximately 650 V1 Mobile satellites provides satellite-to-mobile data, over-the-top voice, and messaging services to approximately 7.4 million monthly unique devices across approximately 30 countries.

• AI. We were the first company to deploy a coherent gigawatt-scale AI training cluster. For complex reasoning and agentic workloads, compute is directly correlated with the quality of intelligence and task completion speed. In under two years, we have established a dual advantage in both cost efficiency and deployment speed at scale. By owning the compute infrastructure and vertically integrating across the full AI stack, we can train and iterate our frontier models at lower cost and higher velocity and accelerate development cycles. This eliminates external bottlenecks and drives rapid, continuous improvements in model performance. We believe this combination of our state-of-the-art AI compute infrastructure, our truth-seeking frontier model, and our access to real-time data on X creates a significant strategic advantage. Our integrated AI platforms across Grok and X have over 1.3 billion supported accounts active in the last twelve months ended March 31, 2026, including approximately 550 million MAUs and generating approximately 350 million daily posts. Of our MAUs, we had approximately 117 million MAUs that used Grok’s AI features as of March 31, 2026. Grok’s deep integration with X enables freshness, relevance, and contextual awareness that we believe is a competitive differentiator. This direct, real-time access to the information and human discourse on X enhances Grok’s truth-seeking capabilities by grounding outputs in up-to-date knowledge and diverse viewpoints. As a result, we believe Grok can deliver the most objective and relevant insights and best serve high-frequency, high-value use cases across consumer and enterprise AI applications.

If AI is a weakness, connectivity is arguably the single biggest reason why we can “believe” in a potential successful outcome here. The team has done an outstanding job in providing a genuinely great service with Starlink and breaking almost every rule of the telco scaling framework. It’s not difficult to see that if somebody can get inference from space done in the next years, it’s most likely to be SpaceX.

Read more

Let's review their announcements, in the order that Google themselves presented them:

AI momentum across the full stack

These stories of how people are using AI are the best measure of progress. To understand the scale at which people are adopting AI, there is another great proxy — tokens, the fundamental units of data our models process, many representing a problem being solved.

Two years ago, we were processing 9.7 trillion tokens a month across our surfaces — a huge number. Last year at I/O, that grew to roughly 480 trillion tokens. Fast forward to today, that number jumped 7x to over 3.2 quadrillion per month.

It tells an important story about our products and how others are building as well — especially developers and enterprises:

• Over 8.5 million developers are now building new apps and experiences with our models monthly.

• Our model APIs are now processing roughly 19 billion tokens per minute.

• Over the past 12 months, over 375 Google Cloud customers each processed more than one trillion tokens, representing incredible demand for AI from across industries.

The AI token factory is no longer a meme, as the total consumption of tokens has ballooned exponentially. The figures themselves, however, are quite revealing, when we account for stats like the OpenRouter usage (where Gemini models have traditionally been widely adopted):

According to this, in May so far the Gemini models have been used to generate around 11T tokens, which can be seen as quite big (as much as the full monthly usage two years ago!) or quite a significant drop (0.35% of the current monthly run). What this tells us is that the significant reduction in Gemini model usage for coding is not imaginary.

Subscribe now

So there is a place where all of these tokens are going, and it's clearly not in coding usage. Google has been redirecting its resources toward powering AI experiences in its search products, most of it essentially subsidized.Natural, conversational AI in products

There’s also a lot of latent productivity to be unlocked. Over the last year, we’ve been bringing the ability to have more natural conversations with Gemini directly inside our products. Recently, Maps got its biggest upgrade in a decade, including a new feature called Ask Maps. People are using Ask Maps for more complex, and much longer questions.

Now we’re bringing more natural conversational AI to more products.

Ask YouTube

People come to YouTube everyday to ask a lot of questions. There’s a lot of great videos, but sometimes it’s hard to know where to start.

Ask YouTube entirely reimagines the experience, making information much more digestible and easy to navigate. You’ll see videos that best match your interest, and most importantly, it jumps right to the part of the video most relevant to you.

We’re starting to test Ask YouTube now, and it will roll out broadly in the U.S. this summer.

Those tokens are now also going to be burned across traditional Google surfaces such as Maps and YouTube. On the question of why I need to go specifically to YouTube to find these insights instead of bringing them into the supposedly widely adopted Gemini app with 900M users, well, that's a question for another day.

Voice-powered Docs Live

There are a lot of times I want to get things done at the speed of my voice. That is much more possible today thanks to technical leaps in our audio models.

A new feature called Docs Live takes this to another level. To create a doc with Gemini before, you had to type out a precise prompt. With Docs Live, you can just verbally “brain dump” whatever is on your mind, and let Gemini do the rest. Here’s a demo in real-time:

In the future, you’ll be able to create new docs and edit them directly, all with your voice. Docs Live is rolling out for subscribers this summer, and powerful voice capabilities will come to Gmail and Keep then too.

We are now into the third announcement for the developer event and Sundar is pitching how you can yell at a docs file and get a formatted version out of it.

While for developers it might not be very exciting to hear the company is going to be spending six times more CapEx this year, it's definitely bullish for The Infra Play portfolio focused on the bottlenecks of the AI buildout:

Gemini Omni looks genuinely exciting, assuming it can actually be productized properly. It’s difficult to overstate how badly Google misplayed the Nano Banana lead that they had, as they offered very limited generations and censored usage significantly. OpenAI’s own image models today are able to deliver much more interesting visual work (including for public figures) without any of these limitations. So a great video world model like this will likely struggle to see actual adoption if the same poor execution post-launch continues.

We are at the sixth announcement, which is focusing on corpo compliance for fake images. Is Google even trying to grab the attention of developers?

At long last, an actual announcement that matters (besides the TPU ramp-up, thank you on behalf of my semiconductor bags). Traditionally the Gemini Flash models have been seen as a great value, offering reasonable capabilities at a significantly lower cost.

According to the internal benchmarks by the team, the model is trading blows with the other leading frontier models.

The first thing that we need to acknowledge is that the "value play" here is completely gone, as the price per token has been raised almost three times and 3.5 Flash ends up at a significantly higher cost envelope than ever before.

If the frontier models are expensive but get the most difficult jobs done, then 3.5 Flash is not making a meaningful step for Google in either direction. It remains more expensive to run than the open-source models and the latest xAI Grok version, while not being best in class either (not even on benchmarks).

Cursor also runs its own evals (CursorBench) and the results do not look particularly exciting either. If anything, the latest Composer 2.5 model completely outplays the Gemini launch, which potentially could be a very interesting early sign of success for SpaceXAI if they acquire Cursor.

User feedback has not been kind either:

Which brings us to the most “hidden” announcement, the push for Antigravity as a product:

Google will actually retire Gemini CLI, an open-source coding harness that was widely used by the open-source model providers to build their own offerings on top. This is replaced with the closed-source Antigravity CLI (born out of the acquihire of Windsurf), which will now be pushed further as the "best way" to utilize Gemini models, which is ironic since the public pushback has been quite negative.

This is not surprising, since Google is essentially getting outplayed on the "surfaces" side, as Anthropic was able to heavily popularize the usage of their Claude app for chat, Cowork, and Code, while OpenAI is trying to expand on its successful launch of the dedicated coding application Codex. If SpaceXAI is able to acquire Cursor, they will also offer their own widely adopted coding surface, leaving Google as the only player struggling to consolidate around a primary product.

There is one silver lining here, which is enterprise agentic usage.

As long as Gemini remains usable for enterprise agentic use cases, the GCP team will be able to keep selling it aggressively at a discount. Still, unless something completely changes with the Gemini 3.5 Pro launch, it's becoming obvious that Google is struggling to be competitive in coding, and at a time when this is leading to exploding revenues at Anthropic and OpenAI, it's a mistake that might be very difficult to recover from. Meta is considering it so existential to training new models that they are now recording all of the computer use of their developers in order to somehow catch up on the high-quality training data required to progress on this (and Musk is willing to spend $60B for Cursor).

The key takeaway

For tech sales and industry operators: Dell is a stable, slow-moving organization that rewards a specific kind of operator: one who delivers outcomes without needing the company to move fast. Realistically speaking, its headline growth is tied directly to repackaging NVIDIA hardware and installing it for customers, while keeping a small portion as margin for itself. The work itself is hard (rack-scale integration, liquid cooling, power delivery, and on-site deployment at speed are genuinely difficult), but only the select few driving the largest customers will benefit from this turnaround. The window for individuals to join the business and benefit is likely over (unless you want to grind the next regular server replacement cycle), similar to the Oracle OCI opportunity in 2025.

For investors and founders: On paper, Dell is a fair business at a fair price, albeit not a wonderful one, to borrow Buffett's framing. It has built a world-class supply chain and proven it can reprice and reroute under stress, and that is not nothing. But value flows to whoever solves the hardest physics, and here that is the chip and the memory, not the chassis around them. For founders, the lesson is to own the bottleneck, not the assembly. If your company's role can be described as integrating someone else's breakthrough, your margin will always be set by someone else's scarcity. For investors, Dell is the lower-risk seat in AI infrastructure: no foundry or CapEx exposure, a cheap valuation, and management beholden to shareholders, with a capped margin as the price of that safety. Personally, I think there are better uses of capital at this stage of the buildout.

The sleeping giant

Jeff Clarke: FY26 was a defining year in our company's history. We delivered record full year revenue and EPS. Revenue reached $113.5 billion, up 19% and EPS grew 27% to $10.30. We converted that performance into a record annual cash flow over $11 billion and returned $7.5 billion to shareholders, including 54 million shares repurchased more than doubled last year. The AI opportunity is meaningfully growing and transforming the company. In FY26, we closed $64.1 billion in AI orders shipped $25.2 billion and exited with a record $43 billion in AI backlog. Powerful proof points that our engineering leadership and differentiated solutions are winning.

We are executing with discipline and speed. We are gaining share in our PC business and strengthening ISG with strong margins in traditional servers and storage, all while positioning the company for the AI era. It was a monumental year. We exit with strong momentum, and I couldn't be more proud of this team. With that, let me turn to the key highlights for the quarter. We delivered a record quarter. Q4 revenue was $33.4 billion, up 39% and earnings per share was $3.89, up 45%, driven by disciplined execution and demand for our AI solutions. While operating in a dynamic environment, we saw a record cash flow generation and above trend capital returns for shareholders.

Now let's move to AI. We kept an already strong year with an exceptional quarter for AI, record orders and broad-based demand. In Q4, we booked $34.1 billion in AI orders, evidence that demand is accelerating as customers deploy AI at scale. We shipped $9.5 billion in AI servers in the quarter. We exited Q4 with a record $43 billion in AI backlog, and our pipeline continued to grow sequentially even after converting $34.1 billion of orders, a clear sign of sustained momentum. For the full year, AI orders reached $64.1 billion. Our customer base surpassed 4,000 with growth across neoclouds, sovereigns and enterprise customers, evidence that demand is broadening across all customer types.

We're winning for the reasons we've outlined all year. Engineering for performance and time to market while optimizing TCO for AI workloads, deployment and installation at speed and scale, ongoing life cycle support that keeps clusters up and running and DFS financing. We're doing this with discipline. Profitability is in line with our mid-single-digit operating margin target. We like our position, the line of sight we have with our backlog and pipeline and the advantage our scale and supply chain bring.

An AI server is a high-performance system built specifically for artificial intelligence work. Unlike regular servers that handle a wide range of tasks, Dell's PowerEdge XE series is designed from the ground up with GPUs in mind. It prioritizes high GPU density, fast connections between the accelerators, strong power delivery, and advanced cooling, including liquid cooling for the densest setups. This makes the servers well suited for both training large models and running inference at scale.

The AI server business line now generates $50B annually, after another exceptional year by the team.

Read more

Over April I rebuilt the positions that now form the Q2’26 Infra Play portfolio (which hopefully will not require a complete directional shift, but you never know what happens with Taiwan). There are some expected names, as well as many new ones.

Power, cooling, custom silicon, memory, lithography, silicon photonics foundry, InP tools, connectivity silicon. Each layer is a separate bottleneck, each bottleneck is structurally short on supply, and each name is the cleanest expression of the layer it sits in.

I recently published the Q2'26 Infra Play portfolio and it's been quite the run, to say the least. Tech market valuations exploded, driving strong support for the "bottlenecks of AI" thesis. The overall return in April is at 47%, which is outstanding by any conventional measure.

There is, however, a bit of a but here.

My actual P/L looks slightly different due to different entries and exits, but obviously making a 25.6% gain on this portfolio within a quarter is not exactly slim pickings.

The Q1 version of the portfolio did very well considering all of the events (tariffs, war, SaaSpocalypse, stagflation), but most of it was carried by the exceptional rally of SanDisk. To make things a bit worse, if I had remained with the same positions going into Q2 and not sold, we would be looking at more than a 100% gain in April alone.

The real driver behind the massive opportunity in the last weeks has been the rally in MU, SanDisk, SK Hynix, Intel, and AMD. The thread between them has been the massive demand for compute, but the star of the show has truly been memory as the key bottleneck of scaling AI.

This raises the question: are we in a memory supercycle, and how do we play it?

Read more

SpaceXAI has signed an agreement with Anthropic to provide access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers.

Built from the ground up in record time, Colossus delivers unprecedented scale for AI training, fine-tuning, inference, and high-performance computing workloads. Colossus 1 features over 220,000 NVIDIA GPUs, including dense deployments of H100, H200, and next-generation GB200 accelerators. The cluster delivers extreme parallel performance for large language models, multimodal systems, scientific simulations, and generative AI at frontier scale.

Anthropic plans to use this additional compute to directly improve capacity for Claude Pro and Claude Max subscribers.

As part of this agreement, Anthropic also expressed interest in partnering to develop multiple gigawatts of orbital AI compute capacity.

The compute required to train and operate the next generation of these systems is outpacing what terrestrial power, land, and cooling can deliver on the timelines that matter.

SpaceX is the only organization with the launch cadence, mass-to-orbit economics, and constellation operations experience to make orbital compute a near-term engineering program rather than a research concept. If engineering challenges can be overcome, space-based compute offers near-limitless sustainable power with less impact on Earth.

Musk has been very aggressive in making rapid changes at xAI, which means he has identified this as the place with the largest bottlenecks in his cluster of companies.

It’s been barely two months since SpaceX and xAI were combined. Then essentially the full founding team left (or was made to leave). Two weeks ago a critical deal was announced: Cursor will train models in the Colossus data center, and will either be acquired for $60B or have its year of work bought for $10B.

Now all of a sudden Anthropic, a company that Musk has had repeated clashes with, gets access to the most important commodity in the world right now, compute?

Let’s try to untangle this.

Read more

I've been tracking hyperscaler earnings since the AI infrastructure buildout started in 2024, and this week was easily the highest-stakes moment of the race so far. The market needed two key pieces of information: their CapEx plans to add more hardware and whether sales performance justified the investment.

Well, we got our answer and the short summary is that there are no breaks on the AI train right now.

GCP

Sundar Pichai, CEO, Alphabet and Google: Hi, everyone, and thanks for joining us today. It was a terrific quarter for Alphabet. Our momentum was on full display at Cloud Next last week, and the month of May brings even more with I/O, Brandcast and GML. I hope you’ll tune in to see our progress.

It’s clear that our AI investments and full stack approach are driving performance across our business.

In Search & Other, revenue grew 19%. People love our AI experiences like AI Mode and AI Overviews, and they’re coming back to Search more.

Cloud accelerated again this quarter due to strong demand for our AI products and infrastructure. Revenue grew 63%, exceeding $20 billion for the first time, and our backlog nearly doubled quarter on quarter to over $460 billion. Gemini Enterprise is seeing tremendous momentum, with 40% growth quarter over quarter in paid monthly active users.

In Subscriptions, this was our strongest quarter ever for our consumer AI plans, primarily driven by adoption of the Gemini App. Overall, the number of paid subscriptions has now reached 350 million, with YouTube and Google One being the key drivers.

And our AI models have great momentum. Our first party models now process more than 16 billion tokens per minute via direct API use by our customers, up from 10 billion last quarter.

After two years of struggling to meaningfully accelerate growth, GCP has had an outstanding six months.

Sundar Pichai: Starting with our AI infrastructure. It’s the foundation of our full stack approach to AI, driving customer growth and product adoption.

Our custom TPUs, Axion CPUs, and the latest NVIDIA GPUs continue to form the industry’s widest variety of compute options.NVIDIA GPUs are a core part of our AI accelerator portfolio, and we will be among the first to offer NVIDIA Vera Rubin NVL72, in addition to the Blackwell- and Hopper-based instances already available.

At Cloud Next, we introduced our eighth generation TPUs, individually specialized for training and serving, and able to take on the most demanding agentic workloads.

TPU 8t provides high performance model training with three times the processing power of Ironwood and two times the performance.

TPU 8i delivers cost effective, low latency inference, with 80% better performance per dollar than the prior generation.

This exceptional infrastructure powers our world class AI research. That includes models and tooling, which continue to progress really well.

While there are meaningful concerns about how TPUs will remain competitive, particularly if Google shifts production away from Broadcom, every accelerated computing rack is worth its weight in gold at a time when Nvidia accelerators remain at full capacity and premium pricing.

Sundar Pichai: Gemini 3.1 Pro continues to push the frontier in reasoning, multimodal understanding and cost. We’ve quickly expanded the Gemini 3.1 series of models to offer more choices for developers, including our cost efficient Flash models.

3.1 Flash Live, our latest audio model, has improved precision and reasoning, making voice interactions more natural and intuitive. It’s now powering conversational features in Search and the Gemini app. Speech to text is now available in 70 languages.

And with 3.1 Pro, our Deep Research agent got a big upgrade, including MCP support and native visualizations.

Our generative media models are incredibly popular. Lyria 3 has generated over 150 million songs since launching on the Gemini App.

Nano Banana 2 reached one billion images in nearly half the time of Nano Banana 1.

And Veo 3.1 Lite is our most cost efficient video model to date.

On top of this, we launched Gemma 4, our most intelligent open model. It’s been downloaded over 50 million times in just a few weeks. In fact, our open models have now been downloaded over 500 million times.

On the model side, Google has clearly taken a step back in the eyes of the "AI power user," as Anthropic and OpenAI continue to trade blows over the best coding model in the world. Still, Google has optimized for a slightly different game, using Gemini for easy API integrations and more straightforward agentic use cases like customer support. Between efficient models, their own silicon, and aggressive pricing, Gemini has clearly found enterprise product-market fit.

Sundar Pichai: Looking ahead, we are focused on pushing the next frontiers of foundation models, including intelligence, agents and agentic coding. And we are using the latest technologies to transform how we work as a company.

For example, with Antigravity, we are shifting to truly agentic workflows. Our engineers are now orchestrating fully autonomous digital task forces, and building at a faster velocity. Much more to come here.

Nobody believes in Antigravity right now except Sergey Brin.

Sundar Pichai: We continue to invest in improvements to AI Overviews, which are driving overall Search growth, and we are also seeing strong growth in both users and usage of AI Mode globally.

Personal Intelligence expanded broadly in the U.S. and we are seeing people ask more personal questions, and getting responses that are uniquely relevant to them.

We also shipped agentic experiences, like restaurant booking, to new countries and new multimodal capabilities like Search Live globally.

We’re also continuing to improve efficiency and speed. Even as we have brought new AI features into our results page, we have reduced Search latency by more than 35% over the past five years.

And since upgrading AI Overviews and AI Mode to Gemini 3, we have reduced the cost of core AI responses by more than 30%, thanks to continued hardware and engineering breakthroughs.

AI Overviews in Google Search, however, remain probably the most powerful "free AI" usage in the industry right now. It has mostly killed the value proposition of dedicated tools like Perplexity, since most quick searches that benefit from AI summaries already work perfectly out of the box with AI Overviews. Whether this drives meaningful commercial conversion through advertisements or subscriptions is a separate question.

Sundar Pichai: Google Cloud is differentiated because we are the only provider to offer first party solutions across the entire Enterprise AI stack. Our growth in revenue, operating margin, and backlog highlights this differentiation. Our Enterprise AI solutions have become our primary growth driver for Cloud for the first time. In Q1, revenue from products built on our gen AI models grew nearly 800% year over year.

We are winning new customers faster, with new customer acquisition doubling compared to the same period last year.

We are seeing strong deal momentum, doubling the number of $100 million to $1 billion deals year on year, and signing multiple billion dollar plus deals. And we are deepening relationships with existing customers. Customers outpaced their initial commitments by 45%, accelerating over last quarter.

At Cloud Next last week, we introduced hundreds of new capabilities across our vertically optimized AI stack that are designed to work together for our enterprise customers.

Coming back to GCP, we are seeing a number of strong signals, including new logo acceleration alongside more large deals. At hyperscaler scale, a big-ticket deal measures in the billions.

Sundar Pichai: We introduced a new Gemini Enterprise Agent Platform that empowers users to build, orchestrate, govern and optimize agents with the controls that enterprise customers need.

Along with new capabilities in Gemini Enterprise app — like Projects, Canvas, Long Running Agents and Skills, every employee can build agents.

In Q1, Gemini Enterprise paid monthly active users grew 40% quarter over quarter. That includes major global brands like Bosch, Citi Wealth, Merck, and Mars, Incorporated.

Our partner ecosystem plays an increasingly critical role in driving Gemini Enterprise adoption. We saw 9x year over year growth both in seats sold with partners, and in the number of partners adopting it for internal use.

This momentum is leading to accelerating usage of our models.

Which brings us to the actual adoption momentum, with large companies still trying to figure out how to scale AI agents for internal use cases while juggling difficult internal budgets. I am directly involved in one of the accounts mentioned here, and I have observed this play out firsthand.

Sundar Pichai: Over the past 12 months, 330 Google Cloud customers each processed over one trillion tokens. 35 reached the ten trillion token milestone.

To give agents business context from enterprise data to help them reason intelligently, we introduced a new Agentic Data Cloud. It includes a cross cloud Lakehouse, Knowledge Catalog and Deep Research Agents, which combine research and analytical skills.

As an example, using our data cloud, American Express is enabling agentic commerce at scale by moving an enterprise data platform, along with hundreds of production applications, to BigQuery. Vodafone is proactively resolving outages, automating network planning and precisely targeting capacity.

Enterprise data has become critical for agents to reason. Our strength with Big Query and Gemini Enterprise has led Gemini powered workflows in Big Query to grow over 30X year over year.

The enterprise AI upsell is, of course, not limited to Gemini inference; it extends across the full technology stack around those applications. BigQuery has been a significant upsell for GCP reps, but I don't know how "sticky, sticky" it will be long term. New players like ClickHouse have been winning workloads by offering better performance at half the price, and that's not something cost-sensitive engineering teams will ignore.

Sundar Pichai: As cybersecurity threats from the use of AI models accelerate, our expertise in AI and Cybersecurity is driving strong demand for our Agentic Defense offerings. In March, we closed the acquisition of Wiz, a leading cloud and security AI platform, which is an incredible fit for the moment we are in. We’ve seen tremendous interest from customers in our unique cybersecurity and AI products and services to protect their IT estate. The performance of Wiz so far has exceeded our expectations.

Together with Google’s Threat Intelligence, Security Operations and AI models, Wiz is helping organizations detect, prevent and respond to threats.

We introduced new Gemini powered agents for threat detection, continuous red teaming, and automated remediation to protect software code and Cloud systems

Customers like Deloitte, Priceline and Shell are using our agentic defense to strengthen their security posture. All of this is powered by the AI Infrastructure I mentioned earlier.

Interestingly, cybersecurity makes a return to an Alphabet earnings call, with a high-level blurb but no meaningful adoption figures. I still think the GCP team doesn't know how to drive proper security deals, and unless something changes, this will remain an underperforming part of the portfolio, regardless of the strong fundamentals.

Michael Nathanson (MoffettNathanson): Thanks. One for Sundar, one for Philipp.

Sundar, if I can connect Brian’s question and Eric’s question and go a little bit higher, I want to understand how are you deciding, how are you allocating which divisions and projects get excess capacity, even though you’re constrained? How do you decide between all the internal projects you have and the external projects? What types of screens are you running to decide who gets the ample capacity?

Then for Philipp, I noticed that you said this on the Gemini app, there’s more and more images that come to you in the shopping journey. Can you talk about your thoughts about adding advertising on that app and what’s guiding your decision making here on adding ads on Gemini? Thanks.

Sundar Pichai, CEO, Alphabet and Google: Thanks, Michael. I think, great question on an ongoing basis, and I’m looking forward to Gemini helping me more and more as I’m thinking that through.

Look, I do think that the foundation where we start with it is, what do we need from a R&D standpoint to develop models at the frontier. So what do you need for training these models? And so, effectively, the compute needed for GDM, because it’s a foundation for everything we do. So that’s a core principle with which we operate.

Then obviously, the ability to plan ahead. We do long range plans on our core areas, be it Search, be it YouTube, and so on, as well as what we see in Google Cloud.

And obviously, in Google Cloud, we are providing enterprise AI solutions, which this quarter had an 800% year on year increase from the prior year. So we’re seeing strong demand for Gemini enterprise, our AI solutions there.

We see strong demand for infrastructure in Google Cloud. And as I said earlier, in certain cases, we are seeing demand for TPU hardware; TPU hardware in other data centers as well.

So we are modeling these out and working to allocate across these areas. Obviously, we are compute constrained in the near term. And as an example, our Cloud revenue would have been higher if we were able to meet the demand.

So we are working through that moment and we are investing, but we have a robust long range planning framework, and we see extraordinary opportunities ahead, and we are allocating with that framework in mind.

Philipp Schindler, SVP and CBO, Google: And to the second part of your question, as I said in my previous answer, we are obviously focused on the user first and creating a really great user experience with all of our product, especially on newer products; and specifically on monetization in the Gemini app, our focus right now is on AI Mode.

But it’s fair to say that we really believe the format that works well in AI Mode would transfer successfully to Gemini app. And so, today in the Gemini app, we’re focused on the free tier and subscriptions and our AI plans were a sizable contributor to our Google One revenue growth.

But let’s also be clear, Ads have always been a big part of scaling products to reach billions of people, and if done well, Ads can be really valuable and really helpful commercial information. At the right moment, we’ll share any plans, as we have said, but we’re not rushing anything here.

It's important to remember that GCP has historically been a largely ignored team inside Alphabet, with most of the leadership coming out of the Ads business. The beauty of sales performance is that at some point they can no longer ignore you, especially when the commercial teams are delivering those large commits with significant compute reserved for customers.

Justin Post (BAML): Thank you for taking my question. I expect a lot of interest in your TPU sales. Can you help us think about how you’re thinking about the opportunity there? And then maybe break down the backlog growth a little bit between TPUs and Cloud.

And then the second question. Just thinking about the margins on these big, generative AI Cloud deals, how do you think about these $100 billion deals coming in and the margins associated with those? Can they be similar to your Cloud business as it is? Thank you.

Sundar Pichai, CEO, Alphabet and Google: Look, overall, I would say we see tremendous interest and there’s tremendous demand for both AI solutions, as well as AI infrastructure, including massive interest in our GPU offerings, as well as our TPUs. So we are proud that we can provide customers with the breadth of our offerings and meet them in terms of where their needs are.

And maybe I’ll pass it to Anat to give some color on the backlog growth.

Anat Ashkenazi, SVP and CFO, Alphabet and Google: So the backlog, the TPU hardware agreements that Sundar referenced in his prepared remarks are reflected in our Cloud backlog, the $462 billion, although the majority of the backlog is still GCP agreements.

Now, as you think about the total backlog, just over half of it will convert to revenue in the next 24 months. And the TPU hardware sales, more specifically, we expect a small percent of them to see coming through as revenue later this year, and then the majority to be realized as revenue in 2027.

In the next two years, GCP will be recognizing $19B of revenue every month even if they don't sell a single new deal (which they seem to be adding to aggressively). We call this world-class in this business, shades of the glory days of Azure. This is not too surprising, as the company recruited a lot of sales leadership from Microsoft, which has brought both benefits (higher expectations) and drawbacks (quality of life and very aggressive growth targets).

The market has mostly rewarded them for this level of performance, and it's not difficult to see why the team is bullish on the next 24 months.

Read more

The actual impact was limited. The war was mostly a one-sided affair, outside of the closure of the Strait of Hormuz. Still, markets did experience a shock, followed by an aggressive rebound.

Depending on how the next few weeks play out, the following companies are on the list to be evaluated for the Q2 version of this portfolio: Micron (MU), Broadcom (AVGO), NVIDIA (NVDA), Alphabet (GOOGL), Constellation Energy (CEG), ASML, Oracle (ORCL), Microsoft (MSFT), AMD, TSMC (TSM), Core Scientific (CORZ), Solaris Energy Infrastructure (SEI), Astera Labs (ALAB), Vertiv (VRT).

One month ago I temporarily exited the markets, as there was no edge to be had in this environment unless you had direct access to the administration ahead of significant announcements. Insider trading has never been more blatant or difficult to navigate unless you like gambling. This is not to say I’m complaining (humans will always be incentivized to take advantage of information asymmetry), but rather to flag that any investments around big events require you to calibrate for the cost of missing vital information in advance.

Over April I rebuilt the positions that now form the Q2’26 Infra Play portfolio (which hopefully will not require a complete directional shift, but you never know what happens with Taiwan). There are some expected names, as well as many new ones.

Power, cooling, custom silicon, memory, lithography, silicon photonics foundry, InP tools, connectivity silicon. Each layer is a separate bottleneck, each bottleneck is structurally short on supply, and each name is the cleanest expression of the layer it sits in.

Let’s get into into this not financial advice.

Read more

In my previous deep dive, I noted that the flood of money into private markets has been both a positive and a negative:

Many of the AI-native companies mostly exist to steal money from their investors in order to fund a LARP founder lifestyle. It's harsh but it's true. Still, for those that are genuinely great new entrants in their categories, there are many interesting attributes to those orgs that are not seen in their incumbent competitors.

While there are obvious cases of companies pushed hard with media exposure despite offering unoriginal, pointless, and ultimately insecure products (Cluely, Delve), there are also plenty with a genuine shot at being generational.

To be fair, the odds of a generational AI-native company emerging are obviously high when almost every new company receiving venture capital today is AI-native.

More interestingly, for the existing crop of massive private companies (Anthropic, OpenAI, Databricks, SpaceX), the funding flowing into late-stage rounds has dwarfed anything we're seeing in the public markets. An IPO today in cloud infrastructure software is seen as a burden rather than the most exciting outcome for a company. This dynamic might finally change in the next 12 months if the "private whales" take the leap into the public markets, starting with SpaceX in the summer.

Subscribe now

This will be critical to reshaping where venture dollars flow, as the top 20 deals in 2025 captured almost half of available liquidity. While there are fair justifications given the compute demands of the whales, realistically this is not beneficial for the broader startup ecosystem, if not outright detrimental.

The flip side is that the time to scale to an outstanding valuation is shortening dramatically. Whether this makes a material difference (nobody is paying $50B to acquire Cursor, and xAI had to be bailed out through SpaceX) is a different topic.

Well, I guess there's one person willing to pay even $60B.

This is a deceptively simple framework, but I think it’s a fair way of looking at things in the current context.

I’m looking for an SDR who wants to make it in AI-native security with Octane Security.

What you’ll do: Own business development with startups and companies bringing secure code review into their workflows. Hit your numbers and you’ll start leading full sales cycles fast, on a direct path to promotion.

What we’re looking for: You’ve worked with developers. You understand security, and you see what AI is doing for teams that move early on the right tools. You know nothing gets handed to you. You show up, you build, you rise to the occasion every day.

Location: NY or SF preferred. Remote considered if you have the track record and the habits to back it up.

Think that’s you? Reply to this email or DM me, referrals welcome too.

The problem for the application layer is defending why anyone should pay for workflows that can either be replicated internally or handled within the model provider itself. Harvey’s legaltech challenge is a clear example.

The problem for infrastructure plays (and cloud infrastructure software specifically) is whether their most important market (enterprise) sees meaningful benefits from adopting their platform. Cursor vs. Claude Code illustrates this well: both are valid approaches, with Cursor scaling coding agent access across organizations at enterprise-grade reliability and security, while Claude Code takes a different path to the same buyer.

Although it's easy to be bearish on the application layer, that view is mostly driven by investors being limited to public markets (the SaaSpocalypse) versus private markets, where we're seeing accelerated outcomes and growth at an extraordinary pace.

The most interesting metric in this new application layer remains ARR per employee. For all intents and purposes, AI-native orgs run at significantly higher productivity per employee, which is also why the deep headcount cuts at public companies haven't drawn much pushback (most agree, if silently, that those orgs are deeply bloated).

Large incumbents in cloud infrastructure software have struggled to offer obvious "best-in-class" products across most of the AI infra challenges like agent security and observability. This leaves significant room for new AI-native companies to enter and separate themselves from the pack.

Regardless of how much we try to highlight the wider ecosystem, at the center of the AI scaling era there are two companies driving the majority of growth: OpenAI and Anthropic. The race for who scales the most in run-rate this year is wide open and highly volatile.

The competition is not sitting still on fundraising, but in practical terms we have not yet seen real challengers emerge from the neolabs. The most interesting one to track is Safe Superintelligence, thanks to Ilya Sutskever, arguably the individual who has had the biggest impact on deep learning progress in AI over the last 15 years. His primary "opponent" is Demis Hassabis, the single most motivated leader in AI research today, running not just DeepMind but also Isomorphic Labs. Funnily enough, OpenAI was co-founded by Musk as the counterpoint to Demis.

Between the SaaSpocalypse and the significant liquidity in private markets, it's fair to say things can be perceived as frothy.

The counterargument is that the growth rate justifies the premium. While that's clearly a risky bet (stalled adoption would be brutal on valuations), it does make sense.

The "growth slowdown" could also come from a different angle than weaker customer demand: overcrowding in certain markets. High valuations only make sense if someone is the obvious winner of a category, and the difficulty curve to become one is much higher today.

Frothy or not, the opportunity to enter the market from a funding perspective is still very much open, and the same goes for talented individuals looking to move to an AI-native company. That said, the window for the highest returns is likely closing in the next 12 months.

Read more

Read more

My strong conviction bet is that computing power is the most important resource, alongside oil. Everything that we build in the worlds of atoms and bits depends on it. What are the most important attributes of computing power?

Obtaining it (The art of turning sand into endless racks of deployed compute)

Utilising it (Building applications powered by AI)

Keeping it running (Orchestrating and observing resilient systems)

Protecting it (Securing computing power through software, people, and processes)

The most underappreciated and underreported part of these is how we keep all of this software running on a daily basis.

There was a time in the industry, when top performing engineering teams would talk about five, six or even seven 9s of uptime. Today, arguably the most important software company in 2026 is barely able to keep a 99%-ile uptime. To put it into perspective, this means that the main Claude website and the Enterprise API will on average be down for four to five full days every year.

This is a serious and difficult problem to solve. The market opportunity for this has been considered one of the most interesting TAMs in software, with the investor darling Datadog becoming the standard bearer of the industry in the public's consciousness.

Keeping systems running however is not something you just quickly abstract behind a SaaS layer. It's a bona fide engineering job, requiring cross-team collaboration, adaptability, domain knowledge and nerves to keep searching for a needle in a haystack while under immense pressure.

Grafana Labs is one of the most trusted tools in the industry to get the job done. Does this mean it's the next $10B IPO?

The key takeaway

For tech sales and industry operators: The Grafana Labs sales motion works because the company created a category and then built the commercial layer on top of it, which is the correct sequencing. The problem is that the category advantage is being arbitraged away faster than the sales org can convert it. What Grafana actually sells, at its most honest, is faster time to value when things go wrong rather than preventative control, something that is omitted in your usual pitch deck about an unlimited TAM. The open-source-to-cloud conversion path assumed that familiarity with the visualization layer would eventually pull users toward the managed stack. That assumption held when the core buyer had an emotional relationship with the open-source project. It does not hold for developers under 30, who never had that relationship to begin with and are querying logs with AI, setting up automated alerting, and expecting the system to surface the anomaly rather than expecting a human to notice it on a dashboard. In observability the biggest revenue is tied to the workloads that generate the most telemetry, something that makes AI-natives the most interesting greenfield segment in the industry. Unfortunately for Grafana Labs, the majority of business is coming from displacing alternative managed solutions, a business that will look a lot shakier in two years if they remain limited to competing as an efficient option for observability workloads.

For investors and founders: Grafana Labs is at peak success and peak strategic vulnerability simultaneously. The important question to ask is whether the core product premise, specifically that engineers need a sophisticated visualization layer to understand their systems, survives the next wave of AI development. I think it does not, at least not in its current form. The companies building the best AI agents today will eventually turn those agents on their own infrastructure, and the output of that process will be a ticket, not a dashboard. Grafana's real asset is the data access and the distribution network, not the interface sitting on top of it. The trick is that while being widely adopted as an open-source tool has given them a lot of distribution, capturing value without data gravity or workflow control is much more difficult. Data platforms, hyperscalers, cybersecurity companies and other observability vendors are all trying to compete for the same underlying telemetry. A standalone IPO at $1B ARR in the next 24 months would require the company to convince public market investors that dashboards are the future of observability at exactly the moment when agentic systems and data platforms are making that case hardest to defend. The founders did almost everything right. The next decision is the one that determines whether this is remembered as a great infrastructure software company or as a cautionary tale about what happens when a category leader waits too long to expand its category.

The grafanista way

NEW YORK CITY — February 3 — Grafana Labs, the company behind the open observability cloud, today announced a year of strong momentum across customer adoption, product innovation, and market recognition. Over the company’s past fiscal year, which ended Jan. 31, Grafana Labs continued to scale its business while helping thousands of organizations turn signals into action across increasingly complex, AI-driven systems.

Sustained Business Growth and Market Leadership

Grafana Labs entered 2026 with accelerating growth across every dimension of the business:

• Annual recurring revenue: Surpassed $400M, driven by continued expansion of Grafana Cloud and growing adoption among large, software-led enterprises.

• Customer growth: Now supports 7,000+ organizations worldwide, including 70% of the Fortune 50.

• Global expansion: The company added 100+ employees and established a new subsidiary in Japan to accelerate local market growth.

• Industry recognition:

  • Named a Leader in the 2025 Gartner® Magic Quadrant™ for Observability Platforms, positioned furthest for Completeness of Vision

  • Ranked #13 on Forbes’ Cloud 100, marking the fifth consecutive year on the list

  • Named a Leader in GigaOm’s Cloud Performance Testing Radar Report

  • Recognized as Best Observability Solution for Grafana Cloud in the 2025 DevOps Dozen Awards

This momentum reflects a clear trend: as software systems become increasingly distributed, AI-driven, and business-critical, teams are opting for open observability – built on open source, open standards, and open ecosystems – to avoid lock-in and move faster with confidence.

The challenge with doing deep dive articles on companies that are not yet public is that there is typically limited information to understand what direction the company is going towards. Let's start with this bland, but unusual press release, trying to communicate the progress made in their last fiscal year.

From a purely commercial perspective, the company is growing aggressively. Since the announcement of their secondaries sale back in October '25, the company added another $100M ARR in run-rate, something we will see reflected positively in the mood across the go-to-market org.

Powering the Next Generation of AI Companies

As AI-native companies push infrastructure to its limits, observability has become foundational. Over the past year, Grafana Labs deepened its role as critical infrastructure for teams building the next era of AI.

“The pace of innovation in software and AI is only accelerating, and our customers are feeling that complexity every day,” said Anthony Woods, co-founder of Grafana Labs. “This past year showed that open observability isn’t just a philosophy, it’s a practical advantage. From AI-native startups to the world’s largest enterprises, teams are choosing Grafana Cloud so they can understand their systems, control costs, and move at the speed their businesses demand.”

AI leaders, including Anthropic, Lovable, and OutSystems, rely on Grafana Cloud to operate complex, fast-moving systems in production. These teams use Grafana to:

• Monitor rapidly evolving AI workloads across metrics, logs, traces, and profiles

• Debug distributed systems where model inference, data pipelines, and user experience intersect

• Control telemetry cost while maintaining deep visibility as usage scales unpredictably

From training pipelines to real-time inference, Grafana Labs is helping AI builders keep their systems reliable, efficient, and understandable as complexity explodes.

The reality of the observability market today is that AI-natives are not really going towards Grafana Labs. The press release tries to tackle this immediately, claiming Anthropic and Lovable as important customers, but the reality is a bit more complicated than that, with a significant amount of workloads going to new players that are much more focused on the efficiency and speed angle of storing and accessing logs for these dynamic systems.

In observability for AI-natives today, what you’ll commonly see is that Datadog is leading in adoption (with an OpenAI consumption contract into the hundreds of millions), Sentry for the smaller startups without significant complexity and Clickhouse for companies that have significant telemetry and technical challenges.

Grafana Labs, Dynatrace, Elastic and Splunk have mostly struggled with that segment, partly because of complexity of the products and partly because of a curious aspect of software engineering, which we will simplify into “culture”.

Often the tools that software engineers choose, are driven more by trends, external influences and age of the user, rather than some perfect benchmark and a PoC experience.

Read more

Read more

For the purposes of this article, I’ll go over their early 2026 market update and how to interpret some of the trends we are seeing today.

If it’s not obvious, we are currently seeing a massive investment cycle in AI hardware and software, which is quickly trending towards the single biggest buildout that’s ever happened in the US. I would like to call it “western nations,” but the reality is that by all measurable metrics (GW of compute, early-stage investments, and revenue growth) the race for AI is an American moment in time.

The biggest signal of demand is driven by the hyperscalers, who have significantly stretched their investments into the infrastructure required to scale AI. The important thing is that this level of investment is increasing, something that just a few months ago was seen as impossible.

The talk of the "AI bubble" has also taken a backseat, as the leading companies in the industry are showing high growth (that appears quite durable) and we remain supply constrained. While outsiders with little understanding of where things are going might still blab about it, for all intents and purposes it's obvious that something different is happening right now (even if the classical euphoria meme is literally "this time it's different!").

A lot of the acceleration in the last few months is driven by "AI coding is both the present and the future of developers" becoming a consensus. While this was accompanied by fear campaigns of "all developers will be out of a job," the reality is that technical knowledge has never been more important, as your average company is now able to create and deliver software that previously was reserved only for organizations with tens of engineers.

Interestingly enough, agentic workflows are barely making a dent yet. For most practitioners it's obvious that human-in-the-loop remains very much mandatory for great outcomes; however, it's also not difficult to see this dynamic shifting quickly in a few years.

This has led to an explosion of new software being created. This is not the same thing as "valuable" new applications being created, but then again, agentic coding has also revealed how simplistic and easily replicable most of the software being sold today truly is.

Read more

Read more