Anthropic Files for IPO: Five Years from Leaving OpenAI to a $637 Billion Compute Lock-In
On October 26, Anthropic filed a prospectus with the U.S. Securities and Exchange Commission, seeking to list on Nasdaq under the ticker ANT. The filing discloses cumulative long-term compute arrangements of approximately $637 billion, including about $161.2 billion with Broadcom, at least $114.5 billion with Google, and at least $110 billion with Amazon. The article recounts its 2021 departure from OpenAI and how, after FTX's collapse, it accepted investments of up to $4 billion from Amazon and $2 billion from Google while being required to purchase the corresponding cloud resources, and states that revenue in the first half of 2026 was $16.286 billion with a book loss of $80.748 billion, of which $72.4 billion came from fair value changes on convertible bonds.
Why read this
AI score 75
Drawing on the long-term compute contracts disclosed in Anthropic's IPO prospectus, this piece shows how large-model companies become bound to the capital and infrastructure of cloud giants.
Topics
Body · AI translation
[[**Anthropic left OpenAI in the name of safety, yet in the compute race it has been deeply bound to giants such as Amazon and Google through investment and procurement contracts; its prospectus exposes massive long-term compute commitments, showing that the threshold for large-model innovation has risen to the hundred-billion-dollar level.**
**Key points**
**1. Safety ideals are forced to yield to compute costs**
After FTX collapsed, Anthropic accepted up to $4 billion from Amazon and $2 billion from Google, but the funds had to be used to purchase the corresponding cloud resources and chips, binding it to the giants' infrastructure.
**2. It rejected OpenAI's merger offer**
In 2023, OpenAI's board invited Dario to succeed as CEO and merge the two companies, but he refused; after Altman returned, he accelerated compute expansion, and Anthropic was forced to follow suit in Claude training.
**3. Enterprise-side commercialization has exploded but depends on rapid iteration**
It entered enterprise workflows through Claude Code and APIs; revenue was $4.585 billion in 2025 and $16.286 billion in the first half of 2026, with more than 11,900 customers at the $100,000 level; in Q3, more than 80% of revenue came from new models released less than six months earlier.
**4. The prospectus reveals capital binding and financial costs**
English text is being prepared.
This article comes from the WeChat public account: Beyond the Page, authors: Huahua, Banjun
Silicon Valley stories often begin with a parting of ways.
Five years ago, a group of core researchers at OpenAI chose to leave. They had different judgments about the direction of AI development and decided to start over for that reason.
That departure later gave rise to an AI company whose valuation and revenue both climbed rapidly: Anthropic.
In 2021, in an old factory building at Pioneer Square in San Francisco, Dario Amodei and his sister Daniela Amodei were packing their belongings into cardboard boxes.
Those who left OpenAI with them included core researchers who later participated in GPT-4 development and key members of OpenAI's safety team.
The reason for leaving OpenAI was not complicated: CEO Sam Altman was pushing OpenAI to transform from a nonprofit research institution into a company pursuing high valuation and commercialization. Dario believed that this rapid sprint was causing safety to spin out of control.
They named the new company Anthropic (meaning human-centered) and formulated a charter, hoping to build a laboratory emphasizing restraint and safety amid Silicon Valley's AI frenzy.
English text is being prepared.
It was precisely this determination to be independent of Sam Altman that pushed it onto a capital game from which there was no turning back.
1. Seed funding and the chips of being chosen
To train frontier models capable of competing with OpenAI, compute is an unavoidable physical threshold, and compute requires enormous funding to support.
To maintain the independence of research, Dario deliberately avoided introducing traditional Silicon Valley venture capital in the early days.
But fate produced a dramatically interlocking turn here. In 2022, the provider of $580 million in early funding to Anthropic was Sam Bankman-Fried (SBF), founder of the cryptocurrency exchange FTX.
Yet just a few months later, FTX collapsed, and SBF was brought to court.
As part of assets that were nearly deemed to be liquidated, Anthropic's equity faced the risk of judicial auction, and the company was once on the edge of survival.
The market waits for no one, and AI competition iterates by the day or even the hour. This sudden funding rupture forced Dario to find, within a very short time, a rescuer large enough and able to provide stable compute.
Cloud giants that smelled the opportunity stepped in.
In the fall of 2023, Amazon announced an investment of up to $4 billion in Anthropic, and Google followed with $2 billion. The funds solved the immediate crisis, but also fixed the real price of this rescue.
The giants' investment did not provide unconditional cash; rather, it required Anthropic to use the vast majority of the funds, untouched, to purchase their cloud computing resources and chips.
To implement this agreement, AWS even built a dedicated large Trainium2 chip cluster for it. Anthropic also agreed to give Amazon early priority tuning rights over its architecture improvements, deeply binding chip design and algorithm optimization together.
To escape the fate of its equity being auctioned, Anthropic formally bound its future to the infrastructure bills of the cloud giants.
2. Rejecting the throne
Just as Anthropic was struggling forward under the giants' infrastructure framework, a turning point that seemed capable of changing the landscape suddenly appeared.
On the evening of November 17, 2023, the OpenAI board abruptly fired CEO Sam Altman. In the ensuing 48 hours, the OpenAI board, plunged into chaos, secretly called Dario Amodei and proposed an unexpected plan: invite Dario to take over as OpenAI CEO and merge the two companies.
This was the closest Dario came to regaining control of his old team and integrating resources.
But he chose to refuse.
He saw clearly the conflicts of interest behind that governance crisis, and ultimately returned to Anthropic's lab to continue advancing Claude's development.
This refusal also led Sam Altman, after returning to OpenAI, to accelerate a computing power expansion plan on the scale of hundreds of billions of dollars.
Sam Altman, skilled at the game of capital, kept squeezing Anthropic's market space with offensive moves. To avoid being left behind, Anthropic had to multiply the scale of its training clusters in the development of the Claude 3 and 3.5 series.
By the time of the Claude 3.5 Sonnet iteration, the computing cost consumed by a single complete training and alignment run alone had already reached several times that of the previous generation's flagship model.
To avoid falling a generation behind OpenAI in model iteration, Dario had no way out and could only follow by pushing Claude's training scale and computing consumption to ever higher levels.
And to support this unprecedented computing expenditure, Anthropic needed to find a business model that could quickly monetize.
3. A business model forced into being
Under the financial pressure of computing costs doubling month by month, Anthropic could not, like OpenAI, slowly cultivate consumer payment habits through a general-purpose chat box aimed at the masses; it went directly to customers willing to pay heavily for efficiency.
So Anthropic made a key strategic pivot, abandoning the fight with ChatGPT for consumer traffic and turning to the enterprise battlefield.
In 2025, Anthropic launched Claude Code, a programming tool that runs directly embedded in the terminal command line.
This breakthrough came from the team's reconstruction of the underlying Prompt and computational graph architecture.
In the past, developers had to manually package and paste the code files of an entire project into a web box, whereas Claude Code can directly mount the local file system, precisely invoke indexes in the command-line background, and automatically complete cross-file refactoring at the tens-of-thousands-of-lines level without leaking core enterprise assets.
It did not spend resources on generic AI conversation; instead, it let the model enter the software development process directly, handling code refactoring, bug hunting, and architecture debugging for enterprises. Engineers no longer treat Claude as simple Q&A software, but as engineering productivity whose input-output ratio can be quantified.
This commercial pivot, forced into being by computing power, unexpectedly hit the pain points of the enterprise market.
Revenue paid by actual computing usage (API) then exploded; in 2025 Anthropic's revenue reached $4.585 billion, and by the first half of 2026 this figure directly surged to $16.286 billion, with enterprise customers spending more than $100,000 annually surging to more than 11,900.
However, the commercial explosion did not resolve Dario's predicament; instead, it pushed him into an even more urgent situation.
Details in the prospectus reveal the coldness of this business model. In the third quarter of 2026, more than 80% of revenue came from new model versions released less than six months earlier.
This means that rapid iteration of model capabilities will cause the commercial value of old versions to decay within a short period.
The more enterprise customers rely on Claude to handle complex tasks, the more Anthropic needs to continuously invest funds to purchase higher-performance computing power to train new models.
The product developed to earn money to pay computing fees ultimately requires the company to invest even greater computing power.
4. Hidden clauses in the prospectus
On October 26, Anthropic formally submitted a prospectus to the U.S. Securities and Exchange Commission (SEC), planning to list on Nasdaq under the ticker symbol ANT.
The document disclosed the interlocking operating results of recent years.
In the first half of 2026, the book loss was as high as $80.748 billion. Of this, about $72.4 billion came from fair value changes in convertible notes caused by valuation increases. Under Non-GAAP measures, its operating profit in the first half had already reached $1.618 billion, and operating cash flow also turned positive to $3.929 billion.
But what truly puts an end to this five-year story is the hundreds of pages of infrastructure procurement contracts in the prospectus notes.
To fulfill its previous computing commitments to the giants and maintain model iteration, Anthropic signed long-term arrangements totaling about $637 billion, including:
About $161.2 billion in equipment leasing with Broadcom;
At least $114.5 billion in cloud service commitments with Google;
At least $110 billion in infrastructure contracts with Amazon.
If other computing contracts with Microsoft, Lambda, and others are added, the overall long-term commitment scale is even higher.
Behind Broadcom's hardware cooperation worth as much as $161.2 billion is customized binding for custom ASIC chips and next-generation optical interconnect buses. This means that Anthropic's future computing foundation is already tightly bound to Broadcom's supply chain at the physical hardware level.
The data also shows that in 2025, 47% of Anthropic's sales (about $2.16 billion) were completed through Amazon AWS and Google Cloud marketplaces, for which it paid $351 million in distribution channel fees.
Simply put, for every $100 of services sold through cloud platforms, about $16 falls into the pockets of the cloud giants as a toll.
Amazon and Google are both shareholders of Anthropic and its largest computing power suppliers and distribution channels, while also developing their own Gemini and hardware chips.
V. Silicon Valley's Fate
Anthropic's IPO is not simply a company going public; it is also a rare collective bet in Silicon Valley in more than two decades.
Since Google's listing in 2004 and Meta's (Facebook's) bell-ringing in 2012, Silicon Valley has not witnessed such a massive listing event with highly integrated equity and infrastructure for a long time.
But this is no longer the Silicon Valley story people once knew.
For the past half century, the myth Silicon Valley prided itself on was the combination of garage startups and venture capital.
Two young people write code in a garage, get a few million dollars in venture capital, and quickly disrupt huge industry giants through product experience.
That was the era of Fairchild Semiconductor, the era of Apple and Microsoft, and also the era of early Google and Facebook.
And Anthropic's listing declares the end of this go-it-alone model.
In the large-model era, the physical barriers built by computing power, power grids, and data have pushed the threshold for innovation to the hundred-billion-dollar level.
No startup can complete this marathon independently without the cloud infrastructure of the giants.
The essence of this listing is that Silicon Valley giants, using capital, chips, and data centers, have brought a rebel that originally sought to challenge the old system back into the existing Silicon Valley industrial order.
Amazon, Google, Microsoft, and Broadcom are no longer just financial investors sitting in the audience; they are funders, lessors, channel providers, and also the biggest winners of this technological revolution.
Even top university laboratories and research institutions, once independent and neutral forces, have had to attach their research projects to the giants' computing clouds because they cannot afford computing clusters made up of tens of thousands of GPUs. This concentration of resources in a few giants is reshaping Silicon Valley's entire innovation model.
Dario Amodei, the idealist who once walked out of an old factory building in San Francisco, spent five years completing a long circuit.
He escaped Sam Altman's battlefield, but ultimately submitted his answer sheet to the new order of all Silicon Valley.
[Beyond the Page]'s note:
Every generation of technological revolution changes the threshold for innovation.
As the era when code alone could launch a company becomes harder, how much computing power one has and how much capital one can mobilize have begun to determine who is qualified to compete.
Technology is still moving forward, but the threshold for innovation is rising.
Silicon Valley's next reshuffle may start here. AI industry signal channel: Frontier Technology. This content comes from the internet and the original link; the views represent only the author and do not represent Huxiu's position. If there is any copyright issue, please contact hezuo@huxiu.com, and we will verify and handle it promptly. Everyone who is changing and wants to change the world is in the Huxiu APP[ ]