Anthropic filed its S-1 with the US Securities and Exchange Commission on Monday. The paperwork shows the company expects to raise more capital in this public offering than any firm in history. It is the first major American business focused solely on artificial intelligence software to go public. Valuation has climbed from $183 billion a year ago to $965 billion in the latest round. Analysts expect the stock to trade near $2 trillion.
The filing reveals staggering figures. Revenue hit $16.3 billion in the first half of 2026. That is more than 14 times the figure for the same period last year. The company has already committed hundreds of billions of dollars to compute infrastructure.
Anthropic is expected to raise more money in its IPO than any company in history.
Financials and Enterprise Focus
Leah Feiger noted the sheer scale of capital involved. She recalled a previous WIRED article about people selling homes in the Bay Area with promises of Anthropic stock. The financials released now suggest that promise was not empty.
Brian Barrett said the numbers put many concerns to rest. He described the entity as a real business that is growing and doing the things a business should do. He acknowledged the company is losing money hand over fist but argued the trajectory is better than expected.
Zoë Schiffer highlighted a key difference between Anthropic and OpenAI. She said the majority of revenue comes from other businesses using their coding product. She noted that Dario Amodei, the CEO, focuses daily on enterprise code. That is the thing that makes money for the business. She contrasted this with the perception that OpenAI is scattered across too many areas.
Revenue in the first half of 2026 reached $16.3 billion.
Reliance on Clients
Brian Barrett raised a yellow flag regarding revenue concentration. Nearly half of the $60 billion in revenue came from three tech companies cloud marketplaces. He argued this creates a tangled nest of reliance. The company depends heavily on a small number of clients that are also competitors. He questioned where the moat is in a market with open source products and other AI tools getting better.
Zoë Schiffer agreed the flag was yellow. She said it was not surprising for a polyamorous worker cooperative to be incestuous in its business dealings. She noted that 80 percent of revenue for both Anthropic and OpenAI comes from the top 1 percent of AI spenders. If those companies cut back, Anthropic is exposed.
However, she added a green flag regarding transparency. The filing included warnings about existential threats. She said this was probably her favorite part of the whole thing.
Existential Risk
The filing uses passive voice to state concerns have been raised by Dario about existential risk. The phrase appears three times in internal sections. Twice it refers to existential risk to humanity. Once it refers to existential risk to society. Brian Barrett interpreted this as a scenario where humans abandon any semblance of a social contract.
Leah Feiger called the stuff incredible. She noted that this is not new for Anthropic. They have shouted about AI killing everybody since they started.
Brian Barrett identified a red flag for humanity. He said the company cites recursive self-improvement as its primary competitive advantage and the most likely cause for the end of human life on earth. He explained that recursive self-improvement means an AI model improves itself without human intervention. He argued that if models keep getting better at improving themselves, they could close the loop until humans do not understand what the AI is doing. That is the scenario where AI is powerful enough to end the world. It is also the scenario where Anthropic plans to get ahead.
Leah Feiger mentioned that former researcher Jacob Coxon resigned because Anthropic could bring about the end of the world. He specifically cited curbing recursive self-improvement at the top of his list.
Brian Barrett noted that the risk factors section is the longest part of the prospectus. It is 94 pages. He said they could have saved everybody by just admitting they have a shadow side.
Compute Costs
Leah Feiger asked for an explanation of the compute bill. She said all the technology costs so much money. She does not personally feel like she has an understanding of where the limit is. She asked how to spend enough to use all the money but also spend enough to not put the business out of existence.
Zoë Schiffer said she does not pretend to understand it. She said the scale is totally unprecedented. She compared it to railroad companies but noted a key difference. Unlike the railroad companies, which were operating on slightly thinner margins, this scale is different.
What it means
The filing confirms Anthropic is a massive business with huge losses but also huge growth. The company relies on a few big clients for most of its money. It is also betting that the ability to improve its own models is the key to winning. That same ability is the reason the company warns the world could end.
