Closing Time: Anthropic opens its books
Semafor reports that Anthropic’s IPO filing shows more than $400 billion in future commitments for chips and data center leases, along with a governance setup in which its seven co-founders control the company through a single supervoting share.
Good afternoon. Anthropic finally let the public see its finances today, and the market liked enough of what it saw to buy chip stocks for the rest of the session. Meta, Microsoft and SpaceX all made big moves of their own to pay for the AI build-out. Here is what mattered today.
Semafor reports that Anthropic’s IPO filing shows more than $400 billion in future commitments for chips and data center leases, along with a governance setup in which its seven co-founders control the company through a single supervoting share.
CNBC reports that Meta plans to raise up to $95 billion by selling new stock to fund its AI infrastructure, and that Nvidia, AMD and Broadcom were interested in buying up to $5 billion of the shares.
Microsoft is offering voluntary buyouts to about 60,000 employees as it trims costs and spends heavily on AI infrastructure, NBC News reports.
Stocks rose as the filing revived the AI trade, Yahoo Finance reports, with chipmakers leading and the Nasdaq closing up 0.8%.
Anthropic opens its books
For years, investors could only guess at how much money Anthropic makes and how much it spends. On Monday the company behind Claude made its IPO filing public, and the numbers show a business that is growing extremely fast while carrying some of the largest bills in the technology industry. According to Semafor, it plans to list on the Nasdaq under the ticker ANT.
The growth is the headline. Anthropic reported about $16.3 billion in revenue for the first six months of 2026, up from about $1.2 billion in the same period a year earlier. Its operating loss narrowed to roughly $700 million over those six months, which is small for a company spending so heavily to train and run its models.
The reported net loss looks alarming at $80.7 billion for the half year, but most of it is an accounting charge rather than cash going out the door. About $72.4 billion comes from the rising value of convertible notes and similar contracts, which grows as Anthropic’s own valuation climbs. Analysts who spoke to CNBC were cautiously positive, and DA Davidson’s Gil Luria pointed to the more than $130 billion in cash and securities the company held at the end of June.
The obligations are the part that may worry investors most. Semafor reports that Anthropic was committed to $417 billion in future compute and infrastructure leases as of June. Its five largest customers made up roughly a quarter of revenue, and nearly half of its revenue comes through cloud providers such as Amazon Web Services, Google Cloud and Microsoft Azure.
The governance is unusual even by the standards of founder-led tech companies. The seven co-founders will control the company through a single supervoting share, and a council of trustees with AI safety expertise will pick a majority of the board. Amazon and Google, two of Anthropic’s earliest backers, will hold nonvoting stock. Because Anthropic is a public benefit corporation, the filing also says plainly that it must weigh its mission alongside shareholder returns.
The risk section reads differently from most IPO filings. It names a possible clash with the U.S. government as a business risk, and it warns that advanced AI systems could pose catastrophic risks if they are not developed with the right safeguards. The share count and price range are still blank, and CNBC reports that a roadshow and listing are expected in mid-to-late November. The listing will be the first real test of how much public investors will pay for a frontier AI lab.
Elsewhere in AI today
CNBC reports that Meta will raise $35 billion through underwritten stock offerings and could sell another $60 billion over time through an at-the-market program. The money will support its AI infrastructure and what the company calls personal superintelligence, and the chipmakers that sell Meta its hardware may end up owning some of the new stock.
MarketWatch reports that SpaceX agreed to a roughly $40 billion financing deal with an Apollo-managed entity to lease GPU pods and other data center equipment. The lease is expected to close before the end of the year, with equipment arriving in the first half of 2027.
OpenAI released GPT-6.1 Terra, a faster and cheaper model built for computer use, such as clicking through apps and filling in forms, The Verge reports. It is available today in ChatGPT Work and Codex for paid users and in the API, at half the price of the previous Terra model.
TechCrunch reports that Apollo chief economist Torsten Slok sees no financial payoff from AI so far at large U.S. companies outside the tech giants. In his charts, revenue per worker and profit margins are rising for the biggest tech companies but are flat for the rest of the S&P 500.
Markets
Stocks rose on Monday as Anthropic’s filing gave the AI trade a lift, according to Yahoo Finance. The Nasdaq gained 0.8%, the S&P 500 added 0.7% and the Dow rose 0.3%. Chipmakers did much of the work, and the Philadelphia semiconductor index climbed 2.8%. Of the 86 AI and tech stocks we track, 61 closed higher.
Optical networking stocks led the winners, with Lumentum up 8.2% and Coherent up 5.9%. Chip designer Arm rose 6.2% and chip design software maker Synopsys added 5.6%, while cybersecurity company SentinelOne gained 6.3%.
On the other side, electronics manufacturer Celestica fell 5.3% and AI cloud provider Nebius lost 3.5%. Voice AI company SoundHound dropped 4.3%, trading app Robinhood slipped 2.4% and Adobe fell 2.0%.
Biggest winners
| Ticker | Name | Price | Day |
|---|---|---|---|
| LITE | Lumentum | 818.10 | +8.19% |
| S | SentinelOne | 25.50 | +6.27% |
| ARM | Arm | 223.35 | +6.19% |
| COHR | Coherent | 221.71 | +5.93% |
| SNPS | Synopsys | 454.33 | +5.62% |
Biggest losers
| Ticker | Name | Price | Day |
|---|---|---|---|
| CLS | Celestica | 246.25 | -5.29% |
| SOUN | SoundHound | 4.24 | -4.29% |
| NBIS | Nebius | 153.27 | -3.45% |
| HOOD | Robinhood | 92.84 | -2.35% |
| ADBE | Adobe | 258.62 | -1.97% |
Prices are Monday’s U.S. closing prices, October 26, 2026, versus Friday’s close. Market data is a snapshot, not live.
Watch
Fireship — China just dropped a new frontier model... and nobody can explain how
Fireship takes a look at Kimi K3.1, a free open-weight model from China with about three trillion parameters that comes within a point of Claude Opus on the Terminal Bench coding test.
The video asks the obvious next question, which is whether the model is actually good in practice or just good on benchmarks.
Watch this for a quick, sharp read on how close open models are getting to the paid frontier.
Latent Space — An Eval for AI Scientists That Knows the Answer Key
Leash Bio cofounder Ian Quigley returns to Latent Space to talk about Minos, a set of pre-measured lab experiments that AI science agents can explore.
Because Leash has already measured how thousands of proteins react to millions of molecules, it can check an agent’s guesses against real results instead of trusting the agent’s own story.
Watch this if you want to understand how researchers are starting to test AI scientists against hard data.
Y Combinator — The New Full-Stack Designer
Marvin Schwaibold, who leads the product design studio at Shopify, joins YC’s Aaron Epstein to walk through how his team designed the new Shopify admin and its Sidekick assistant.
He also shows a weekend project that rebuilt real grass as a shader and a carousel that took 91 attempts to get right, before the two review websites sent in by viewers.
Watch this for a look at how design work changes when the designer can also build.