Today's October 27 AI briefing rounds up the five developments from the past 24 hours most worth reading: Anthropic has formally filed its listing documents, OpenAI has announced three scientific results that still need verification, and Meta and Microsoft are each freeing up money for the AI compute bill in different ways.
Anthropic files its S-1: $16.3 billion of first-half revenue, with huge compute commitments ahead
On October 26, Anthropic filed a Form S-1 registration statement with the U.S. Securities and Exchange Commission, planning to list on Nasdaq under the ticker ANT. The number of shares and the price range have not yet been set.
The filing shows revenue of $16.3 billion for the first half of 2026, about 14 times the same period a year earlier. Second-quarter revenue was $11.6 billion, up 1,368% year on year, with operating income of $974 million for the quarter. The first-half net loss was $80.7 billion, of which $72.4 billion was a non-cash fair-value charge on convertible notes and SAFEs; the operating loss was $693 million.
The compute bill draws more attention: seven named agreements add up to roughly $637 billion, including up to $120.5 billion with SpaceX's AI unit, at least $114.5 billion with Google, $110 billion with Amazon, and $44.9 billion with Microsoft. Once the SpaceX agreement is fully delivered, Anthropic would pay about $3.1 billion a month for NVIDIA GPU compute. The seven co-founders control 50.1% of the votes through a single Class F share. Market reports say the listing could value the company at more than $2 trillion, but the filing itself sets no price, and the real test will come in the roadshow.
OpenAI announces three scientific results: do not treat them as settled yet
On October 26, OpenAI announced one result each in mathematics, computer science, and physics from its internal models: a proof of the local smoothing conjecture in every dimension, which would also settle the Kakeya conjecture and Stein's restriction conjecture; a disproof of the Exponential Time Hypothesis; and a rigorous existence proof for quantum chromodynamics and Yang-Mills theory.
The mathematics and computer science results have been formalized in Lean and their proof materials made public, while the physics materials are being held back for now, with a group led by Nobel laureate in physics David Gross writing up an explanation first. OpenAI describes these manuscripts as raw research material for the academic community to digest. None of the three results has been independently verified, so "announced" and "confirmed" need to be kept apart.
Meta plans a $95 billion raise: $35 billion first, $60 billion next year at the earliest
In a filing with the SEC on October 26, Meta set out a financing plan of up to $95 billion: about $35 billion now, comprising a $22.5 billion forward sale of Class A common stock and $12.5 billion of mandatory convertible preferred stock; then an at-the-market programme of about $60 billion, selling a new non-voting Class C share, starting no earlier than the first quarter of 2027, which has not yet been established and may not happen.
NVIDIA, AMD, and Broadcom have given non-binding indications of interest of up to $5 billion in total. Meta has guided 2026 capital expenditure of $135 billion to $145 billion, and this money is meant first for the data-center and compute bill.
Microsoft offers a voluntary exit to about 60,000 employees
In an internal memo on October 26, Microsoft Chief People Officer Amy Coleman introduced a time-limited "Voluntary Exit Offer" for employees at level 70 and below in the United States, Canada, the United Kingdom, and India. Bloomberg reported, and Microsoft confirmed to TechCrunch, that about 60,000 people are eligible, roughly 27% of the company's global workforce.
Participating teams include AI experiences, Copilot, Xbox, and Windows + Devices, while marketing, commercial business, and security teams are among those not taking part. Microsoft has not published the specific compensation terms, saying only that they resemble a standard severance arrangement, and it has set no target for departures. This is not a layoff; the decision rests with each employee. But it comes as Microsoft has forecast $175 billion of capital expenditure for calendar 2026, and every dollar of payroll saved points at the same AI bill.
GPT-6.1 Terra launches: half the price, built for computer use
On October 26, OpenAI released GPT-6.1 Terra, a fast model combining visual reasoning with computer use. Its API pricing is $1 per million input tokens and $5 per million output tokens, with cached input at $0.05, 50% lower than GPT-5.6 Terra and GPT-6.1 Sol.
In OpenAI's own tests, Terra solved 50.3% of Terminal-Bench 4.0 tasks at maximum reasoning effort, against 21.5% for GPT-5.6 Terra, and scored 15 percentage points higher than its predecessor on Agents' Last Exam at about one-third of the cost. The model is available to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and developers can call it through the API as gpt-6.1-terra.
