AI Wire Daily, Oct. 26: Anthropic files for IPO, Meta seeks $95B, Microsoft offers buyouts

Anthropic opens its books for a stock-market listing, banks line up $63 billion so it can rent Broadcom chips, Meta asks investors for up to $95 billion, Microsoft offers buyouts to tens of thousands of staff, OpenAI cuts the price of an AI that uses your computer, Siri finally speaks Portuguese, and OpenAI publishes data on how people use ChatGPT for work.

This is the AI Wire roundup for Monday, October 26, 2026, with stories that were not in the weekend editions. Each item links to the company, regulator or government text it is based on, or to the news report that first published it. Company benchmark results are the companies’ own claims until others test them.

1. Anthropic files to go public, and shows its numbers

Anthropic filed a registration statement on Form S-1 with the U.S. Securities and Exchange Commission on Monday, the paperwork a company needs before it can sell shares to the public. It plans to list its Class A shares on the Nasdaq Global Select Market under the ticker “ANT.” The number of shares and the price range are still blank. Goldman Sachs and Morgan Stanley are named in the announcement.

The filing gives the first official look at the business. Revenue was $16.29 billion in the first six months of 2026, against $1.16 billion in the same period of 2025. For all of 2025 it was $4.59 billion. More than 11,900 customers paid Anthropic $100,000 or more over the 12 months to September 30, up from more than 950 a year earlier.

The loss is the headline number: $80.7 billion for the half year. Most of it is an accounting charge. The filing records a $72.4 billion loss from the change in fair value of convertible notes, derivatives and SAFEs, instruments whose value on paper rises as the company becomes more valuable. The operating loss was $693 million, down from $2.1 billion a year earlier. Anthropic held $131.7 billion in cash and short-term investments at the end of June.

Control stays tight. A single Class F share held by a company owned by the seven co-founders, including CEO Dario Amodei and President Daniela Amodei, will carry 50.1% of the votes on most matters. The Long-Term Benefit Trust, through Class T shares, elects a majority of the board. The risk section says the company “may not achieve or sustain profitability” and warns that advanced AI could “pose catastrophic or existential risks to humanity” if not developed with proper safeguards.

2. Banks and Blackstone line up $63 billion for Anthropic’s Broadcom chips

The same day, Bank of America, Blackstone, Citi and Morgan Stanley announced a $63 billion financing for AI computing capacity built on Broadcom technology for Anthropic. A new special purpose vehicle will buy Broadcom XPU AI accelerators and networking equipment and lease them to Anthropic. Deliveries are expected to start in 2027.

The three banks arrange the senior, first-lien debt, and Blackstone is the sole provider of the committed second-lien portion. Broadcom backs the vehicle’s residual value up to the amount of the first-lien debt, which means the chipmaker carries part of the risk if the hardware loses value. Broadcom CEO Hock Tan said the deal “validates that there is capital available to finance such growth.”

3. Meta plans to raise up to $95 billion from shareholders

Meta announced a financing plan of up to $95 billion to pay for AI infrastructure and what it calls “personal superintelligence.” It starts with about $35 billion in underwritten offerings: about $22.5 billion of Class A stock sold through forward agreements and about $12.5 billion in mandatory convertible preferred shares. Later, Meta plans a roughly $60 billion “at-the-market” program, selling new nonvoting Class C shares over time, not expected to begin before the first quarter of 2027. Meta says it has not signed any agreement for that program yet.

Chipmakers Nvidia, AMD and Broadcom indicated interest in buying up to $5 billion of the Class A shares. The indications are not binding, and they may buy more, less or none. Meta expects $135 billion to $145 billion in capital spending in 2026, said it is building four “superclusters” of more than one gigawatt each, and reported more than 10 million Muse users in the U.S. and Canada in the early weeks of its AI agent. CNBC reported the stock fell 1.6% in morning trading.

4. Microsoft offers buyouts to about 60,000 employees

Microsoft introduced a “Voluntary Exit Offer” for eligible employees in participating teams in the United States, Canada, India and the United Kingdom, at its internal level 70 and below. Chief People Officer Amy Coleman’s memo says the offer is time-limited and comes with financial support and career-planning help, and that some organizations are offering it while others are not.

Microsoft did not give a number. CNBC, citing a person familiar with the matter, reported that about 60,000 people are eligible, about 27% of a workforce of 223,000, that employees must decide in early December, and that those who accept leave on December 31. CNBC ties the move to the cost of Microsoft’s data-center build-out for AI customers such as OpenAI and Anthropic. Microsoft reports quarterly results on Wednesday.

5. OpenAI launches GPT-6.1 Terra, a cheaper model that uses your computer

OpenAI released GPT-6.1 Terra, a mid-range model built to read screens, click through apps, fill in forms and write. Standard API pricing is $1 per million input tokens and $5 per million output tokens, which OpenAI says is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol. Tokens are the small chunks of text that developers pay for.

OpenAI’s own numbers: on its Agents’ Last Exam benchmark, Terra matches GPT-6 Astra at one-tenth the price per token and finishes tasks more than three times faster than Astra and GPT-6.1 Sol. On an internal set of error-inducing prompts, its error rate at low reasoning effort fell from 53.6% to 16.7%. On Terminal-Bench 4.0, a coding test, it solves 50.3% of tasks at maximum effort. OpenAI says the model meets the “Critical” cybersecurity threshold of its Preparedness Framework and keeps the safeguards built for GPT-6 Astra. It is available now to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and in the API as gpt-6.1-terra.

6. Siri AI now speaks Portuguese, Spanish, French, Japanese and Korean

Apple released its 27.2 software updates on Monday, and with them Siri AI, the rebuilt version of its assistant, now works in French, Japanese, Korean, Portuguese and Spanish, in addition to English. It is still labeled beta and needs a device that supports Apple Intelligence. Apple says Chinese (traditional), Norwegian, Turkish and Vietnamese will follow by the end of the year.

There are limits. Siri AI remains unavailable in the European Union on iPhone, iPad and Apple Watch while Apple talks with EU regulators, and it is not available to users under 13. Features that rely on Apple’s server-side models, including Siri AI, have daily usage limits, and Apple says “expanded access to such features will be available for a fee in the future.” For Brazilian iPhone users, this is the first time the new Siri is available in their language.

7. OpenAI publishes data on how people use ChatGPT for work

OpenAI’s economic research team released OpenAI Signals v3.0, a dataset that maps work-related ChatGPT messages to 165 standard work activities from O*NET, the U.S. occupational database, across nearly 100 countries, quarter by quarter from early 2025 to the third quarter of 2026. It covers individual ChatGPT plans only, not enterprise accounts or Codex.

The main finding: work use is spreading out. Across 74 countries observed in every quarter, the ten most common activities made up an average of 51.1% of classified work messages in the third quarter of 2026, down from 56.0% in early 2025, with different countries specializing in different kinds of work. OpenAI says researchers do not read individual conversations; messages are classified automatically after identifying information is removed, and statistical noise is added.

Why it matters

Today was about money and who pays for AI. Anthropic’s filing shows first-half revenue up about 14 times year on year, and the $63 billion Broadcom deal shows how some of its chips will be paid for: with debt, through a vehicle that leases hardware instead of buying it outright. Meta is asking shareholders for up to $95 billion, and Microsoft is offering buyouts while it spends on data centers. Even chipmakers are offering to put money back into a big customer.

On the product side, the price of capable agents keeps falling. GPT-6.1 Terra halves the cost of computer-use work, and Apple is bringing a far more capable Siri to speakers of five more languages, Portuguese among them. OpenAI’s work data suggests people are already spreading AI across more kinds of tasks.

What we still don’t know

Anthropic’s price range, how many shares it will sell and what valuation it will target. Whether Meta’s chip suppliers actually buy the shares they showed interest in. How many Microsoft employees accept the buyout, and whether layoffs follow. How GPT-6.1 Terra holds up in independent tests. When Siri AI reaches the European Union, and how much Apple will charge for expanded use.