- Anthropic's IPO is multiple times oversubscribed, according to Bloomberg, which reports it seeks up to 90.4 billion dollars through the offering and a private placement.
- Anthropic says Google and Microsoft are strategic investors in its IPO as it launches Claude workplace software.
- TSMC's board approved up to 1 billion dollars in investments in Anthropic and OpenAI, according to The Wall Street Journal.
- Anthropic has committed to about 10 billion dollars more in Microsoft and Google cloud spending through 2033, according to a filing.
- Alibaba's Qwen team announced Qwen4, with Qwen4-Max in preview.
Top stories
Anthropic IPO multiple times oversubscribed as company seeks up to $90.4 billion
Bloomberg reports that Anthropic's IPO is multiple times oversubscribed, with demand for at least twice the shares available. Anthropic seeks up to 90.4 billion dollars through the IPO and a private placement, valuing it near 1.4 trillion dollars, with 26 billion dollars already committed privately. Anchor-sized orders of 5 to 10 billion dollars come from long-only investors. The deal could exceed SpaceX's proceeds but falls short on valuation.
Bloomberg Tech; also Bloomberg Technology
Alibaba's Qwen announces Qwen4, with Qwen4-Max in preview on Token Plan
Qwen, Alibaba's model team, announces Qwen4, describing deeper knowledge, richer world understanding, and stronger coding and agentic capabilities for real-world tasks. Qwen4-Max is in preview on Alibaba's Token Plan, with early access also available on QwenWork, Qoder, Qwen APP and QwenStudio.
Qwen (@Alibaba_Qwen); also AI News / TestingCatalog
Anthropic Launches Workplace Software With Claude; Microsoft Shares Fall More Than 3.5%
Anthropic launched workplace applications on Tuesday that combine its Claude AI with tools to create and edit spreadsheets and presentations and to take notes in video calls. Microsoft shares fell more than 3.5 percent before paring losses, while Google shares were down less than 1 percent. The tools include Claude Sheets, Claude Whiteboards and a preview of Claude Meetings. Anthropic said Google and Microsoft are strategic investors in its IPO.
The Information (newsletters)
TSMC approves up to $1 billion investment in Anthropic and OpenAI
TSMC's board has approved a plan to invest up to 1 billion dollars in Anthropic and OpenAI, according to The Wall Street Journal. The report says the chipmaker is buying stakes in the two model developers, a move that shows its wish to broaden its AI exposure beyond manufacturing advanced chips. Reuters reported the split as 500 million dollars in OpenAI preferred stock and 500 million dollars in Anthropic common shares at its expected IPO price.
The Wall Street Journal (Technology); also Ars Technica
Anthropic Adds About $10 Billion in Microsoft and Google Cloud Commitments; Lambda Deal May Be Cancelled
Anthropic said in a Tuesday filing that it has committed to roughly 10 billion dollars more in Microsoft and Google cloud spending through 2033, bringing its totals to 49.2 billion dollars on Azure and 121.6 billion dollars on Google Cloud. It said a 34.9 billion dollar Lambda lease could be cancelled because Lambda missed an agreed data center deadline, though it does not currently intend to cancel.
The Information (newsletters)
Companies
Fractile raises $720M Series C for AI chips
UK chip startup Fractile raised a $720M Series C backed by the Sovereign AI Fund, led by Thrive Capital, Founders Fund, Lightspeed and Redpoint; valuation over £5B.
TechCrunch
Alibaba plans Hong Kong listing of chip design arm T-Head
Alibaba plans to list chip design arm T-Head in Hong Kong, per SCMP.
South China Morning Post (Technology)
Mercor raises $750m Series D at $20B valuation as annualized revenue reaches $3B
Mercor raised a $750M Series D at a $20B valuation, Latent Space reports.
Latent.Space
Torus completes $500 million Series C led by Magnetar and launches Foundry modular AI data center
Torus completed a $500M Series C led by Magnetar.
36Kr
Compute
CoreWeave plans at least $4 billion investment-grade loan after raising sales forecast
CoreWeave plans to raise at least $4B via an investment-grade loan offering, Bloomberg reports, after raising its annual sales forecast that eased debt concerns.
Bloomberg Technology
Models
Google releases Gemini Decision 1 decision model in Gemini API at $0.10 per 1M input tokens
Google released Gemini Decision 1 in the Gemini API, per Logan Kilpatrick.
Logan Kilpatrick (@OfficialLoganK)
Runway releases Praxis-1 world action model with open weights
Runway released Praxis-1, a 27B mixture-of-experts diffusion transformer predicting video and actions for robot control, open weights under Apache 2.0.
Runway (@runwayml)
Policy
Trump authorized China chip crackdown three times; Lutnick stalled it, sources say
Axios reports Trump approved action against suspected Nvidia chip diversions to China, but Lutnick and Kessler resisted.
Axios (AI) · according to six anonymous high-level sources, reported by Semafor
Microsoft and trade groups urge Trump administration to halt planned ban on Chinese AI data center component imports
Reuters: Microsoft and others pressed the Trump administration to halt plans to bar Chinese imports of a key AI data center component.
Reuters (Technology)
Products
ElevenLabs previews fine-tuning of open-source and voice models at NYC Summit
At its New York Summit, ElevenLabs said it will help businesses fine-tune open-source language and voice models.
ElevenLabs
United Airlines bans AI agents and assistants from booking its flights
United Airlines will bar AI agents and assistants from booking flights via its app, website, travel agencies and aggregators, following similar bans by banks and retailers over privacy.
The Information
Safety
Microsoft Research finds backdoor can be planted by reprogramming a few dormant neurons in an AI agent
Microsoft Research: reprogramming a few dormant neurons can backdoor an AI agent, which behaves normally until a trigger.
Microsoft Research (@MSFTResearch)
Anthropic tells investors an AI safety alarm may be imminent days before its IPO
In its pre-IPO filing, Anthropic warns investors that AI accelerating its own research could soon trigger one of its key safety alarms, possibly imminently, Yahoo Finance reports.
Hacker News; also The Information (newsletters)
Sent by the Observatory.