USV raises $900M to lead more AI rounds, cuts its partnership to four

Share
USV raises $900M to lead more AI rounds, cuts its partnership to four

A storied early-stage shop writes its biggest check yet and shrinks the room that spends it; a state regulator calls human-vs-robot fighting unsanctioned; CoreWeave turns the agent improvement loop into a product.

Union Square Ventures has raised $900 million across two funds — including a $500 million early-stage fund, nearly double its 2024 raise — while cutting its general partnership to four investors, and the firm says it intends to lead more AI rounds. Bloomberg reported the partnership will announce the money Thursday: the early-stage fund jumps from the $275 million USV raised in 2024, and the active general partnership shrinks to four. The firm has spent its career arguing the opposite — small funds, a small partnership, conviction over committee — so doubling the check size and halving the decision-makers is a genuine strategy change, not a routine close. The read: AI rounds now demand more capital per deal, and USV would rather write bigger checks with fewer voices than watch its best companies get led by megafunds. It has already been leaning into the theme — the firm took a position in Nous Research's $90 million Series B earlier this week — but a $500 million fund exists to lead, not follow.


California's State Athletic Commission has told REK, the self-styled "humanoid robot fighting league," that it cannot hold any contest or exhibition involving a human in the state without sanction. The cease-and-desist targets the September 18 bout in San Francisco between influencer Frankie LaPenna and a $100,000 humanoid — an EngineAI machine with a Terminator-style head, piloted by a person over a virtual-reality link — which the commission called unsanctioned; the New York Times broke the story, and REK's chief executive Cix Liv posted the letter to X himself with "We made it boys." The letter's wording bars fights involving "any human," which leaves the league's next event untouched: Friday's card features two armed robots fighting each other. Either way, this is the first US regulator treating robot combat as a combat sport requiring a license rather than a tech demo — a category question every humanoid stunt from here on will have to answer.


CoreWeave is packaging the agent improvement loop as a product: Forge connects run, observe, curate, improve, evaluate in one environment, open across models, frameworks and other clouds. Unveiled around its Fully Connected event, Forge bundles Agent Lens for inspecting what agents actually did, a registry for checkpoints and agent configurations, reinforcement-learning rollouts and model distillation, with CoreWeave's ARIA research agent now generally available; the company says MasterClass and Canva are already building on it, alongside a partner network of tested, co-engineered integrations. VP of product marketing Susanne Seitinger's framing is the pitch — "go from first agent to your very best agent as quickly as possible" — and the underlying claim is that the winning cloud won't just train and serve models but compress the loop between a production run and the next better version of it.

What to watch: whether REK's robot-vs-robot event proceeds without a human-participation ruling becoming a precedent, and which AI round USV leads first with the new fund.

Is shrinking a partnership to four investors the sharpest way to bet on AI — or the riskiest? Tell us in the comments.

Read more

Today in AI — October 8, 2026

Today in AI — October 8, 2026

A day of pushback and product moves: mathematicians told OpenAI to stay out of their field, Anthropic rewrote the rules for treating Claude, and the leaderboard business showed it can pay. Models & Research * Mathematicians call for an OpenAI boycott. The Association for Human Mathematics published a statement urging researchers to stop working with the company after OpenAI dumped hundreds of AI-generated proofs, calling the release "not a demonstration of scholarship, but a demonstration

OpenAI's revenue is $20B below prior reports, AI stocks sink

OpenAI's revenue is $20B below prior reports, AI stocks sink

A revenue clarification rattled the AI trade on Thursday, a leaderboard startup nearly doubled its valuation, and Anthropic picked a side in critical-infrastructure defense. Here's the afternoon in brief. OpenAI told investors its annualized revenue was approaching $50 billion at the end of September — far below the roughly $70 billion that had circulated from investor documents, and the market moved accordingly. The Financial Times broke the story Thursday, and the afternoon session saw Nvidi

Cantwell's frontier AI framework makes safety rules mandatory

Cantwell's frontier AI framework makes safety rules mandatory

Washington's AI policy split cleanly this week: the labs signed a voluntary pledge, and now a Senate committee leader is writing the same ideas into law — with penalties attached. Sen. Maria Cantwell unveiled a six-principle framework for regulating frontier AI on Wednesday — the most detailed Democratic counter-proposal to the White House's voluntary accord so far. The plan asks Congress to set enforceable federal safety standards, with NIST charged with defining protection against catastroph

Google's universal Gemini agent arrives in private preview

Google's universal Gemini agent arrives in private preview

The enterprise agent race found its biggest entrant today, while a security research team showed just how badly shared cloud agent infrastructure can fail. Google Cloud announced a universal Gemini agent that works across your apps, devices, and even other companies' tools. Unveiled at the Gemini at Work 2026 event on Thursday, the agent lives inside the Gemini Enterprise app and runs in the cloud, so it keeps the same context whether you talk to it from your phone, desktop, the web, or a thir