Executive search, now with Pay While They Stay: 25% of 'base salary' only, and the payments stop if your new leader leaves.See how it works
Hurren & Hope 020 3833 0334 Talk to a hiring expert
San Francisco edition

San Francisco tech: the week of 22 September 2026

The week in San Francisco technology, read and picked by hand. No algorithm, nothing in here because someone paid for it.

OpenAI seeks new funding round at $1.5 trillion valuation

According to The New York Times, OpenAI is seeking a new round at a $1.5 trillion valuation, roughly double its most recent $730 billion mark.

Reported by StrictlyVC, 17 September

Arcee AI raises $150 million Series B at $1 billion valuation

Photograph: Fortune

Arcee AI has raised at least $150 million in Series B funding at a $1 billion pre-money valuation. Vista Equity Partners, Cambium Capital and Emergence Capital led, with Microsoft's M12, AI10 Ventures, Hitachi, IAG, P7 and Wipro also taking part.

Reported by Fortune, 16 September

Angle Health raises $600M at a $2.7B valuation

Photograph: TechCrunch

Angle Health raised a $200 million Series C alongside a $400 million tender offer, at a $2.7 billion valuation. Vitruvian Partners led.

Founded in 2019, Angle provides level-funded health plans for small businesses and says its AI-powered platform serves more than 5,000 businesses and is profitable.

Reported by TechCrunch, 18 September

TypeSafe AI releases Jev, a non-LLM model for calibrated decisions

Photograph: TechCrunch

Diogo Almeida, a former OpenAI researcher who helped build ChatGPT, has released Jev, a transformer-based model that outputs probabilities rather than text, so it cannot hallucinate.

Vercel reported Jev was five to 18 times quicker and more accurate than a leading LLM for safety classification. Expect it to sit alongside LLMs, monitoring agent behaviour and routing work between models.

Reported by TechCrunch, 18 September

Anthropic opens a biology wet lab in the Bay Area

The lab will test what AI can contribute to biology and drug discovery, with Claude directing and carrying out research alongside people.

Reported by MTS, 19 September

Palantir, Nvidia and Booz Allen curb use of Anthropic and OpenAI models

The three are limiting use over concerns about proprietary data. Palantir wants zero-data-retention guarantees, and Nvidia is steering sensitive work to its own Nemotron models.

Reported by StrictlyVC, 15 September

Editor's note

The messaging hype around AI has changed but the fundamentals remain: compute power alone will not add the intelligence in AI. It doesn't change the power of the tech and the impact it can have, but it does question the viability of the big players in the field over the long term.

The weekly edition

Get the week in tech, by city

Free, every Tuesday. Picked by hand by Mark Hurren, with no sponsors.

We email you a link to confirm. One click to unsubscribe. Privacy

Book a call

Are you hiring, or looking for a role?

So we can point you in the right direction.

Grow the NetworkRefer a friend, you both win