← Daily AI news

Daily AI news, Sunday, 8 November 2026

Daily AI / Tech / VC Briefing: Sunday 8 November 2026

Anthropic logo and brand image
Image: Anthropic (anthropic.com)

Good afternoon, here's today's briefing. Anthropic is pressing on with its IPO at a lower price, OpenAI's finance chief says the build-out continues despite the midterms, Microsoft is quietly handing back rented AI capacity, and two notable people moves landed at AI companies.

LEAD: Anthropic pushes ahead with a pre-Thanksgiving IPO at a lower valuation

Background: All week, investors have been pushing back on Anthropic's hoped-for $1.5 trillion valuation, and Thursday's Nasdaq-100 drop of about 6.4% raised doubts about whether the roadshow could start on time. The prospectus still leaves the price range and share count blank.

What happened: The Information reported that Anthropic plans to go ahead with its investor roadshow in the week starting 9 November. Investors cited by the Financial Times say the company still wants to list before Thanksgiving and raise $80–100 billion at a $1.25–1.75 trillion valuation, below the $2 trillion some had hoped for but above its last private mark of about $965 billion. Its annualised revenue is now about $81 billion, up from roughly $65 billion at the end of July. Under SEC timing rules, the earliest it could price is Friday 13 November and the earliest first trade is Monday 16 November. Renaissance Capital's weekly note says the IPO "remains on track" but flags that third-quarter growth slowed to 47% quarter on quarter, from 144% in the second quarter.

Why it matters: Anthropic has decided that going public now at a lower price beats waiting for a better market. That would make it the first pure frontier AI lab to list, and how well the shares trade will set the mood for every AI valuation, OpenAI's included.

Sources: The Decoder on the FT report · The Information · Yahoo Finance timeline · Renaissance Capital

OpenAI's CFO: we keep building, and the business grew 70% last quarter

Background: Yesterday's briefing covered OpenAI closing $11 billion of a smaller round at $1.25 trillion, and its new pledges to communities hosting data centres after voters opposed data-centre measures in the midterms.

What happened: Chief financial officer Sarah Friar told CNBC on Friday that OpenAI will press ahead with its US data-centre expansion: "When I think about the midterms, I need to keep thinking about how we keep investing to meet customer demand." She said the business grew 70% quarter on quarter in the third quarter, with enterprise demand doubling, and pushed back on the idea that OpenAI has scaled back its fundraising. Separately, OpenAI signed security firm Hacktron, whose researchers took over ChatGPT and Codex accounts via a single sign-on flaw in July, as a partner to test how OpenAI's systems connect to outside services, and raised top bug-bounty payouts to as much as $30,000 for critical findings.

Why it matters: OpenAI is answering the political backlash and the smaller raise with growth numbers, and it's giving its first public sense of momentum the same week Anthropic starts selling its own growth story to public investors.

Sources: Reuters via BusinessDay · OpenAI news · WebProNews on Hacktron

Microsoft winds down its Oracle AI cloud rental

Background: On Friday we covered Microsoft tightening oversight of data-centre projects run by smaller cloud providers such as Nebius, IREN and Nscale. Oracle signed a multi-year deal in 2023 to supply Microsoft with AI computing for Bing.

What happened: Bloomberg, relayed by Reuters, reports that Microsoft declined to renew the Oracle arrangement and has been switching off rented capacity in recent months, because the hardware runs on older Nvidia chips that no longer fit its needs. Oracle says demand for its AI infrastructure "remains strong, including for earlier-generation GPUs." IREN also said it has delayed some deliveries after Microsoft asked for network design changes, pushing more than $4 billion of annual revenue capacity from year-end to January, possibly late February.

Why it matters: The biggest buyer of rented AI computing is getting choosier. That's a warning for anyone whose business depends on leasing older chips, at a time when investors are already nervous about AI spending.

Sources: Reuters via Indian Express

Key hire: Meta's strategy chief Dave Wehner becomes CFO of Fireworks

Background: Fireworks is a four-year-old company that lets businesses customise and run open AI models. It was valued at $17.5 billion in July and is backed by Nvidia and Sequoia. Wehner joined Facebook in 2012 and was its finance chief from 2014 to 2022.

What happened: Wehner announced on Friday that he's leaving Meta to become Fireworks' CFO, in charge of scaling its finances and operations. Business Insider reports Fireworks is in early talks to raise again at a $30–40 billion valuation. Fireworks disclosed a security incident in October involving misuse of internal credentials, and said it found no evidence that customer models or data were accessed.

Why it matters: A senior Big Tech finance leader is moving to an AI infrastructure startup, which usually signals preparation for a much larger raise or, eventually, a listing. It also shows that companies serving open models are drawing serious talent.

Sources: CNBC

Key departure: Codex product lead Rohan Varma leaves OpenAI

Background: Varma was Cursor's first product manager before leading product for Codex, OpenAI's coding agent. This week OpenAI also hired Cat Wu, Claude Code's former product lead, to lead product management for Core Products, the group behind Codex.

What happened: Varma posted on X on Friday that he has left OpenAI, describing his time on Codex as spanning three or four company "evolutions." He didn't name his next employer and said he'd take a few days off first.

Why it matters: Product leadership at the top coding-agent teams keeps changing hands between Cursor, OpenAI and Anthropic. Where Varma lands next is worth watching for any competitor in coding agents.

Sources: RuntimeWire · The Next Web on Cat Wu

Competitor product: Claude Code cloud sessions now set up a repo by themselves

Background: Claude Code's cloud sessions run Anthropic's coding agent on remote machines. Until now, users had to configure dependencies, network access and secrets themselves before an agent could run a project.

What happened: On 6 November Anthropic said that on its Pro, Max and Ultra plans, Claude now works out which dependencies, network access and API keys a repository needs, then saves that environment for next time. Keys are stored so "Claude never sees them": Anthropic's proxy attaches them to outgoing requests after they leave the session's machine. Separately, Anthropic said it will remove users' account email addresses from Claude Code's context after a public bug report showed the address could end up in outgoing web requests.

Why it matters: Getting a repo to run is often the slowest part of starting a cloud agent. Anthropic is cutting that setup time, the same ground Cursor's cloud agents and OpenAI's Codex are competing on.

Sources: SysNote · Aivor (German)

SpaceX publishes the full Flight 15 plan: catch both stages, 22 hours in orbit

Background: SpaceX is targeting Friday 13 November for Starship Flight 15, and Musk said on Friday it would try to catch both stages with the tower.

What happened: SpaceX's official mission page now gives the details. The 75-minute window opens at 4:45pm Central, which is 10:45pm UK time, pending FAA approval. It will be the first catch attempt for a Super Heavy V3 booster. Starship will deploy 26 Starlink V3 satellites, three of which are modified to photograph the ship's heat shield, then make about 14 orbits over roughly 22 hours. If checks pass, it will fly back over northern Mexico and southern Texas for a catch at Pad 2.

Why it matters: Recovering both stages intact would be the first full recovery of an orbital rocket, which is the key step towards the cheap, frequent launches SpaceX's Starlink and AI-compute plans depend on.

Sources: SpaceX · KRGV

It was a quiet weekend for new funding: no new AI round above $2 million was announced on 7 or 8 November (AI Funding Tracker).

← Saturday, 7 November 2026 Tuesday, 10 November 2026 →

Get the weekly sift

The best new AI and agent releases, with one-line verdicts. Occasional picks, no spam.

Privacy policy: your email goes to our email provider, Resend. Unsubscribe any time.

News is editorial and never paid for. Sponsored content on ToolSift is always labelled.