Model

Goldman Sachs pulls Claude Code and GitHub from its bankers two weeks before Anthropic's $1.5 trillion IPO

Internal guidance updated in October ends months of trader and banker access to Claude Code and GitHub, after a year in which Goldman's own executives championed employee vibe coding. For Anthropic and Microsoft, seat expansion at a marquee buyer just stopped, two weeks before Anthropic's IPO pricing.

Vincent JiangVincent Jiangthree-minute read
Share
Dario Amodei at President Trump's super intelligence lunch, White House, September 2026
1 / 6Slide 1 of 6
Dario Amodei, Anthropic's co-founder and chief executive, at President Trump's luncheon on super intelligence at the White House on September 29, 2026. His company's IPO is expected to price in November.

Goldman Sachs−2.68% — Goldman Sachs, down 2.68 percent today has cut most of its traders and bankers off from advanced AI coding tools. Internal guidance updated in October limits Anthropic's$350B — Anthropic, private, latest valuation $350B Claude Code and Microsoft's−0.74% — Microsoft, down 0.74 percent today GitHub to engineers only, after months in which non-technical staff used them to follow client trading flows, simulate market shifts and draft text and visuals 12. Only the "strats," a group whose work blends software engineering, quantitative research and trading, kept access alongside the engineers 2. Bloomberg reported the change on November 2, roughly two weeks before Anthropic's IPO pricing 2.

The bank that preached vibe coding just revoked it

In July, Goldman chief information officer Marco Argenti praised "vibe coding" by non-technical staff: a trader builds a quick prototype with AI, then hands it to the bank's roughly 12,000-person engineering pool for polishing 23.

In August, Goldman partner Chris Churchman, who co-chairs the bank's Global Banking and Markets AI working group, warned that the spread of AI across Wall Street risks degrading the analytical abilities of the next generation of bankers 2.

By October the risk camp had prevailed, and this is the bank's second retreat this year: it earlier restricted Hong Kong staff from using Claude 2. A spokesperson called it the end of the experimentation phase, with tools going to those "best suited to leverage them for the long term" 1. The October guidance implies the bank could not yet confirm the tools paid off for non-engineers; it does not report a measured productivity verdict.

Who pays when a marquee buyer rations seats

The restriction means seat expansion at one of the most prestigious enterprise AI buyers has stopped for both Anthropic and Microsoft's GitHub business. The read-through lands two weeks before Anthropic's IPO, expected to price in November at a valuation around $1.5 trillion 4.

Anthropic's sequential revenue growth fell from 144% in Q2 to 47% in Q3

  • Sequential revenue growth
  • Estimate
0%50%100%150%Q4 '25Q1 '26Q2 '26Q3 '26144%
Data
Sequential revenue growth
Q4 '2548%
Q1 '26132%
Q2 '26144%
Q3 '26 (estimate)47%
Quarter-over-quarter revenue growth from Anthropic's amended S-1, as reported by The Motley Fool, November 2, 2026. Q3 dashed: figure from the amendment filed ahead of the IPO.4

And the tape is already skeptical. Anthropic's amended prospectus shows third-quarter revenue just shy of $17 billion, up about 47% sequentially after 144% in Q2 4.

The filing cites capacity constraints, summer seasonality and price cuts as explanations, and it discloses that price reductions on new models and contractual discounts cut Q3 revenue by about $600 million 4. The same week, ServiceNow−3.02% — ServiceNow, down 3.02 percent today fell 26.9% on October 29 even after raising its AI contract target to at least $1.7 billion, on a forecast of slower backlog growth 5. Enterprise AI is no longer graded on adoption; it is graded on whether adoption pays.

What to watch

Anthropic's amended prospectus discloses customer concentration but does not break out seat-level retention, so its roadshow disclosures are the checkable test: enterprise net revenue retention and any named financial-services customer. When a bank this prestigious rations the tool after its own review, other enterprise buyers' renewal math is the next thing to price.

Get Model by email, free

The AI labs, their models, and the business behind them. In our daily email, every weekday.

Loading GS

Reader comments

0 comments

    Model

    Meta's Q3 earnings fell to $3.83 against $5.46 expected, on a $9.89 billion teen-harm legal charge

    Excluding the charge Meta beat, and Businessweek's settlement pricing implies only about half the pending teen-harm claims are covered, with 2,000 school districts unpriced. The reserve note in January's filing decides.

    Model

    Nvidia shows how to run Z.AI's GLM-5.3 on $190,000 of Nvidia hardware, with no per-token bill to Z.AI

    The tutorial walks enterprises through serving GLM-5.3 on two DGX Stations, roughly $190,000 at listed reseller prices, with no per-token bill to Z.AI. For a stock carrying 118.6 times sales, that reads as a price list for everything Z.AI sells.

    Model

    Anthropic's Claude models tie at 58 on a coding index Cognition partly built, days before its Nasdaq IPO

    A third of the refreshed index runs on FrontierCode, a benchmark built by Cognition, whose Devin agent announced one of the two tied winners, Claude Fable 5.2, on the day of its release. Anthropic's model lists on Nasdaq on November 19; Cognition's Devin has offers from $70 billion on the table.

    Model

    Anthropic's Fable 5.2 passes 41.3% of real freelance jobs before its Nasdaq debut

    The one independent test of finished freelance work puts Anthropic's Fable 5.2 first at 41.3%, and the model still fails most of it, days before a Nasdaq debut priced at up to $1.4 trillion.

    Model

    Hedra cut Krea 3's edit time from 22.9 to 3.3 seconds on AMD, and the app doesn't own the speed

    The eight-second edit Krea advertises is manufactured in the serving stack, and a partner's measurement shows the app does not own the speed. Inference efficiency accrues to the optimizer and the chip vendor, which caps a per-image app's margin and hands AMD a demand channel beyond hyperscalers.

    Model

    xAI ships AI teammates at $30-$200 a month, ten days after Musk said serving them loses money

    The published card sets a chatbot price against an always-on machine the CEO says loses money on every subscriber; SpaceX shareholders fund both sides of that trade in one stock.

    Subscribe to TheInference for free