Goldman Sachs has limited Claude Code and Microsoft’s GitHub coding tools to about 12,000 engineers, leaving most traders and bankers without access, according to Yahoo Finance. The move lands alongside signs that growth in business spending on AI tools is slowing, raising a pointed question for companies selling AI beyond engineering teams: will those new users keep the tools once the novelty meets a productivity review?
Watch Desk analysis
What happened
Yahoo Finance says Goldman changed its internal guidance in October after reviewing productivity and risk. Bloomberg had reported the restriction on 2 November, the article says. Traders had been using the tools to follow client flows and model market moves; the report says the tools are now limited to engineers.
The article also cites Ramp’s AI Index, which draws on transactions from more than 85,000 US businesses. Ramp found overall AI spending grew by less than 2% in October compared with September, while spending growth on Anthropic tokens slowed to 0.5%. Ramp economist Ara Kharazian described engineering as the largest category of enterprise AI spending and said it was starting to plateau. These are reported figures and one month’s snapshot, not proof of a settled trend. Yahoo Finance’s report also discusses possible growth in customer support and cybersecurity, but presents those as potential next markets, not demonstrated replacements for engineering demand.
Why it matters
Goldman offers a useful real-world test of how companies assess AI tools outside technical teams. The restriction does not show that the tools failed, but it does show that access can be narrowed after a productivity and risk review, even when staff have found uses for them.
For AI providers, expansion beyond software engineering may matter if spending growth in that core category cools. Yet wider deployment is not simply a matter of handing every department a licence. It depends on whether the tools deliver enough practical value to survive scrutiny from the people paying the bill.
Our read
This is a caution against treating enterprise AI adoption as a one-way ratchet. Goldman’s decision and Ramp’s October figures make a worthwhile signal, not a verdict on the market: one company and one month cannot establish that demand has peaked. The useful test is whether AI earns durable use in non-engineering work, rather than merely collecting seats and enthusiastic demos.
What to watch
- Whether Goldman restores access to traders or other teams after its review.
- Whether Ramp’s November data shows another slowdown in business AI spending.
- Whether customer support or cybersecurity becomes a measurable source of enterprise AI demand.
Discussion spark: Should companies restrict general AI access until teams can show clear productivity gains, or does that risk cutting off the experiments that might find them?
Sources and evidence
- Goldman Pulls Claude Code From Its Traders as AI’s Biggest Spend Category Is “Starting to Plateau” – Yahoo Finance (12 November 2026, 22:20 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.