Goldman, JPMorgan and Cisco Put Budgets and Guardrails on Staff AI Use

JPMorgan capped some engineers' use of Anthropic's Claude Code at $2,000 a month in September — one of the sharpest signals yet that corporate America is moving from open-handed AI adoption to disciplined, gated deployment.

Goldman Sachs, Cisco and JPMorgan are all putting new guardrails on workplace AI, according to a Business Insider report. The goal isn't necessarily to cut spending. It's to make sure companies squeeze real value from technology they're pouring ever-larger sums into, while trimming security risk.

"We need frontier models for certain things, particularly cyber," Goldman CEO and chairman David Solomon told Bloomberg TV this month. But "there are a lot of things where we don't need the frontier" to boost productivity. Solomon said the bank is rapidly expanding the range of AI tools offered to its 45,000 employees — just not the most advanced models for everyone, every day.

Employers are realizing AI use is "really picking up" and that they need guardrails, said Lari Hämäläinen, a senior partner in McKinsey & Company's technology, media and telecommunications practice.

How are companies rationing AI access?

Cisco took a staged approach. Chief product officer Jeetu Patel compared the company's early policy to learning to ride a bicycle: coding teams got months of unlimited tokens to build fluency before efficiency mattered, he said on the CXOTalk podcast last month. Now that usage has grown, managers across his 32,000-person team will receive token budgets.

JPMorgan tied its $2,000 Claude Code cap to productivity, quality, risk reduction and business impact, a spokesperson said. Rippling, a workforce-management software startup, set monthly caps in August after projecting token spending could hit 40% of its R&D headcount budget. Limits and more efficient model routing cut that projection to between 10% and 15% — while token use stayed near its previous peak.

"The point isn't to cut AI costs," Rippling chief product officer Matt MacInnis said in a statement. "We're actually fine with cost growth, provided it's productive."

Why did the tone shift from 'use more' to 'use right'?

Months ago, employers bragged about how many tokens their workforces consumed. Visa said it doubled monthly token use from 1 trillion in February to nearly 2 trillion in March, and gave prizes to its AI power users. Nvidia CEO Jensen Huang said in March the company aimed to spend up to $2 billion annually on tokens for engineers — and that he'd be "deeply alarmed" if a $500,000 engineer used less than $250,000 worth.

Dashboards followed. Disney's showed its heaviest AI users making hundreds of requests per day and burning tens of millions of tokens, Business Insider reported in April. KPMG US launched a dashboard in May for the roughly 10,000 workers in its advisory business to encourage more "frequent and sophisticated" use — and has since discontinued it, a spokesperson said, now that AI is part of staffers' everyday work.

What do the numbers say about ROI?

The spending curve is steep. Enterprise AI deployment spending hit $177 billion by September, up from $33 billion in all of 2025, according to Menlo Ventures. Yet only 13% of roughly 2,000 tech executives polled by Gartner from May through July said their organizations had gained significant value from AI.

Part of the problem is incentive. Workers with unlimited access have little reason to curb usage, said Alex Sobol, cofounder of the Millennium Alliance, a C-suite community. "They're going to use it in every possible way they can," he said.

Security adds pressure. A 2025 KPMG survey of more than 1,000 US workers found 44% used AI at work without authorization or in inappropriate ways, and 46% had uploaded sensitive company information or intellectual property to public AI platforms.

The message from the C-suite is now consistent: AI budgets will keep growing, but only where they're tied to measured productivity, contained risk and demonstrable business impact.