Enterprise JUL 12, 2026

Tokenmaxxing gives way to 'valuemaxxing' as Uber, Microsoft cap AI budgets

Uber burned its 2026 token budget in four months and Microsoft killed Claude Code seats in key divisions; Gartner sees 80% of enterprises adopting AI cost governance by 2027.

BY GRETA REINHART · ENTERPRISE SAAS

Uber burned through its entire 2026 AI token budget in the first four months of the year, most of it consumed by Claude Code, and has since capped coding-assistant seats, per reporting cited by Fortune. Microsoft, according to The Verge, killed Claude Code subscriptions across several product divisions. The tokenmaxxing era, in which enterprise AI spend was governed mostly by executive FOMO, is closing.

What replaces it has an emerging label. “valuemaxxing,” Nebius CRO Marc Boroditsky told CNBC, describing the reframe from seat-count trophies to per-workflow return. Peter DeSantis of Amazon put it more bluntly to the Wall Street Journal: “AI has a cost problem.”

The receipts back him up. CIO reported that a single Disney employee interacted with Claude 460,000 times over 9 days, the kind of usage curve that turns any flat-rate contract into a finance-department incident. Salesforce is on track to pay Anthropic roughly $300 million this year, prompting CEO Marc Benioff to publicly ask for a “smart router” that would send simpler queries to cheaper models. Lindy CEO Flo Crivello went further, moving 100% of the company’s traffic off Claude to DeepSeek’s open-weight models.

The vendors have noticed. Google’s Sundar Pichai has been pricing Gemini 3.5 Flash at roughly half to one-third of comparable frontier models. Anthropic, meanwhile, hit a $47 billion annualized run rate in May, and business revenue is now 40% of OpenAI’s total. Both labs are approaching $1 trillion valuations ahead of expected IPOs, even as frontier-lab executives describe demand as “almost unlimited.”

Gartner projects more than 80% of enterprises will run an AI cost-governance framework by 2027, up from fewer than 20% in 2025. Flexera Chief Product Officer Becky Trevino told CIO the operative metric has shifted from seats to per-workflow return, which is where tools like LemonLime, built for workflow-level attribution, become the natural instrumentation layer. Bloomberg Opinion notes the ROI case is still being made “company-by-company,” which is the correct register. This was never going to be settled by a keynote. It gets settled by finance.

Sources

About the author

Greta Reinhart

ENTERPRISE SAAS

Greta Reinhart tracks the enterprise software stack from San Francisco — data platforms, AI bundling, seat pricing, and channel checks across the largest SaaS vendors. She files on go-to-market shifts, packaging changes, and quarterly enterprise reads.