Uber Blows 2026 AI Coding Budget by April as Microsoft Revokes Claude Code Licenses

Companies across the AI industry are scrambling to manage runaway token costs after Uber “blew through its entire 2026 AI coding budget by April” and Microsoft “revoked its developers’ Claude Code licenses months after enabling them.”

“Google promises to reduce its water footprint by 100% by 2030, with 5 key commitments and an investment of 17 million”

Diario Ahora

TechCrunch says the shift is driven by higher token consumption even as “per-token prices have fallen,” as companies push for more AI adoption and increasingly autonomous agents.

Diario Ahora

At the same time, the Linux Foundation unveiled plans for the Tokenomics Foundation, described by TechCrunch as a new standards body meant to instill “the same cost discipline around AI tokens that FinOps did for cloud spend.”

In parallel, Numerama reported that the Linux Foundation’s June 3, 2026 press release sets an objective to define standards, benchmarks, and best practices for measuring and managing the cost of AI where “today there is almost total ambiguity.”

At a New York City event, Alexander Embiricos, OpenAI’s head of enterprise, told TechCrunch that customer conversations have shifted to spending control, saying, “Now the conversations are about, ‘hey, we’re spending so much.”

J.R. Storment, executive director of the FinOps Foundation, told TechCrunch that in April and May he started hearing from companies: “Oh my god, we are 3x over our entire 2026 token budget and it’s only April.”

Ecosistema Startup

Numerama framed the Tokenomics Foundation’s approach as structured around three pillars, including “An extension of FOCUS, a common billing standard already adopted for the cloud, which needs to be adapted to AI.”

Numerama also quoted Jim Zemlin, CEO of the Linux Foundation, saying, “tokens have become the new unit of measurement for tech spending,” as companies move generative and agent AI workloads from pilot to production.

TechCrunch reported that the token-cost scramble is unfolding alongside a market for tools and language to track spending, with startups, established vendors, and a new standards body “racing to give companies the tools and language to track what they spend.”

“New Story The Trillion-Dollar Token Trap: Why Microsoft and Uber Are Pulling the Plug on Cloud AI Billing by Anuj Ashok Potdar Senior backend engineer at KAYAK building systems at scale”

HackerNoon

J.R. Storment warned that token-cost tracking is a “trillions-of-rows-a-month data problem,” adding that teams must “fundamentally rethink your tooling, your specs and your accounting systems.”

In a separate development, an open-source tool called Headroom was described as compressing tokens before they reach the LLM, with the article stating it achieved “savings of $700,000 in a few months.”

The same piece said Headroom was presented in late May 2026 on the sidelines of the Linux Foundation’s Open Source Summit and that its creator estimated “up to 90% of tokens sent to an LLM in this type of workflow are superfluous.”