The AI Labs Face a Budget Reckoning

OpenAI announced three new AI models on Friday—GPT-5.6 Sol, Terra, and Luna—but only a "small group of trusted partners" approved by the U.S. government will get access . The White House requested the restricted rollout because of the models' advanced capabilities, a source told CNN . Sol is OpenAI's strongest offering yet, with improvements across coding and biology, and the company calls it its most capable model for cybersecurity .

The timing is awkward. In April, Uber's CTO revealed the ride-sharing company blew through its entire annual AI budget in just four months . Flo Crivello, CEO of AI startup Lindy, switched his company off Anthropic's Claude models entirely this month, moving 100% of traffic to DeepSeek—and watched the cost curve "crash to the ground" . Jeff Henry, president of consulting at Highspring, said some clients are pulling back until they "can really start to prove an ROI," while others are waiting 12 to 18 months before making big spending decisions. "Everybody is experiencing the same spend crunch on AI" .

Can the Labs Grow When Customers Are Cutting Back?

The era of so-called tokenmaxxing—where employers incentivized developers to use as much AI as possible without worrying about the results—is ending . Uber implemented spending tiers on some AI tools this month, starting at a base level of $1,500 per month, though employees can request access to higher levels . Leading model developers, including Anthropic and OpenAI, have been slower to slash prices in recent months , even as the average cost per million tokens across major providers fell from roughly $10 to $2.50 in a single year, according to Ramp's enterprise spending data cited by research firm Artefact .

The pressure is showing up in market share. ChatGPT's share of global generative AI web traffic fell from 77.6% in May 2025 to 53.7% by April 2026, and for the first time, more companies tracked by the Ramp AI Index are paying for Anthropic than for OpenAI . OpenAI's annualized revenue grew from roughly $13 billion in 2025 to $25 billion by February 2026 —strong by any measure. But Anthropic's annualized revenue jumped from $9 billion at the end of 2025 to $30 billion by April 2026, a pace in a different category entirely .

Over 1,000 enterprise customers now spend more than $1 million per year on Claude. That number was 500 when Anthropic announced its Series G in February. It doubled in under two months . Claude Code, launched publicly in mid-2025, reached $1 billion in annualized revenue within six months, and since the start of 2026, business subscriptions to Claude Code have quadrupled .

Why Is the Government Vetting Model Releases Now?

The request to OpenAI comes after the administration placed an export control order on Anthropic, leading the AI company to pull its latest advanced models Mythos and Fable, which raised fears in Washington over their advanced cybersecurity capabilities . OpenAI and the administration view OpenAI's latest model as "on par" with Mythos .

President Donald Trump signed an executive order earlier this month that asked AI companies with advanced models to voluntarily submit them for government review 30 days before release, but the framework for that has not been established . In the interim, there's confusion among AI companies on who or which agency is directing AI regulation. The request to OpenAI came from the White House, whereas the export control ban on Anthropic came from the Commerce Department .

"We don't believe this kind of government access process should become the long-term default," OpenAI said. "It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them" . OpenAI said it's working with the Trump administration to help establish a framework for such assessments and to develop a "repeatable process for future model releases" .

What's Driving the Memory Chip Crisis?

The AI spending crunch is colliding with a supply crisis. DRAM prices have risen 80 to 90 percent so far this quarter, according to Counterpoint Research . Data centers now consume an estimated 70% of all memory chips produced worldwide, and analysts say the shortage will persist well into 2027 .

The voracious demand for HBM by hyperscalers such as Microsoft, Google, Meta, and Amazon has forced Samsung, SK Hynix, and Micron to pivot their limited cleanroom space toward higher-margin enterprise-grade components. This is a zero-sum game: every wafer allocated to an HBM stack for an Nvidia GPU is a wafer denied to the LPDDR5X module of a mid-range smartphone .

IDC's research manager Jitesh Ubrani has warned that PCs, tablets, and smartphones could see price increases of 10% to 20% by the end of 2026 . In Q1 2026 alone, consumer RAM prices inflated by up to 110% and SSDs surged by 147%, triggering a desperate stockpiling race among PC manufacturers . CNBC reported that DRAM prices have roughly doubled since early 2025, smartphone shipments are projected to decline 12.9% in 2026, and the PC market faces an 11.3% contraction .

What Changed This Week

The U.S. government asserted direct control over frontier AI model releases for the first time, forcing OpenAI to gate access to its newest models while the framework for doing so remains undefined. Meanwhile, the enterprise customers both OpenAI and Anthropic depend on are tightening budgets and demanding ROI after a year of unconstrained spending. The memory chip shortage that began as an AI infrastructure problem is now hitting consumer electronics hard enough to force price hikes and production cuts across the industry.

What to Watch

OpenAI plans to make GPT-5.6 Sol, Terra, and Luna generally available "in the coming weeks," according to the company's blog post. The Commerce Department and White House have yet to clarify which agency will lead AI model oversight or publish the regulatory framework Trump's executive order promised. Anthropic is reportedly targeting an October 2026 IPO, which would put both its $30 billion revenue run rate and its enterprise customer concentration under public market scrutiny for the first time. Memory manufacturers Micron, Samsung, and SK Hynix are expected to report earnings in the next two weeks, with analysts watching for guidance on when new fab capacity might ease the supply crunch.