Corporate America Caps AI Spending, Exposing Bubble
OpenAI weighs drastic token price cuts as enterprises cap spending and Chinese rivals undercut Western prices, signaling the end of the AI gold rush and return of rational capital allocation.
OpenAI is weighing drastic token price cuts to win back enterprise customers, a move that signals the end of the AI gold rush as corporate America caps spending and Chinese rivals undercut Western prices. The structural collapse of AI as a premium-priced monopoly has arrived. Major corporations are slamming the brakes on unsustainable token expenditures while market indicators flash red.
The Wall Street Journal reported Wednesday that OpenAI is considering "drastically lowering the prices it charges users" in anticipation of similar reductions from rival Anthropic. The pricing war marks more than a competitive maneuver. It signals the end of the tokenmaxxing era, where enterprises threw billions at AI tools with little regard for return on investment.
Silicon Data's LLM Token Expenditure Index has dropped for seven consecutive days, falling to levels last seen in mid-January before the agentic AI craze. Macro strategist Andreas Steno Larsen calls the index "the single most important chart in the market right now." He warns that weakening token pricing could signal the end of hardware and data-center trades for this cycle. Institutional investors are hunting for what JPMorgan's Mark Schilsky calls the "second derivative kink" in AI revenue growth, a warning sign that the music may stop for the AI trade.
Corporate America has slammed the brakes on tokenmaxxing, the practice of throwing tokens at every workflow regardless of ROI. Uber burned through its entire 2026 AI budget in four months, with 95 percent of engineers using AI tools monthly and about 70 percent of committed code originating from AI. Yet Uber COO Andrew Macdonald admits the disconnect between spending and results remains stark. "That link [between AI token usage and shipping successful products] is not there yet, right?" Macdonald told a podcast interview May 26.
Uber now caps employee spending at $1,500 per month per tool. Coinbase instituted weekly price caps from $500 to $5,000 per employee based on job level after usage went "parabolic" following Claude Opus 4.6's February launch. Salesforce CTO Parker Harris articulated the corporate awakening to capital allocation reality: "We gotta run a business, we're a public company. We can't tell our investors like, 'Yeah, sorry, we gave half of our upside this year to Anthropic so they can go public.'"
The valuation disconnect between corporate budgeting and AI startup ambitions could not be more stark. Anthropic raised $65 billion in Series H at a $965 billion valuation May 28, while OpenAI filed a confidential S-1 June 8 at an $852 billion valuation. Yet OpenAI CEO Sam Altman acknowledged the cost equation has fundamentally changed. "The issue never came up. People were totally happy with the amount they were spending — to all of a sudden, a huge issue," Altman told an enterprise event June 2.
OpenAI is projected to spend roughly 14 times more cash than Anthropic before reaching profitability, with breakeven projected around 2030 compared to Anthropic's 2028 target. Both companies are projected to spend nearly $65 billion combined this year on computing, training, and operational expenses.
As Western firms slash prices, Chinese competitors are poised to exploit the collapsing price floor. DeepSeek permanently cut its V4-Pro pricing by 75 percent May 23, while Xiaomi slashed API costs for its MiMo-V2.5 model by up to 99 percent May 27. Xiaomi's MiMo-V2.5 model now processes 1.7 trillion tokens in seven days, representing growth exceeding 999 percent from the prior week.
OpenAI confirmed seeing "some evidence" of distillation from Chinese groups, while White House AI czar David Sacks said there was "substantial evidence" DeepSeek distilled knowledge from OpenAI models. DeepSeek founder Liang Wenfeng's philosophy underscores the strategic threat: "What China and Chinese companies lack is not capital, but rather confidence and the ability to organize and manage talents to realize true innovations."
Citadel Securities flagged the fundamental shift in a research note titled "Tokenomics," arguing that "adoption is no longer about what AI can do in principle. It's becoming about the price and scarcity of the inputs needed to run it at scale." Hedge fund manager Jim Chanos noted the inherent contradiction: "Price cuts anywhere in the AI ecosystem when demand for 'compute' is supposedly infinite, seems... problematic."
Florida Governor Ron DeSantis captured the enterprise sentiment shift: "Businesses are increasingly concluding that the value of the AI model is not worth the extravagant costs." GitHub switched to per-token billing June 1, triggering user reports of costs surging from under $45 monthly to over $847 — a 19-fold increase.
Goldman Sachs' Rich Privorotsky predicts 80 percent of AI workloads will migrate to models that are 99 percent cheaper within 12 to 18 months. Coinbase CEO Brian Armstrong projects the same migration, with only 20 percent of tasks requiring extreme intelligence remaining on frontier models. The bubble has burst. Rational capital allocation is finally returning to the technology sector, and American businesses are reclaiming control over their spending.