Billions poured in. Expectations soared. Then the bills arrived.
Early this month, Uber capped its employees’ use of AI coding tools. The annual budget for one popular system lasted just four months. A senior leader admitted the expense had grown “harder to justify,” with little evidence it produced better features for riders. Similar stories now surface from conference rooms across tech and beyond.
Amazon killed an internal ranking of AI token consumption after staff gamed it with pointless queries. One executive sent a blunt note: “Please don’t use AI just for the sake of using AI.” GitHub shifted its Copilot coding assistant to usage-based pricing. Developers suddenly confronted the real expense of heavy reliance. These aren’t isolated gripes. They mark a broader turn.
The AI investment wave that lifted markets through 2025 now faces its payback phase. Companies that rushed to embed the technology everywhere discover a stubborn gap. The models perform impressively in narrow applications. Scaled across thousands of workers and workflows, the economics falter. Savings fall short of forecasts. Costs mount faster than returns.
From Mania to Measurement
Analysts once debated whether AI would deliver any meaningful automation. Then tools like Claude Code and autonomous agents ignited a frenzy. Enterprises scrambled to maximize adoption, track usage, and broadcast productivity wins. That phase has given way to hard questions about value.
A Bain & Company survey of 951 large firms found AI-driven savings running below projections even as most planned higher outlays. “The technology worked. The value didn’t arrive,” the report stated. OpenAI Chief Executive Sam Altman called the revenue-visibility issue “the most fair criticism” of the moment, according to reporting by Axios (https://www.axios.com/2026/06/06/ai-bubble-economy-growth).
But the discomfort runs deeper than one survey. A February 2026 study from the National Bureau of Economic Research found that 90% of firms reported no measurable impact from AI on workplace productivity. Executives nevertheless projected gains of 1.4% in productivity and 0.8% in output. The pattern echoes the old productivity paradox of the 1980s and 1990s, when computers spread widely before their economic benefits became clear.
Spending continues at startling scale. US mega-cap tech companies were expected to commit $1.1 trillion on AI infrastructure between 2026 and 2029, with total AI-related outlays surpassing $1.6 trillion, Wikipedia’s compilation of analyst estimates shows (https://en.wikipedia.org/wiki/AI_bubble). Morgan Stanley projected global data-center investment near $3 trillion from 2025 to 2028, half of it backed by private credit. OpenAI alone has talked of $1.4 trillion over eight years for new data centers while reporting just $13 billion in revenue. The company anticipates annual losses through 2028, including a $74 billion operating loss that year. Deutsche Bank analyst Jim Reid has estimated cumulative losses for OpenAI at $140 billion from 2024 to 2029. Documents obtained by The Wall Street Journal show the firm projecting significant profits only in 2030.
Ray Dalio warned this month that the AI surge displays classic bubble traits. “All great technology changes produce bubbles,” the Bridgewater founder said in a Bloomberg interview (https://www.bloomberg.com/news/videos/2026-06-03/dalio-ai-bubble-to-burst-as-wealth-converts-to-money-video). He expects wealth built on paper to convert into harder money, with painful consequences.
Gary Marcus, longtime AI critic, declared in early June that 2026 would likely see the bubble unwind. Whether suddenly or gradually, the economics no longer add up for many deployments, he argued on Substack (https://garymarcus.substack.com/p/the-ai-bubble-is-all-over-now-baby-blue). Cory Doctorow made a sharper distinction on Medium. The early internet grew more profitable with each new user. Generative AI grows less profitable. “Every new AI user makes AI less profitable, as does every new use for AI, and each generation of AI loses more money than the last,” he wrote (https://doctorow.medium.com/https-pluralistic-net-2026-05-26-the-ai-will-continue-until-morale-improves-2582915489de).
Market concentration tells its own story. In late 2025, the five largest companies accounted for 30% of the S&P 500 and 20% of the MSCI World index, the highest levels in half a century. AI-related stocks drove roughly 80% of US market gains that year. Nvidia’s market value topped $5 trillion in October 2025, exceeding the GDP of every nation except the United States and China.
Yet cracks have appeared. When Broadcom reported strong AI revenue but held its longer-term outlook steady, the Nasdaq dropped 4.2% and the Philadelphia Semiconductor Index fell 10.3%. Investors wanted proof that demand kept accelerating. They received reminders that optimism carries limits.
Executives inside the builders see the tension. Sundar Pichai, Alphabet’s chief, has spoken of “elements of irrationality” in the trillion-dollar investment boom. Sam Altman himself has conceded overexcitement. The technology remains potent. The question is whether the current spending trajectory can sustain itself without clearer returns.
Some voices push back. Fidelity Investments noted in early 2026 that several classic bubble signals had not yet materialized. Free cash flows had not shrunk despite heavy capital expenditure. Corporate cross-holdings of US stocks had not spiked. Valuations, while elevated, sat below dot-com peaks on certain measures. JPMorgan and Goldman Sachs analysts have argued that AI ties to genuine enterprise revenue and profit growth, distinguishing it from pure speculation.
Still, the internal pushback from adopters carries weight. Uber did not abandon AI. It simply stopped treating it as free and unlimited. Amazon did not reject the technology. It rejected mindless deployment. These adjustments suggest a maturing market. Precision over proliferation. Targeted gains rather than blanket transformation.
The original Yahoo Finance piece that framed this shift as “revenge” captured the mood perfectly (https://finance.yahoo.com/sectors/technology/articles/revenge-ai-bubble-125911121.html). The bubble’s revenge is not a market crash. It is the quiet realization inside conference rooms that the easy wins were never easy. The models impress in the lab. They demand discipline, curation, and realistic expectations when loosed on real operations.
Productivity may yet rise. Certain workers already achieve dramatic output lifts when tools are applied with care. Chipmakers, model developers, and specialized users capture clear value today. The broader economy has started its slower absorption phase. History shows such technologies eventually pay off. They rarely do so on the timelines or at the multiples that fueled the initial mania.
So the reckoning continues. Companies trim wasteful usage. Vendors adopt usage-based pricing to reflect true costs. Boards press for measurable returns. Investors parse every earnings call for signs that capital expenditure translates into sustainable profit. The technology isn’t disappearing. The fantasy that it could be sprayed everywhere and pay for itself just did.
But the real test lies ahead. If 2026 brings slower adoption, tighter budgets, and more honest accounting, the industry may emerge stronger. If losses keep mounting without credible paths to breakeven, the unwind could prove more abrupt. Either way, the era of unchecked exuberance has ended. Measurement has begun.
WebProNews is an iEntry Publication