AI Spending Backlash Deepens: Tech Stocks Plunge as Investors Question the $725 Billion Capex Wave

The $725 Billion Question Wall Street Stopped Ignoring

For three months, the trade was effortless. Buy anything touching AI. Chip designers, data center operators, power utilities. The thesis needed no defending because the prices kept climbing. The wider tech sector had more than doubled since its 2022 trough as capital flooded into the AI supply chain 1.

Then Alphabet disclosed the scale of its planned AI capital expenditure, and the momentum broke. The tech-focused Nasdaq slid roughly 3% in a single session, with semiconductor stocks absorbing the heaviest losses. Nvidia and Intel dropped, pulling a global chip index lower 1. Asian semiconductor shares followed the sell-off the next morning, with SoftBank sinking 9% and Tokyo Electron losing more than 8% as losses spread across Pacific markets 2.

The combined AI capital expenditure projected for 2026 across Microsoft, Alphabet, Meta, and Amazon has reached an estimated $725 billion 3. That money is flowing now, not in some distant fiscal year. It is buying chips, building data centers, and financing the energy infrastructure to run them. And the market's response suggests investors are no longer willing to assume the buildout pays for itself.

Bank of America's Vivek Arya has framed the current moment as a shift: the industry is moving away from defending early returns on investment and toward tackling the physical bottlenecks in power supply and infrastructure, with demand strong enough to keep pushing forecasts upward 1. In that reading, the capital expenditure is not optional. It is table stakes.

A growing contingent of skeptics argues that shrinking enterprise technology budgets, combined with broader economic headwinds, signal the end of easy market gains 1. Market watchers are now asking whether genuine enterprise adoption of AI can support the stock prices that powered the rally 1. If companies are not yet generating meaningful revenue from AI tools, each quarterly report makes the spending harder to justify.

Chinese AI firms are producing capable models at a fraction of the cost their American rivals are absorbing 4. If lower-cost alternatives compress what US companies can charge for AI services, the return on hundreds of billions in capital expenditure shrinks before the infrastructure is even fully operational.

The damage has been concentrated precisely where the bets are largest. Danni Hewson, head of financial analysis at AJ Bell, noted that the FTSE 100 held above water partly because London markets carry relatively few technology stocks, cushioning them from the sell-off sweeping Wall Street 1. The companies absorbing the losses are the same ones writing the checks.

Analysts remain divided on what comes next. Some read the pullback as a routine correction after an extended rally. Others see the beginning of a deeper repricing of AI assets. What is clear is that the sheer scale of the investment thesis has made it impossible to ignore. For nearly two years, spending announcements functioned as a signal of conviction. Companies that disclosed the largest capex plans were rewarded with higher valuations. The logic was circular, but it held because nothing forced anyone to test it.

The coming earnings season changes that dynamic. Companies will need to show whether AI revenue is materializing into real earnings or whether the spending remains a wager on a payoff that keeps receding 1.

For anyone whose company is being told to spend on AI or risk irrelevance, the market is sending a more specific signal. The question was never whether to invest. It is whether the investment produces revenue before patience, and capital, run out. The companies that can demonstrate that will set the terms for whatever comes next. The ones that cannot are about to face a market that has started asking harder questions.