Big Tech's June selloff was not a verdict on AI failing. It was a warning that investors are done pretending the bill does not matter.
The AI trade finally hit the part of the story where someone has to pay for the data centers, chips, power contracts and borrowed money. The Magnificent Seven, Broadcom and Oracle have lost roughly $2.7 trillion in market value this month, according to Yahoo Finance analysis, after investors began treating AI infrastructure less like a promise and more like a cost center with a deadline.
Alphabet made the anxiety visible on June 22. MarketWatch reported that Google's parent lost about $225 billion in market value in one session, its largest single-day wipeout on record, after the stock fell 5%. The Times put the hit closer to $240 billion, with shares closing down 5.1% at $348.78. Either way, you don't need to split hairs. A company worth more than $4 trillion does not lose that much in a day because investors are mildly concerned.
The trigger was not one clean thing. Noam Shazeer, a Gemini co-lead and one of the authors of the transformer paper behind the generative AI boom, left for OpenAI. John Jumper, the Nobel prize-winning DeepMind researcher known for AlphaFold, left for Anthropic. At the same time, investors were staring at Alphabet's planned AI spending, which the New York Post reported could reach $180 billion to $190 billion in fiscal 2026, much of it for compute and data centers. Talent is leaving. The bill is rising. That is a bad pairing.
Then the selling spread. The Guardian reported that the Nasdaq closed 2.2% lower on June 23, while the S&P 500 fell about 1.4%. Nvidia dropped 4.2%, Micron fell more than 13%, and the pressure moved through Asia, where South Korea's Kospi briefly plunged about 10% as SK Hynix and Samsung Electronics were hit. This was not just a Google story by lunchtime. It had become a referendum on the whole AI stack.
The cash problem is no longer theoretical
For two years, investors were happy to buy the companies selling the picks and shovels. Nvidia became the most valuable public company in the world because every major platform needed its GPUs. Broadcom, TSMC, Micron and SK Hynix rode the same logic. If Microsoft, Amazon, Meta and Alphabet were going to build the AI economy, the suppliers would get paid first.
That logic still works, but it has a harder edge now. If every hyperscaler keeps spending at once, margins get squeezed. If any of them slows down, suppliers get questioned. That's why a selloff in Alphabet can hit Nvidia and memory stocks in the same wave. The market is not only asking who wins AI. It is asking who funds it without damaging the business investors already own.
Amazon shows you the problem cleanly. MarketWatch reported last month that Amazon's free cash flow for the 12 months ended March 31, 2026 fell to $1.2 billion from $25.9 billion a year earlier, largely because property and equipment spending jumped by $59.3 billion. That is not accounting trivia. Free cash flow is what supports buybacks, acquisitions, debt reduction and patience from shareholders. When it disappears, the story changes.
The capex numbers explain why the market is jumpy. Tom's Hardware, citing first-quarter earnings compiled by the Financial Times, reported that Alphabet, Amazon, Microsoft and Meta are expected to spend about $725 billion on capital expenditure in 2026, up 77% from $410 billion last year. Goldman Sachs has gone even further, with Business Insider reporting that the firm expects AI spending by those four companies to reach $5.3 trillion by 2030. Frankly, a number that large should make you pause even if you believe every bullish claim about AI.
There is still real revenue here. Microsoft has turned Copilot into a broad enterprise push. Amazon is selling AI infrastructure through AWS. Google Cloud has been growing fast. Meta is trying to use AI across advertising, recommendations and consumer assistants. These are not science projects being run out of a spare office. They are central to the biggest companies in the market.
But investors are no longer paying only for ambition. They want proof that AI spending turns into durable cash, not only higher depreciation and bigger power bills. You can see why. A data center full of accelerators is useful only if customers pay enough, often enough, to justify building the next one. Otherwise the spending looks less like dominance and more like an arms race nobody knows how to stop.
The uncomfortable part is that Big Tech may have no real choice. If Alphabet spends less, OpenAI and Anthropic look more dangerous. If Amazon slows AWS investment, Microsoft can use capacity as a weapon. If Meta stops buying chips, it risks falling behind in ads and consumer AI at the same time. So the companies keep writing checks, and investors keep asking when the checks come back with interest.
The story is current because the market is still sorting out that answer this week. The selloff may pass. These companies are not weak, and AI demand has not vanished. But the easy phase is over. From here, you should watch free cash flow as closely as model demos, because Wall Street has started doing exactly that.
Also read: Goldman Sachs leads $110 million bet on Taktile as banks move from AI pilots to autonomous decisions • OpenAI and Broadcom unveil Jalapeño, a custom inference chip that puts Nvidia's pricing power on notice • Vinod Khosla wanted every dollar of Runlayer's $30 million round and that tells you everything about where enterprise AI is heading
