Staff Writer | NUPIAO
Tech titans are cranking up their AI infrastructure spending like there’s no tomorrow, and “burning cash” has become the only game in town for the U.S. tech sector right now.
Throughout July, the so-called “Magnificent Seven” tech stocks, along with AI computing and semiconductor firms, painted a picture that was both strikingly uniform and deeply divided. The month kicked off with a sector-wide pullback—the Philadelphia Semiconductor Index took a brutal 11% nosedive in a single week, with nearly 70% of semiconductor names falling over 20% from their highs. Investors were sweating bullets over whether these astronomical capital outlays would keep eating into corporate cash flow.
But then the late-July earnings floodgates opened, and the narrative flipped on its head. Microsoft surged 15.51% in a single day—its biggest one-day pop since 2008—igniting a full-blown rally across memory, equipment, and AI chip stocks. The Philadelphia Semiconductor Index skyrocketed 8.19% in one session, with Micron, SanDisk, and AMD all posting single-day gains north of 13%.
Every major player’s core business moves in July shared one unmistakable thread: they kept raising their full-year capital expenditure guidance, with nearly all the money funneled into AI servers, data centers, custom silicon, and memory capacity. A rough tally by NUPIAO shows that big tech companies, including Amazon, Alphabet, Microsoft, and Meta, collectively unveiled annual capex plans approaching $800 billion in just one month.
Google parent Alphabet fired the first shot in mid-July, hiking its full-year capex guidance to between $195 billion and $205 billion. CFO Anat Ashkenazi told analysts on the earnings call that Google Cloud’s backlog has ballooned past $514 billion, and the compute gap is likely to persist for at least two more years.
Microsoft held the line on its existing $190 billion capex plan, but its quarterly capital spending hit $41 billion—a staggering 70% year-over-year jump—all earmarked for GPUs to expand its Azure compute clusters. With paid Copilot users growing at a blistering pace, the investments are paying off in real revenue, which is exactly why Microsoft’s stock exploded at the end of the month.
Meta and Amazon both dropped heavy hints of even more spending in late July. After Meta released its Q2 earnings, it bumped its full-year capex guidance from $125 billion up to $130-145 billion, with quarterly capital expenditures of $31.08 billion—a massive year-over-year surge.
Meta CFO Susan Li told analysts on the call that the industry is staring at a long-term compute supply crunch, and the company plans to max out its data center construction pace between 2026 and 2027. That includes pushing forward its massive supercomputing campus projects in Texas ($14 billion) and Louisiana ($50 billion).
Here’s the kicker though: unlike other cloud giants, Meta doesn’t have a cloud business to sell compute to outsiders. Every single AI dollar has to be recouped through advertising. Its Q2 free cash flow cratered 91% year-over-year to just $784 million, and the stock got clobbered—down 8% in after-hours trading—making it the laggard in July’s capex expansion party.
Amazon, meanwhile, dropped its own bombshell expansion plan. CEO Andy Jassy announced on the July 30 earnings call that the company is raising its full-year capex expectation once again to $220 billion—a $20 billion bump from the initial February forecast.
Jassy’s take: the compute supply-demand imbalance is going to stretch all the way to 2028. “Even at this level of spending, we still can’t satisfy all of 2026’s compute demand,” he said. “And the demand for 2027 and 2028 remains red-hot.”
Amazon Web Services posted Q2 cloud revenue of $42.2 billion, up a scorching 37% year-over-year—the strongest growth in 18 quarters. Its AI business topped $25 billion in annualized revenue, growing triple digits. That gusher of cash flow more than offset the capex pressure, and Amazon’s stock popped nearly 10% after hours, standing in stark contrast to Meta’s stumble.
Hardware titan Apple also dropped its latest numbers on July 30. For its fiscal Q3 2026, hardware revenue grew across the board, but the weak Q4 production guidance and slower-than-expected services growth sent the stock tumbling 6.33% after the report.
Apple hasn’t gone all-in on building its own massive AI data centers, but it’s been pouring R&D money into custom on-device AI chips while locking in long-term supply deals for memory and advanced chips upstream. It’s essentially the same cost-expansion logic playing out at the top of the AI supply chain, dovetailing with the other giants’ compute buildout from the other end.
And let’s not forget Tesla, which reaffirmed its plan to spend over $25 billion in capex this year. Its AI chip factory in Texas is up and running, mainly relying on autonomous driving’s edge AI to drive compute expansion.
Semiconductor companies, as the direct beneficiaries of all this capex, spent July completely tethered to the tech giants’ moves, creating a neat supply-demand loop.
Intel reported on July 24 with quarterly revenue of $16.1 billion, up 25% year-over-year, and its data center AI revenue surging 59%. It raised its full-year capex to $20 billion, doubling down on advanced process fabs in the U.S. and locking in orders via ten long-term server CPU supply agreements with cloud providers.
Nvidia’s stock had a choppy July, dipping as low as $190 intraday as investors fretted about chip oversupply from the cloud giants’ relentless expansion. But end-market demand stayed red-hot, and the stock rebounded with the broader market late in the month, closing up 2.65% while holding its market cap near the $4.7 trillion stratosphere.
AMD, memory makers Micron and Western Digital, and equipment supplier Applied Materials all landed massive long-term orders from hyperscale cloud providers in July. Memory chip companies, riding the AI storage wave, saw their stocks explode late in the month, with SanDisk soaring nearly 26% in a single session.
As for this latest round of compute expansion, analysts point out that money is rotating from the Magnificent Seven into semiconductor names. The market is rewarding the beneficiaries of capex while questioning the cloud providers doing all the heavy spending—and this divergence is likely to stick around for a while.
BlackRock, the world’s largest asset manager, thinks the selloff was “overblown,” arguing that the market is confusing “a shift in the AI competitive landscape” with “an AI investment collapse.” BlackRock stresses that cheaper AI won’t kill investment—it’ll actually accelerate adoption across industries, expand the total addressable market, and ultimately drive even higher demand for data centers, memory chips, and power infrastructure. The firm maintains its overweight stance on U.S. equities and recommends keeping an eye on AI bottleneck areas like chips and electricity.
