The $300 Billion Number
The four largest hyperscalers reported earnings within an 80-second window on Wednesday afternoon. The combined 2026 AI capital expenditure guidance: somewhere between $625 billion and $700 billion across the group. Meta alone guided $115-135 billion. Alphabet guided $175-185 billion. Amazon roughly $200 billion. Microsoft is on track for similar numbers.
For context: the entire US defense budget is about $895 billion. Big Tech is now spending close to that on data centers, GPUs, and AI infrastructure in a single year.
Three of the four stocks fell on the news.
The Beat-but-Disappoint Pattern
All four companies beat Wall Street consensus. Microsoft revenue of $82.9 billion, up 18% year-over-year. Net income up 23% to $31.8 billion. Alphabet net income of $62.58 billion — up 81% year-over-year. Meta net income of $26.8 billion, up from $16.6 billion. Amazon AWS posted strong cloud growth.
These are some of the best earnings reports in corporate history. The market sold three of the four anyway.
The reason: expectations were already in orbit. After 18 months of "AI will transform everything" hype, investors needed not just a beat but a guidance raise. Meta said revenue growth would be flat in Q2. Microsoft and Amazon delivered mixed cloud margins. Alphabet was the only one to deliver clean acceleration — and it was rewarded with a 6% pop.
The Capex Question Nobody Wants to Answer
$300 billion in capex needs to generate returns. The math is brutal: at a 15% return on invested capital, that capex needs to add $45 billion in annual operating profit. Across the four companies. By when?
Microsoft and Alphabet have a clear monetization path through cloud and search. Meta's path is murkier — they are spending more than Apple, Google or Microsoft combined while their revenue is half the size of any of them. Amazon is somewhere in between.
If AI revenue does not show up in the next 18 months at the scale that justifies the spend, this is the largest capex bubble since the dot-com era. The market is starting to ask the question. The answer arrives in 2027 earnings.
The Sell-Side Pattern
Bank of America cut Meta's price target. Morgan Stanley reiterated buy on Alphabet. Goldman flagged "expectations risk" across the cohort. Every major sell-side desk now has the same observation: AI capex is real, AI revenue is real, but the gap between the two is widening.
That gap has to close one way or another. Either revenue accelerates dramatically, or capex pulls back. Neither outcome is good for current valuations.
What Happens to the Supply Chain
The biggest beneficiary of $300 billion in capex is Nvidia. Then TSMC, ASML, Broadcom, and the rest of the AI supply chain. If Big Tech capex holds, the supply chain rallies even if the end-customer stocks struggle.
If capex starts pulling back — even modestly — the supply chain unwinds fast. Every $1 of capex cut is multiplied through Nvidia's 60-70% gross margins, ASML's monopoly pricing power, and TSMC's utilization assumptions.
Watch for capex revisions in the next two quarters. That is the leading indicator for whether the AI trade has 18 more months of runway or whether April 2026 was the inflection point.
The Apple Question
Apple reports tonight after the bell. They have been the laggard in AI capex — Tim Cook has refused to throw billions at infrastructure that competitors are committing to. The market has rewarded Apple with the worst Mag 7 returns of the year.
If Apple announces a major AI capex commitment, the entire group rerates higher. If they hold the line on disciplined spending, the divergence from Google/Meta/Microsoft/Amazon widens further.
Apple has a new CEO coming in September. John Ternus replaces Tim Cook. Many investors are quietly waiting to see whether Ternus reverses Cook's capex restraint. Tonight is the first hint.
The Bottom Line
Big Tech delivered the best quarterly numbers in corporate history and three out of four stocks fell. That tells you everything about where expectations are. The AI bull market has not died — but it has moved past the "beat estimates" stage. From here, it requires acceleration, monetization proof, and capex discipline.
None of those are guaranteed. The next 18 months will determine whether the $300 billion bet was visionary or reckless.
