Analysis
Meta, Microsoft, Amazon and Apple all report quarterly earnings this week, carrying into the reports a combined roughly $725 billion in planned 2026 capital expenditures -- up 77% from $410 billion in 2025 -- almost all of it earmarked for AI data centers, chips and power. It is, by a wide margin, the highest-stakes earnings week of the year for the AI trade, arriving just as investor patience with that spending has visibly started to run out.
The mood shift is already showing up in how the market treats capex guidance rather than results. Intel fell as much as 11% after its own earnings last week despite beating estimates and raising its quarterly profit forecast, purely because of concerns tied to its capital spending plans -- a reaction that would have been unthinkable a year ago, when almost any AI capex increase was read as a bullish signal. One analyst summarized the flip bluntly: "it used to be the more the better, but now it is the less the better."
Alphabet has already set an uncomfortable bar for the rest of the group this earnings season, reporting its first-ever negative free-cash-flow quarter as AI infrastructure spending outran its cash generation for the first time in the company's history as a public company. Individually, the spending plans are enormous even by this cycle's standards: Amazon's 2026 capex is projected near $200 billion, Meta's could reach $135 billion (up 79% year-over-year, with Barclays modeling close to a 90% hit to free cash flow), and Microsoft's trailing-twelve-month spend of roughly $97 billion is expected to climb toward $119 billion in the next fiscal year before jumping again the year after.
“The mood shift is already showing up in how the market treats capex guidance rather than results.”
The competitive logic underneath all this remains unchanged even as investor sentiment sours: none of the four companies can afford to be the one that under-invests in AI capacity if a rival's bet pays off, which is exactly the dynamic that turned Nvidia into a de facto lender to its own customers this month via the SK Group and OpenAI financing deals. But a market that spent 2025 rewarding capex increases is now punishing them, and this week's four reports will be the clearest test yet of whether hyperscalers can credibly show revenue and margin catching up to spending, rather than simply promising it will.
For VCs and founders, the read-through is direct: the hyperscalers' own investors are now demanding the same ROI discipline on AI infrastructure spend that portfolio companies get grilled on in every board meeting, and any pitch premised on "infinite hyperscaler capex forever" needs a plan B if this week's earnings trigger further capex skepticism. GPs marking AI infrastructure positions should treat this week's four reports, not the next funding round, as the more important valuation input.
The bear case is straightforward: if even one of the four misses on margin or offers cautious capex guidance, the market could extend Intel's post-earnings reaction across the sector broadly, dragging down every AI-infrastructure-adjacent name regardless of that specific company's fundamentals. The bull case is equally simple -- any hyperscaler that can show AI revenue clearly outrunning its spending this week could reset the entire market's mood heading into the back half of the year.
What to watch: Meta and Microsoft's specific AI-revenue disclosures (as opposed to just capex totals), whether Amazon's AWS growth reaccelerates enough to justify its spending pace, and whether Apple -- the group's smallest AI spender by far -- faces different scrutiny than the other three given its comparatively conservative capex posture.
