January 29, 2026 delivered a textbook fork in big tech. Microsoft fell 10% after its earnings report — its worst single day since March 2020 — erasing more than $357 billion in market value. The same day, Meta surged more than 10%, adding over $176 billion. Both companies are pushing AI infrastructure spending to record highs, and the market handed down opposite verdicts. Reuters said it plainly: investors are now punishing AI spending “that delivers slower growth.”
This is not market gossip. How capital markets price AI capex will flow straight into how much compute is available to rent over the next year, where API prices settle, and which internal AI projects survive the next budget review.
Microsoft: A $357 Billion Single-Day Wipeout
Microsoft’s problem was not a lack of spending — it was spending without visible acceleration. Azure growth slowed to 39% from 40% the previous fiscal quarter. Quarterly capital expenditure including finance leases hit $37.5 billion, up 66% year over year and above the $34.31 billion Visible Alpha consensus. CFO Amy Hood’s explanation was that demand still outstrips supply: if every GPU delivered in the first and second quarters had gone to Azure, growth would have hit 40% that quarter. In other words, revenue is capacity-gated, not demand-starved.
The order book is the more uncomfortable detail. Microsoft’s commercial backlog rose to $625 billion, up 110%, including a $250 billion cloud agreement with OpenAI — OpenAI alone accounts for 45% of commercial remaining performance obligations. Evercore ISI argued the selloff was fueled by “concerns about OpenAI’s ability to meet funding commitments,” while calling that sentiment “overblown.”
Meta: Spending Nearly Doubles and the Stock Rallies
Meta’s script read completely differently. Fourth-quarter 2025 revenue grew 24% year over year, still driven primarily by online advertising. Then the company guided 2026 capital expenditure (including principal payments on finance leases) to $115–$135 billion — nearly double the roughly $72 billion spent in 2025. Mark Zuckerberg framed the outlay as “building personal super intelligence.”
The market’s response: more than $176 billion of market value added in a day. CNBC’s reading is blunt — investors saw evidence that AI spending is converting into ad revenue. When the return path connects visibly to the income statement, doubling spend becomes a reason to buy.
Two Control Cases: IBM and ServiceNow
Widen the lens and the same week offers two control cases. IBM beat expectations and rose 5%; its AI book of business more than doubled to $12.5 billion from about $5 billion a year earlier. Goldman Sachs said IBM looks on “track to complete its pivot to long-term growth,” and JPMorgan called it a “relatively defensive name.”
ServiceNow is the counterexample. It beat earnings estimates and still dropped 10% — down 25% since the start of 2026 and roughly 50% over the past year. What the market fears there is AI eroding the software subscription model itself. CEO Bill McDermott’s rebuttal: enterprise AI will be the biggest source of returns from the multitrillion-dollar AI infrastructure investment supercycle. Whether that lands remains an open bet.
Why Identical Spending Got Opposite Verdicts
Unpacked, the market priced three separate variables in one day. First, directness of the return path: Meta’s AI spending plugs straight into ad monetization, while Microsoft’s spending first becomes capacity resold to OpenAI — one extra layer of risk. Second, whether spend and revenue move together: a one-point Azure slowdown was enough to turn 66% capex growth from commitment into question mark. Third, counterparty credit risk: with OpenAI at nearly half of commercial backlog, OpenAI’s financing capacity became an implicit variable in Microsoft’s share price.
It tracks with what we argued in our 2026 opening outlook for AI: the AI infrastructure story is shifting from “how much got built” to “how much comes back.”
What It Means for Builders
Three practical effects. First, compute supply will not loosen soon — Microsoft itself says GPU deliveries are falling behind demand, so any product architecture heavily dependent on a single cloud needs a fallback plan. Second, the cost of capital will show up in pricing: as markets demand demonstrated returns on AI spending, cloud and model providers will pass costs into API and lease prices faster; teams building cost models should budget for that variable. Third, AI ROI has graduated from slide-deck vocabulary to a valuation variable — and the same logic applies internally, where AI spending without a legible payback path will be first in line when the next budget review comes around.
Sources
- Microsoft plunges, Meta rallies as investors demand AI payoffs — Reuters
- Meta and Microsoft stocks head in opposite directions after earnings — CNBC
- Meta Reports Fourth Quarter and Full Year 2025 Results — Meta Investor Relations
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.