Summary: On January 29, 2026, Microsoft fell 10% and lost over $357 billion in market cap while Meta rose more than 10% and added over $176 billion. Microsoft’s Azure growth slipped to 39% despite $37.5 billion in quarterly capex, and $250 billion of its $625 billion commercial backlog came from OpenAI. Meta’s 24% revenue growth and $115–135 billion 2026 capex guidance were read as spending with a clearer ad payoff.
Big tech staged a clean natural experiment on January 29, 2026. Microsoft dropped 10% the day its earnings landed — its steepest single-session slide since March 2020 — wiping out more than $357 billion of market cap. Meta, on the same trading day, jumped more than 10% and picked up over $176 billion. Each is putting record sums into AI infrastructure, yet the market ruled in opposite directions. Reuters put it bluntly: the punishment now falls on AI spending “that delivers slower growth.”
This is more than trading-floor chatter. The price capital markets attach to AI capex feeds directly into how much rented compute capacity will exist over the coming year, where API prices end up, and which internal AI projects keep their funding when budgets are next reviewed.
Microsoft: One Trading Session, $357 Billion Gone
The ailment was never too little spending — it was money out the door with no acceleration to show for it. Azure’s growth rate slipped to 39% from 40% in the prior fiscal quarter. Capital expenditure for the quarter, finance leases included, came to $37.5 billion, a 66% year-over-year jump that cleared the $34.31 billion Visible Alpha consensus. CFO Amy Hood’s account: demand continues to run ahead of supply. Had the GPUs newly delivered across the first two quarters all been pointed at Azure, the quarter would have grown 40%. Put simply, revenue here is gated by capacity, not starved by demand.
What the order book is made of is the sharper worry. Microsoft’s commercial backlog climbed to $625 billion, up 110%, and $250 billion of that comes from the cloud agreement with OpenAI — a single customer carrying 45% of commercial remaining performance obligations. Evercore ISI read the selloff as fueled by “concerns about OpenAI’s ability to meet funding commitments,” though the firm judged that mood “overblown.”
Meta: A Doubling Budget and a Rallying Stock
Meta told the opposite story that same day. Revenue for the fourth quarter of 2025 posted 24% year-over-year growth, with online advertising still doing the heavy lifting. Then came the 2026 capital expenditure guidance (including principal payments on finance leases): $115–$135 billion, close to twice the roughly $72 billion of 2025 spend. For Mark Zuckerberg, the money is about “building personal super intelligence.”
Wall Street answered with more than $176 billion of added market value in a single session. CNBC’s interpretation is plain — investors finally had proof that AI spending converts into ad revenue. Once the return path visibly terminates in the income statement, a doubled budget reads as a buy signal, not a risk.
Two More Readings from the Same Week: IBM and ServiceNow
Pull the lens back and that same week supplies two more data points. IBM topped forecasts, adding 5%; its AI book of business has doubled and then some in a year, from roughly $5 billion to $12.5 billion. Goldman Sachs judged IBM on “track to complete its pivot to long-term growth,” while JPMorgan tagged it a “relatively defensive name.”
ServiceNow sits on the other side of the trade. The company cleared earnings estimates yet still lost 10% — leaving it down 25% since 2026 began and roughly 50% over the past year. The fear priced in here is AI eating away at subscription software itself. CEO Bill McDermott’s counter: enterprise AI is set to be the biggest payoff of the multitrillion-dollar AI infrastructure investment supercycle. Whether the market buys it remains to be seen.
Same Money, Opposite Calls: Three Variables the Market Priced
Take the day apart and the market turns out to have repriced three distinct variables. One, how direct the return path runs: Meta’s outlay lands directly in ad monetization, whereas Microsoft’s money first turns into capacity that is resold to OpenAI — an added layer of exposure. Two, whether spend and revenue step together: just one point of Azure slowdown turns that 66% spending jump from a vow into a question mark. Three, counterparty credit risk: with nearly half of the commercial backlog riding on OpenAI, its ability to raise funding becomes an implicit input to Microsoft’s share price.
The pattern lines up with the case we made in our 2026 opening outlook for AI: the AI infrastructure narrative is rotating from “how much got built” to “how much comes back.”
What Builders Should Take from This
Three takeaways with practical teeth. First, don’t expect compute supply to ease anytime soon — Microsoft concedes GPU deliveries trail demand, so an architecture that leans hard on one cloud should already have a fallback in place. Second, funding costs will leak into pricing: once markets require AI spending to show real returns, cloud and model vendors will bake those costs into API and lease prices at a faster clip; anyone doing cost modeling should set money aside for that factor. Third, AI ROI has moved out of pitch-deck language into valuation math — and the same test now applies in-house, where AI spending with no clear payback story goes first when budgets next come under review.