Microsoft and Alphabet report fiscal Q1 earnings tonight after the New York close. Meta and Amazon report Wednesday and Thursday. The four companies will spend roughly six hundred billion dollars on AI this year combined. That is the biggest single-year capital expenditure in corporate history.

The question Wall Street is asking is no longer whether the bets will pay off. The question is when. And after Tuesday's session, the market answered: not soon enough.

The Russell 2000 fell. The Nasdaq fell. Oracle fell seven percent at one point on intraday questions about its three hundred billion dollar OpenAI compute commitment. SoftBank fell ten percent in Tokyo. CoreWeave got hit. AMD got hit. Even TSMC, the universal AI infrastructure beneficiary, finished red. Roughly one hundred eighty billion dollars in market cap evaporated from the top decile of AI-exposed names in a single session. It was the biggest one-day repricing of the AI trade since DeepSeek R1 in January 2025.

What broke the run.

Three things landed inside twenty four hours. Wall Street Journal reported OpenAI missed its quarterly revenue target. Bloomberg confirmed the Microsoft-OpenAI deal had been amended to end exclusivity, a structural change that made every previous capex assumption suspect. And Reuters published a long preview of the megacap earnings cycle that asked, in print, whether the four companies plowing six hundred billion dollars into AI infrastructure are seeing a corresponding lift in revenue, margins, or free cash flow. The answer, based on the consensus estimates floating into tonight, is mixed at best.

Microsoft's Q3 FY26 numbers.

Consensus expects roughly eighty one billion dollars in revenue, sixteen to seventeen percent year over year. Earnings per share around four dollars and six cents. Capex for the quarter around thirty five billion dollars, which is sixty four percent year over year growth. Polymarket has Microsoft at ninety four percent on a beat. Options markets are pricing a seven percent move in either direction.

What Wall Street actually wants is not the headline numbers. They are roughly known. What they want is three things.

First: Azure growth. Bank of America's Tal Liani wrote that Azure growth is gated by capacity, not demand. If Nadella confirms that thesis on the call, the entire FY27 capex envelope gets revised up again, and the market has to absorb a fresh increase to the spending line.

Second: the OpenAI exposure quantification. Microsoft now has a license to OpenAI's technology through 2032 but no longer pays revenue share on resold products and no longer has exclusivity on hosting. The accounting for that change is not in the public filings yet. Tonight Amy Hood walks through it for the first time.

Third: Copilot penetration. Consensus has Copilot at roughly three percent of Microsoft 365 commercial seats. Anything above four percent triggers a step change in how analysts model the AI revenue line. Anything below three suggests the most-hyped AI product of 2025 is still struggling to convert.

Alphabet's Q1 numbers.

Consensus expects roughly one hundred seven billion dollars in revenue, nineteen percent year over year. Earnings per share around two dollars sixty three. The story is Google Cloud. Analysts are looking for fifty percent growth and a substantial increase in remaining performance obligations, the future-revenue backlog that signals enterprise commitment. The Anthropic bake-off win, the forty billion dollar investment, and the TPU 8 launch all need to convert into RPO numbers tonight or the narrative cracks.

What the call won't say but the numbers will.

Combined Microsoft plus Alphabet capex this year will land in the range of two hundred eighty to three hundred billion dollars. Add Meta's one hundred fifteen to one hundred thirty five billion and Amazon's roughly two hundred billion, and the four-company total clears six hundred billion. For context, the largest capex line in corporate history before 2024 was ExxonMobil's roughly twenty four billion dollar 2014 number. Microsoft and Alphabet alone will outspend the entire S&P 500 oil and gas sector this year.

The question is whether AI revenue is growing fast enough to justify it. Microsoft AI is running at a roughly thirty billion dollar annual run rate. Anthropic's annualized revenue crossed thirty billion this quarter. OpenAI is north of fifteen billion but missed its target. Google is harder to disaggregate but Cloud's growth rate is the proxy. The aggregate AI revenue across the entire industry is roughly one hundred and fifty billion dollars annualized. The aggregate capex is six hundred billion. That is a four-to-one spend-to-revenue ratio.

Either AI revenue accelerates dramatically over the next eighteen months or capex gets cut. The market has been assuming the former. Tuesday was the first day it priced in the possibility of the latter.

What to watch on the call.

If Nadella raises the FY26 capex envelope above one hundred fifteen billion, the stock falls. If he holds it flat at one hundred ten billion, it might rally. If Hood says the words "capacity catching up to demand" without raising the spending line, the stock rallies hard.

If Pichai signals that Google Cloud's growth is self-funding, the stock rallies. If he signals that Anthropic compute commitments are accelerating beyond plan, the stock might still rally because investors want growth even at the cost of capex. The double-edge of the AI trade is that the market punishes you both for spending too much and for spending too little. There is no winning answer. There is only the answer that survives the next quarter.

Tonight is the night the AI capex thesis gets its first real stress test. The market that fell on Tuesday wants to know which side wins the framing war.

Sources: Reuters, Yahoo Finance, Investopedia, CNBC, Bloomberg, S&P Global, Seeking Alpha, TradingKey, CoinDCX, Blockonomi, Bank of America research, TipRanks, Polymarket.