The projections are now on the income statement. Q1 2026 earnings season has converted the hyperscaler capex story from analyst speculation into reported fact, with three of the five largest AI infrastructure investors disclosing quarterly capital expenditure figures that confirm the scale of the infrastructure build-out.
According to Q1 2026 earnings reporting, Alphabet reported capital expenditures of approximately $35.67 billion in the first quarter, roughly double the year-earlier figure. Amazon reported the largest quarterly capex figure among the group at approximately $44.2 billion. Microsoft disclosed approximately $30.88 billion in capital expenditures in its fiscal third quarter, representing an approximately 84% increase year-over-year, according to the same earnings reporting. Meta, for its part, raised its full-year 2026 capex guidance to $125 billion to $145 billion, according to Q1 earnings disclosures. Combined 2026 capital expenditure plans across the group have reportedly reached approximately $700 billion, based on Q1 earnings disclosures and full-year guidance.
One figure from the Amazon disclosure warrants a separate read. AWS’s custom chip business, built around the Trainium and Inferentia lines, reportedly reached an annualized $20 billion revenue run rate. That’s a run-rate projection, not a separately disclosed revenue figure in formal filings, and should be read accordingly. It is, however, the clearest signal yet that the chip investment is generating a revenue line, not just reducing third-party vendor spend.
Google Cloud added a different dimension. Its reported contract backlog reached approximately $460 billion, according to Q1 earnings disclosures, roughly double the year-earlier figure. Backlog is a forward commitment metric, not recognized revenue, but a $460 billion pipeline is a constraint on how much of the capex story is speculative. Somebody has already signed.
Why does this matter to investors and infrastructure strategists? The Q1 earnings cycle closes a specific evidentiary gap. The $700 billion figure had circulated in analyst notes and earnings previews for weeks. What earnings season provides is the per-company reported granularity that lets analysts assess whether the collective commitment is being funded by revenue growth, debt, or workforce reallocation. The answer, based on Q1, appears to be all three simultaneously.
This is the fourth consecutive quarter in which hyperscaler capex has accelerated on a year-over-year basis. The prior cycle’s question was whether the infrastructure commitment was real. Q1 answers that. The emerging question is whether it’s generating returns at a pace that justifies the scale. Google Cloud’s backlog figure is one early data point. AWS’s chip run rate is another. Neither is a definitive answer, but both suggest the revenue conversion story has begun.
Watch Q2 guidance revisions. If any of the five companies pulls back on full-year capex projections, that will be the most significant market signal in the AI infrastructure story since the commitments were first announced. Conversely, further upward revisions would confirm that the build-out phase is extending, not peaking.
The infrastructure bet has moved from intention to record. What Q2 earnings will reveal is whether the returns curve is bending in the same direction.