After Blowout Q1 Earnings, Hyperscalers Raise AI Capex Guidance to $725 Billion for 2026
Following record Q1 2026 earnings, Alphabet, Microsoft, Meta, and Amazon have collectively raised their annual capital expenditure guidance to approximately $725 billion for 2026 — up 77% from last year — with analysts projecting the combined AI infrastructure bill could breach $1 trillion by 2027. Alphabet's stock had its best April since 2004, while Meta faces investor skepticism despite raising its own capex ceiling by $10 billion.
The Q1 2026 earnings season is over, and the verdict is unambiguous: the four hyperscalers that have bet most aggressively on AI infrastructure are now spending more than ever before — and they just told investors they plan to spend even more.
When analysts tallied the updated capital expenditure guidance provided by Alphabet, Microsoft, Meta, and Amazon across their April 29–30 earnings calls, the combined 2026 figure came in at approximately $725 billion, up 77% from the roughly $410 billion the same four companies spent in 2025. Several analysts, including those at Evercore ISI and Bank of America, placed their revised 2026 estimates even higher, in the $800–$900 billion range, when accounting for announced but not yet fully disclosed data center contracts.
More striking still: executives at multiple companies signaled that 2027 spending will be materially higher than 2026. CNBC reported that some Wall Street analysts now project the combined AI infrastructure bill could exceed $1 trillion in 2027 — a figure that would have seemed fantastical just three years ago.
Company by Company: The New Capex Landscape
Alphabet revised its 2026 capital expenditure guidance to $180–190 billion, up from a prior range of $175–185 billion. The increase was announced alongside the strongest quarterly performance in Google Cloud’s history: $20 billion in revenue, up 63% year over year, with a backlog that has nearly doubled to $462 billion in a single quarter. CFO Anat Ashkenazi said 2027 capex would “significantly increase” relative to 2026, without providing a specific figure. Alphabet’s stock climbed as much as 6% in after-hours trading and closed April up 34% — its best monthly performance since 2004.
Microsoft set its 2026 capex guidance at $190 billion, with management noting that roughly $25 billion of that figure reflects elevated component costs — particularly memory — rather than purely incremental capacity additions. Azure’s AI business now runs at a $37 billion annualized revenue rate, up 123% year over year, giving the company the strongest visible return on AI investment of any of the four. Despite the strong cloud numbers, Microsoft shares fell about 3% after the report, with investors fixating on sequential growth rates that some felt were moderating too early.
Amazon held its previously announced capex budget at $200 billion for the year — more than any other hyperscaler — while reporting AWS revenue of more than $37 billion for Q1, up 28% year over year. CEO Andy Jassy used the earnings call to characterize the current moment as a “once-in-a-generation” infrastructure buildout, and Amazon’s budget figure reflects that conviction. Amazon shares also fell roughly 3% post-earnings, with investors concerned about whether the massive capex will convert to revenue growth fast enough.
Meta raised its 2026 capex range to $125–145 billion, up from its prior range of $115–135 billion, explicitly citing higher component pricing and additional data center costs. CEO Mark Zuckerberg remained bullish on long-term AI returns, but the stock fell more than 5% after the company said it expected revenue growth in Q2 to remain roughly flat compared to Q1 — a softer outlook that investors took as a sign that Meta’s AI advertising improvements may be approaching a near-term ceiling.
The Question of Returns
The central tension at this earnings season was not capability — all four companies delivered record revenues — but the credibility of returns on AI investment. Google provided the clearest answer: its Cloud segment, supercharged by AI, grew faster than AWS and Azure in Q1, and the company’s AI Overviews product in Search has demonstrably increased user engagement metrics. For Google, the capex story is relatively straightforward: the AI investment is directly traceable to cloud and search revenue acceleration.
Microsoft’s case is almost as clean. Azure’s $37 billion AI annualized run rate, growing 123% year over year, provides a direct revenue line item against which investors can measure the $190 billion capex commitment. The company’s GitHub Copilot, Microsoft 365 Copilot, and Azure OpenAI Service together represent the most mature enterprise AI monetization stack in the industry.
Amazon’s return story is harder to tell in the near term. AWS’s 28% revenue growth is strong by historical standards but softer than Google Cloud’s 63% and Azure’s implied AI-driven gains. The company is betting heavily on its custom silicon story — Trainium and Inferentia chips designed to run AI workloads at lower cost than Nvidia GPUs — but those chips have yet to drive a visible revenue premium.
Meta’s challenge is the most complex. Its AI investments are primarily improving ad targeting and content recommendations — improvements that flow into revenue but are not separately measurable as “AI revenue.” The company’s long-term bet on AI-native hardware (Ray-Ban smart glasses, Orion AR prototypes) and frontier AI research (its Superintelligence Labs) remains difficult to value on a quarterly timeframe.
The Nvidia Effect and the Risk of Overcapacity
Underlying all four spending plans is a common assumption: Nvidia’s GPU supply will remain the key constraint on AI deployment, and securing capacity early is a competitive necessity. That logic has been validated quarter after quarter — but it also carries embedded risk.
If AI scaling laws plateau sooner than expected, or if the next generation of custom silicon (Google’s TPU 8, Amazon’s Trainium 3, Microsoft’s Maia 2) dramatically reduces per-token inference costs, the economics of the current buildout could shift dramatically. Several analysts noted that the hyperscalers are now ordering hardware based on projected demand two to three years out, meaning any miscalculation compounds.
The bear case, articulated by some skeptical investors during the earnings calls, is that the four companies are collectively building a $725 billion field of AI capability whose monetization depends on enterprise adoption curves and consumer AI behaviors that remain genuinely uncertain. The bull case — the one embedded in Alphabet’s 34% April stock rally — is that we are in the early innings of the largest infrastructure buildout in computing history, and the companies that under-invest today will lose the next decade.
What Comes Next
With Q1 earnings behind them, the hyperscalers now face the six-month question: can Q2 and Q3 revenue growth keep pace with the accelerating capex? The market gave its preliminary verdict in after-hours trading: Google deserved a premium; Microsoft and Amazon needed to prove their return story; Meta needed to stop talking about 2028 and deliver something in 2026.
Beneath the quarterly drama, the structural reality is that the AI infrastructure spending wave has now moved beyond a point of easy reversal. Data centers under construction, semiconductor supply agreements, and power procurement deals signed in 2025 and early 2026 lock in substantial spending regardless of quarterly sentiment. The $725 billion figure for 2026 is less a decision than a trajectory — and $1 trillion in 2027 is looking less like an outlier forecast and more like a floor.