Hyperscaler Capex H1 2026: Variance vs Guidance and the Widening ROI Gap

When evaluating hyperscaler capex H1 2026, Amazon spent more on infrastructure in Q1 2026 than it did in the entire first half of 2025 — and that single data point captures the scale of the hyperscaler capex story entering mid-year. This is the hyperscaler capex H1 2026 story in one number.

But it does not answer the question that now defines the investment thesis: is the spend generating proportionate returns? With Q2 2026 earnings for all four major hyperscalers (Amazon, Microsoft, Alphabet, and Meta) expected in the 22–28 July window, this is the analytical framework for reading what those reports will reveal. IBM’s push into streaming infrastructure via its Confluent acquisition reflects the same capex pressure.

Hyperscaler Capex H1 2026: What the Q1 Actuals Show

In the context of hyperscaler capex H1 2026, Amazon reported Q1 2026 capex of $44.2 billion — a 77% year-on-year increase from $25 billion in Q1 2025. Amazon Web Services (AWS) revenue grew 28% year-on-year to $37.6 billion, with an operating margin of 37.7%. Amazon’s AI services run rate exceeded $15 billion in Q1, growing at triple digits year-on-year.

Full-year 2026 capex guidance stands at $200 billion, up from $125 billion in 2025. Read against revenue growth, this hyperscaler capex figure is the number investors are now scrutinising most closely. That pace makes hyperscaler capex H1 2026 the largest single variable in the cycle.

Adding to total hyperscaler capex H1 2026, Alphabet’s Q1 2026 capex of $35.7 billion represented a 107% year-on-year increase from $17.2 billion in Q1 2025 — the sharpest acceleration of the four. Google Cloud revenue grew 63% year-on-year to over $20 billion, and the company raised its full-year 2026 capex guidance to $180–190 billion. Google Cloud’s remaining performance obligations (RPO) — contracted future revenue — topped $460 billion, the clearest signal that demand is outpacing supply.

Microsoft’s fiscal third quarter of FY2026 (calendar Q1 2026) capex reached $34.9 billion, driving hyperscaler capex H1 2026 higher. Azure grew 40% year-on-year; Microsoft’s AI revenue run rate surpassed $37 billion annually. Commercial cloud RPO hit $392 billion, up 51% year-on-year. Full-year 2026 capex is expected at approximately $190 billion, up 61% from 2025. Notably, Microsoft has $80 billion in unfulfilled Azure orders constrained not by demand but by power availability — a supply-side bottleneck, not a monetisation problem.

Rounding out the initial hyperscaler capex H1 2026 data, Meta’s Q1 2026 capex of $19.84 billion came in significantly below the analyst consensus estimate of $27.57 billion. The company subsequently raised its full-year 2026 capex guidance to $125–145 billion (from $115–135 billion), signalling heavy back-half loading.

Meta’s total Q1 revenue grew 33% year-on-year to $56.31 billion — but that revenue is advertising, not cloud services. There is no direct infrastructure-revenue line to measure capex efficiency against in the same way as AWS, Azure, or Google Cloud.

Variance vs January Guidance: Uniformly Upward

Projections for hyperscaler capex H1 2026 shifted when the four companies delivered Q4 2025 earnings in late January and early February 2026, per-company full-year 2026 capex guidance was as follows. By Q1 2026 results on 29 April, each had revised upward.

Combined 2026 guidance is now tracking toward $670–700 billion — a $60–90 billion upward revision across the group in under 90 days. The gap between hyperscaler capex and measurable returns is what the July reports will test. Guidance was raised across the board, and hyperscaler capex is now the core of the variance story.

CompanyJanuary 2026 GuidanceApril 2026 (Q1 Actuals) GuidanceVariance
Amazon$200B$200B (held)—
Alphabet$175–185B$180–190B+$5B midpoint
Microsoft~$118B annualised~$190B full-year trajectory+~$72B
Meta$115–135B$125–145B+$10B midpoint
Combined~$610–640B~$670–700B+$60–90B

Spending Variance 2026

$44.2B Amazon Q1 Capex Run-Rate (+77% YoY)

Surging hyperscaler capex H1 2026 outpaces cloud revenue growth rates, widening the structural capital return deficit across tier-one infrastructure providers.

The massive reality of hyperscaler capex H1 2026 challenges the “digestion thesis” — the expectation that hyperscalers would eventually pause to allow software monetisation to catch up with hardware depreciation — has not materialised.

The variance from January guidance is uniformly positive (more spending, not less), driven by three factors: higher component pricing, particularly for graphics processing units (GPUs) and high-bandwidth memory (HBM); accelerating backlog and contracted demand signals; and competitive dynamics that make a unilateral pause strategically untenable for any single player. Whether the digestion thesis holds depends on how quickly hyperscaler capex growth converges with revenue growth.

Examining hyperscaler capex H1 2026, analysts at CreditSights and Evercore ISI have flagged that aggregate free cash flows for the group will decline 28–100% year-on-year as capex absorbs operating cash flow. Amazon’s trailing 12-month free cash flow has already collapsed 95% to $1.2 billion, as disclosed in Amazon’s Q1 2026 earnings press release. The capital is being deployed; the question is the return trajectory.

The ROI Divergence Taking Shape

Understanding hyperscaler capex H1 2026 requires recognizing that the most analytically useful frame for Q2 is not which companies are spending more, but which are generating improving revenue leverage per dollar of capex.

For a broader look at how hyperscalers are financing this buildout — debt issuance, revolving credit facilities, and balance sheet strategy — see our earlier analysis: Hyperscaler Capex Debt Financing and the AI Infrastructure Bet (May 15, 2026). The answer decides whether hyperscaler capex is still a growth engine or becoming a cash drain.

Within hyperscaler capex H1 2026 returns, AWS and Google Cloud are the clearest cases for improving efficiency. AWS generated $37.6 billion in quarterly revenue against $44.2 billion in quarterly capex — a ratio of approximately 0.85x (NextWave Insight calculation from Q1 2026 capex and segment revenue disclosures) that will improve materially if AWS revenue continues compounding at 28% while capex growth moderates from its Q1 peak.

Google Cloud’s position is structurally stronger: the $35.7 billion Q1 capex figure is company-wide (covering Search, YouTube, and all other infrastructure), while the $20 billion Cloud quarterly revenue figure is Cloud-only. The capex-to-cloud-revenue ratio is already favourable and improving with each quarter of 63% revenue growth.

Azure’s position is slightly more complex. Microsoft’s $37 billion AI annual run rate against $34.9 billion in quarterly capex implies the annualised revenue-to-capex ratio is approaching parity (NextWave Insight calculation). The $80 billion in demand-constrained backlog is a leading indicator — once power constraints ease, that revenue should accelerate without proportionate new capex.

Across hyperscaler capex H1 2026 commitments, Meta occupies a structurally different position. Its capex is financing AI capability underlying its advertising products, not a cloud platform sold to enterprise customers.

Better AI-driven ad targeting raises revenue per user and advertiser pricing power — a legitimate business case, but not comparable to AWS or Google Cloud on a revenue-per-dollar basis. Investors underwriting Meta’s $125–145 billion full-year commitment are betting on ad-AI monetisation without the cloud-revenue benchmark that makes the other three measurable. This concentration of capital in a single monetisation model has broader implications for the sector — explored in our analysis of AI funding concentration and the resulting startup runway compression.

Microsoft’s Fiscal Year Alignment

One labelling note: Microsoft operates on a fiscal year ending 30 June. “Fiscal Q3 FY2026” covers the January–March 2026 calendar period — the same window as Q1 2026 for the other three. The calendar alignment holds for cross-company comparison; the fiscal labelling does not.

What to Watch in Q2

Hyperscaler capex is now seen topping $1 trillion by 2027, with surging hyperscaler capex H1 2026 showing no signs of slowdown. Three signals will determine whether the ROI narrative shifts or hardens when Q2 results land in late July.

First, AWS revenue growth rate versus capex growth rate. If AWS revenue accelerates above 28% while Q2 capex is flat or declining versus Q1’s $44.2 billion, the efficiency curve is bending in the right direction. If capex remains elevated without a corresponding revenue step-up, the “digestion” question resurfaces with more force. Tracking AWS revenue against hyperscaler capex in Q2 gives the clearest read on efficiency.

Second, Google Cloud RPO drawdown. The $460 billion backlog is the largest optionality signal in the group. Watch how quickly contracted revenue converts to recognised revenue — fast conversion validates that the supply build is meeting real demand; slow conversion suggests the backlog includes deals too early-stage to execute.

Third, Meta’s H2 capex phasing. Q1 came in well below consensus, with heavy back-half loading signalled. If Q2 capex surges toward the $35–40 billion range implied by full-year guidance, that sharpens the question of what return each incremental data centre dollar is generating for an ad business already growing at 33%.

How capex-to-revenue ratios move across these four companies in Q2 will define the final assessment of hyperscaler capex H1 2026 and whether mid-2026 is remembered as the quarter the AI infrastructure build started paying off — or the quarter the divergence between cloud-revenue hyperscalers and Meta’s ad-infrastructure model became impossible to ignore. Comparing hyperscaler capex intensity on the same quarterly basis is what separates signal from noise.

This article was produced with AI assistance and reviewed by the editorial team.

Sarah Chen

Sarah Chen

Sarah Chen covers the economics of the technology industry for Next Waves Insight — venture funding, enterprise AI adoption, and the business models emerging around frontier technology. Her work traces capital flows and corporate strategy to explain not just what companies are building, but why, and what it signals about where the industry is heading. She has a background in technology and financial journalism.

2 thoughts on “Hyperscaler Capex H1 2026: Variance vs Guidance and the Widening ROI Gap”

Leave a Comment