Cloud Giants Deploy $600B Capex Blitz as AI Demand Overwhelms Infrastructure
TL;DR: Amazon, Google, and Microsoft are collectively spending nearly $600 billion on infrastructure in 2026, with demand still outpacing supply. Hyperscalers admit they cannot build capacity fast enough to meet AI workload requirements.
The Capex Surge: When $200B+ Still Isn’t Enough
Amazon Web Services is escalating capex from $200 billion to $220 billion for 2026, and CEO Andy Jassy flatly stated the company will still face capacity shortfalls through 2027. AWS generated $42.2 billion in Q2 revenue—a 36.7% year-over-year increase and the fifth consecutive quarter of accelerating growth. At current run rates, AWS alone constitutes a $169 billion annualized business, which would rank 24th on the Fortune 500 as a standalone company.
Google is matching this intensity, raising capex guidance to $195-205 billion from $180-190 billion, with CFO Anat Ashkenazi explicitly acknowledging that “demand still outpaces that investment.” Cloud revenue surged 82% to $24.8 billion, with AI infrastructure and solutions emerging as primary drivers.
Operational Reality: Supply Constraints Define the Market
Both hyperscalers are encountering a supply-side constraint unseen since early cloud adoption—except compressed into months rather than years. Memory costs alone forced Amazon to revise estimates upward, while Google faces pressure from accelerating chip delivery demands and energy infrastructure scaling.
This differs fundamentally from first-era cloud buildout, where demand accumulated gradually. AI workloads compressed the adoption curve, creating bottlenecks across GPU procurement, power delivery, and datacenter construction.
Margin Dynamics and Long-Term Returns
Jassy drew parallels to early cloud economics, noting that “margins and returns in AI [are] tracking what we saw with core at the same point of evolution.” This suggests hyperscalers expect AI capex intensity to normalize—but only after establishing dominant market positions.
Google’s depreciation expense and operational costs will increase proportionally, with additional hiring across AI, cloud, and marketing offsetting efficiency gains from scale.
Background: The $600B Question
Amazon Web Services dominates cloud infrastructure with 32% global market share. The division’s Q2 acceleration reflects enterprise AI adoption across training, inference, and managed services. AWS capex elevation signals confidence in sustained AI demand despite recession warnings.
Google Cloud Platform trails AWS and Microsoft Azure but is gaining share through aggressive infrastructure investment and bundled AI solutions. The 82% cloud revenue growth indicates successful positioning in competitive midmarket.
Microsoft Azure, while not quoted in this report, is similarly investing $120+ billion across its own capex requirements. The three hyperscalers collectively control 65% of cloud infrastructure spend.
The AI Inflection arrived in 2023-2024 with large language models entering production. Enterprise demand for inference capacity—not just training—created unexpected density requirements. Chip shortages, power constraints, and cooling infrastructure became operational constraints rather than procurement challenges.
Investment Implications
- Vendor concentration deepens: Only three companies can sustain $200B+ capex. Competitors cannot match this burn rate.
- Margin compression near-term: Depreciation and operational costs will suppress EBITDA growth in 2026-2027.
- Moat strengthening: First-mover advantages in power, land, and chip allocation compound over 18-month cycles.
- Utilization risk: If demand normalizes, excess capacity becomes stranded capex. Both companies are betting heavily on sustained enterprise AI adoption.
What to Watch
Monitor utilization rates and average revenue per compute unit (ARPU) trends. If capex exceeds demand growth in Q4 2026, guidance will become more cautious. Watch for power infrastructure bottlenecks becoming the binding constraint—generators and transmission upgrades move slower than server deployment.
Microsoft’s next earnings call will clarify if Azure faces identical constraints. If not, the company gains relative advantage in pricing power and customer acquisition.
