Morgan Stanley just revised hyperscaler capex estimates upward — again. The five largest cloud companies will spend $805 billion this year and $1.1 trillion in 2027. That’s more than the GDP of Saudi Arabia.
Morgan Stanley’s latest revision (post Q1 2026 earnings) raised estimates across the board:
2026: $805B (up from $765B prior estimate)
2027: $1.116T (up from $951B prior — a 17% revision)
The five hyperscalers — Amazon, Google, Meta, Microsoft, and Oracle — are now expected to collectively spend $1.1 trillion in a single year on AI infrastructure.
Capex Growth by Company (2024-2027 CAGR)
Oracle116%
Microsoft69%
Google59%
Meta54%
Amazon48%
Oracle Is the Surprise
116% CAGR. Oracle — the company most people associate with enterprise databases — is growing AI infrastructure spend faster than any hyperscaler. Faster than Microsoft. Faster than Google. Faster than Meta.
Why? Oracle Cloud Infrastructure (OCI) has quietly become the backend for some of the largest AI training runs. They’re not competing on the model layer — they’re competing on the infrastructure layer, offering cheaper GPU clusters than AWS and Azure. Classic Harness Theory: Oracle isn’t building AI. It’s building the picks and shovels.
A bubble is when capital chases returns that don’t exist. This is different. Cloud revenue is growing. API demand is growing. Agentic workloads are 1,000x more compute-intensive than chat. The capex is following real demand — it’s just demand that hasn’t fully monetized yet.
2. THE INFRASTRUCTURE LAYER IS EATING EVERYTHING
$1.1 trillion in one year is more than the GDP of the Netherlands, Saudi Arabia, or Switzerland. The AI infrastructure layer is now larger than most national economies. This is the Map of AI’s bottom layer — and it’s expanding faster than every layer above it.
3. THE TOKENMINIMIZING PARADOX
The same companies cutting internal token budgets are increasing infrastructure capex by 80% YoY. This isn’t contradictory — internal AI is a cost center. External AI infrastructure is a revenue engine. The capex tells you which one they’re betting on.
Business Engineer Framework
The Map of AI — Infrastructure Layer
The Map of AI tracks 9 layers and 200+ companies. The infrastructure layer — compute, data centers, chips — is the foundation everything else runs on. $1.1 trillion in annual capex means this layer is now the single largest capital allocation in technology history.
Every quarter, the capex estimates go up. Every quarter, analysts revise higher. At some point, someone will have to explain to shareholders what $1.1 trillion bought them.
But right now, the hyperscalers are making a bet: that the demand for AI compute will grow faster than the supply they’re building. If they’re right, this is the most consequential infrastructure buildout since the internet. If they’re wrong, it’s the most expensive.
Either way, the Map of AI’s infrastructure layer just became the most important chart in technology.
Source: Morgan Stanley Research, Altimeter Capital (April 2026)
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
