Morgan Stanley: 2027 Global Cloud Capex Consensus of $1.2 Trillion Still Too Low
Taylor Wilson
Morgan Stanley estimates 2027 global cloud capex at $1.4 trillion — roughly 17% above market consensus. For at least ten straight quarters, the Street has underestimated how fast cloud operators actually spend.
The Street already raised its forecast — why does Morgan Stanley say it's still too low?
Post-Q2 earnings, the 2027 cloud-capex consensus rose to roughly $1.216 trillion, up $512 billion from early 2026 — a 72.8% jump.
Morgan Stanley's own cash-capex estimate sits at $1.4 trillion, about 17% above that consensus. This means → even after a massive upward revision, the market may still be undershooting.
Across 14 major cloud spenders, total capex forecasts have risen quarter-over-quarter for at least 10 consecutive quarters. In plain terms = the Street has been chasing actual spending for over two years and still hasn't caught up.
Where is the Big Four's money going, and why isn't it enough?
Google, Microsoft, and Amazon all said in Q2 results that demand exceeds available capacity. Meta echoed that industry-wide compute will stay tight for the foreseeable future.
Amazon said 2026 additions still won't meet demand, and much of its 2027 capacity is already pre-booked. Microsoft's Azure additions are absorbed almost immediately after coming online.
Cloud revenue growth is accelerating too: Azure grew 43% YoY, Google Cloud 82%, and AWS 37% — the fastest pace in 18 quarters. This means → the more they spend, the more they earn — the spending logic has not broken down.
What are the three most common market misreads?
Misread #1: Treating 29% growth as "a clear slowdown." That 29% comes on top of roughly 97% growth in 2026 — the absolute dollar increase is still enormous. Morgan Stanley itself forecasts 2027 growth above 50%.
Misread #2: Confusing tight supply with demand risk. All four hyperscalers say demand outstrips supply. The bottleneck is build-out speed, not a shortage of customers. In plain terms = it's not that nobody wants it — it's that they can't build fast enough.
Misread #3: Equating AI capex with GPU purchases. Spending now extends to custom chips (ASICs), networking, storage, data centers, and power. AI creates a multiplier effect across the entire infrastructure stack — far beyond graphics cards alone.
Where does the money flow? How should investors read the supply chain?
Morgan Stanley screened 50-plus tech companies and found an average of roughly 45% of revenue tied to cloud capex. This reflects how cloud spending has become the central thread of the entire tech supply chain, not just a few names.
High-exposure names include Wiwynn, Nvidia, Zhongji Innolight, Western Digital, Foxconn Industrial Internet, and Quanta, spanning servers, chips, optical modules, storage, and thermal management.
But Morgan Stanley cautions: high exposure means strong order elasticity — and greater sensitivity to customer concentration and spending cadence. This means → outperformance won't be decided by who books the most orders, but by who converts high-growth orders into sustained profitability.
Content is for reference only, not financial advice.