Morgan Stanley projects $1.1T capex

The story here is AI infrastructure — the physical buildout underneath all the models, copilots, and cloud services people actually use. The stakes are huge because this is no longer a normal tech spending cycle. It is turning into a capital race measured in hundreds of billions, with a handful of companies deciding how fast the whole stack gets built. What changed over the past few days is that Morgan Stanley, after the latest earnings from the biggest cloud platforms, pushed its forecast for hyperscaler capex to $805 billion in 2026 and $1.1 trillion in 2027. (newsdefused.com) ### Who is actually spending this money? The core group is Amazon, Alphabet, Microsoft, Meta, and Oracle — the companies Morgan Stanley bundles as the main hyperscalers in this call. These are the firms building giant data centers, buying accelerators and networking gear, and expanding clou(newsdefused.com)raises budgets, it changes demand for chips, power, financing, and construction all at once. (newsdefused.com) ### Why did Morgan Stanley raise the number now? Because the latest earnings gave Wall Street fresh evidence that the spending wave is not slowing. Analysts broadly moved 2026 estimates into the $800 billion to $900 billion range and 2027 above $1 trillion after the April 29, 2026 earnings (newsdefused.com)or $765 billion, and about $1.1 trillion for 2027 versus roughly $951 billion before. (cnbc.com) ### What in the earnings made that believable? Revenue finally started to catch up enough to support the story. Alphabet said Google Cloud revenue grew 63% year over year and raised 2026 capex to as much as $190 billion, while Amazon reported AWS revenue growth of 28% and kept talking up long-term infrast(cnbc.com)e. (cnbc.com) ### How big is the spending right now? Already enormous. In Q1 2026 alone, Amazon, Alphabet, and Microsoft spent a combined $112 billion on capex — $45.17 billion, $35.67 billion, and $30.88 billion respectively. That means the industry is not talking about a future ramp. The ramp is already here, quarter by quarter, and Meta plus Oracle push the aggregate even higher. (officechai.com) ### So why are investors still uneasy? Because spending is rising faster than revenue forecasts in some cases, and free cash flow gets squeezed when companies build this aggressively. Morgan Stanley warned earlier this year that hyperscaler capex intensity could exceed dot-com-era peaks, with cash capex-to-sales ratio(officechai.com)ow up immediately. (finance.yahoo.com) ### Who benefits first? Chipmakers, networking vendors, data-center builders, utilities, and the banks financing all of this. Morgan Stanley’s own March note framed the AI capex boom as a tailwind for banks because huge infrastructure programs create financing demand and deal flow. In plain English — when hyperscalers spend li(finance.yahoo.com)ower. (morganstanley.com) ### What is the real risk? Concentration. A tiny number of companies are now responsible for an outsized share of total market capex, and Morgan Stanley has already flagged that hyperscalers could drive about 40% of Russell 1000 cash capex over 2026 to 2028. That makes the AI boom feel powerful, but also brittle — if demand, regulation, or financing conditions wobble, the whole supply chain feels it fast. (finance.yahoo.com) ### Bottom line This is not just a bigger forecast. It is a sign that AI has moved from software excitement to industrial-scale buildout. Morgan Stanley’s $1.1 trillion 2027 number tells you the market now expects the biggest tech platforms to keep spending at a pace that would have looked absurd even a year ago — and to keep doing it because, so far, demand is giving them permission. (newsdefused.com)