# Morgan Stanley Projects $1.58 Trillion in AI Computing Investment for 2027

By John Kim

Canonical URL: https://www.tokenpost.com/news/business/36755
Published: 2026-10-21T14:39:38.000Z
Updated: 2026-10-21T14:39:38.000Z
Section: Business

> The cash-flow gap is expected to be sharpest in 2027, while U.S. data centers could face a 32-gigawatt power shortfall by 2028.

AI computing investment by large and emerging cloud operators could reach $1.58 trillion globally in 2027, Morgan Stanley projects, as borrowing needs and electricity limits pressure the buildout.

That forecast is about 60% above the roughly $980 billion estimate for 2026. Spending could climb further, to about $1.8 trillion in 2028. The tightest mismatch between cash generated and investment is expected in 2027, so the projected cash-flow pinch does not mean total spending peaks that year.

U.S. companies issued about $255 billion in unsecured corporate bonds so far in 2026, equivalent to roughly 26% of projected AI infrastructure spending for the year. The comparison gives a sense of scale but does not establish that the bonds financed AI projects.

In a modeled scenario with more borrowing, combined interest coverage for Amazon, Alphabet, Meta and Microsoft falls from about 31 times to 12 times. For Oracle and SpaceX, the modeled figure declines from four times to 1.8 times. These are scenario estimates, not reported changes in the companies’ actual borrowing or interest costs.

Electricity is another potential limit. Morgan Stanley estimates U.S. data centers could face a 32-gigawatt power shortfall by 2028, against projected demand of 97 gigawatts. If chips and servers arrive before power connections and data-center construction are ready, some equipment could sit unused.

The investment also has a wider earnings footprint. Large cloud providers and emerging AI cloud operators account for about 5% of S&P 500 company revenue but could contribute roughly a quarter of revenue growth, the bank estimates. Its model links 91.7% of Nvidia’s projected 2027 revenue to AI infrastructure demand.

The bank also points to the 2022 NAND market, when shipments fell about 5% while prices dropped 20% to 30%, as an example of how demand changes can affect pricing as well as sales volume.

Amazon’s capital-spending plan rose from $200 billion to about $220 billion partly because memory prices increased. The higher estimate does not imply a comparable expansion in planned data-center construction. Whether revenue and cash flow from the buildout keep pace with its spending, financing costs and power requirements remains central to the forecast.