A new Morgan Stanley projection estimates that global AI compute investment by hyperscalers — the largest cloud infrastructure operators — and neo-cloud providers will reach approximately $1.58 trillion in 2027, a sum the bank expects to account for roughly 5% of S&P 500 sales that year and about 25% of the index's sales growth. The S&P 500 is the benchmark index of large US-listed companies, which is what connects the spending plans of a handful of cloud firms to the growth profile of the broader market.

A defining feature of the spending wave is how it gets paid for: half of the projected investment is expected to be financed by debt.

The Numbers Behind the Buildout

The 2027 forecast marks a steep climb from around $980 billion in 2026, and Morgan Stanley expects the total to keep rising, reaching $1.81 trillion in 2028.

The headline figure also captures more than cash. It folds in lease commitments, which the bank estimates will contribute roughly $195 billion in lease capex in 2027, rising to $246 billion by 2028. Including leases pushes the projection 14% above earlier estimates that tracked only cash outlays.

Oracle leads the lease pack at $57.2 billion, followed by Microsoft, Meta, Amazon, and Google. Across the group, hyperscalers currently hold over $1.1 trillion in pending lease commitments.

Who Collects the Checks

If AI spending is a gold rush, NVIDIA is still the one selling the shovels. The report forecasts the chipmaker capturing around two-thirds of AI-related inflows, amounting to nearly $690 billion.

Morgan Stanley estimates that broader "AI enablers" and "adopters" will together drive around 78% of S&P 500 earnings growth in 2027.

Goldman Sachs has reached a similar conclusion looking backward. The bank found that AI-related capital expenditure has driven nearly half of the recent earnings growth for the S&P 500.

The Accounting Wrinkle

Data-center leases function much like debt-financed asset purchases, but they are not recorded as capex. That means traditional cash-flow statements may understate the real scale of the buildout. The commitments are not invisible, though: lease obligations and their payment schedules are disclosed in standard financial filings.

The research notes that high-rated hyperscalers still retain substantial borrowing capacity. They are, however, facing free cash flow pressure because so much of the spending is front-loaded.

What This Means for Investors

When around 78% of projected 2027 earnings growth leans on AI enablers and adopters, the index's fortunes become closely tied to one investment cycle, and when a quarter of sales growth flows from a single theme, diversification looks different.

The research anticipates capex growth rates decelerating after 2027, even as total spending keeps climbing toward $1.81 trillion in 2028.

With over $1.1 trillion in pending lease commitments and half the projected investment expected to be debt-financed, investors may want to look past headline capex and check lease disclosures and borrowing trends. The hyperscalers' strong credit ratings provide a cushion for now, but front-loaded spending paired with free cash flow pressure is a combination worth monitoring. Quarterly earnings updates and annual filings are where those figures surface in detail, offering a recurring checkpoint on whether the debt-financed buildout is tracking the forecast.