Morgan Stanley Report Projects Hyperscaler AI Spending to Surpass $2 Trillion as Free Cash Flow Declines

Morgan Stanley has projected that the five largest hyperscale cloud providers will spend more than $2 trillion on capital expenditures over the next three years, pushing capital intensity beyond the peak of the dot-com bubble era. The forecast indicates that Amazon, Alphabet, Meta, Microsoft, and Oracle will see their combined capex-to-sales ratio reach 34% in 2026, 39% in 2027, and 37% in 2028, surpassing the roughly 32% peak recorded during the internet bubble .

The report, published on February 26, 2026, states that when financing leases are included, the capex-to-sales ratio climbs further to 38%, 44%, and 45% for the same period . These figures represent a structural shift in the technology sector’s business model, moving from asset-light operations to a heavy-capital cycle.

Morgan Stanley analyst Todd Castagno noted that spending is growing much faster than revenue. Over the past six months, consensus estimates for 2026 and 2027 capital expenditures were revised upward by more than $630 billion, while revenue expectations saw only limited adjustments .

Free Cash Flow Under Pressure

The rapid expansion in capital spending has not been matched by equivalent revenue growth. Morgan Stanley forecasts that Amazon’s free cash flow for 2026 will be negative $17 billion, while Bank of America projects a wider shortfall of $28 billion .

Alphabet’s capital expenditures are projected to reach $250 billion by 2027, placing further pressure on its cash generation . On a trailing four-quarter basis through the first quarter of 2026, Amazon, Alphabet, Microsoft, and Meta spent a combined $433.9 billion, with quarterly outlays growing 80% year over year .

Amazon reported first-quarter free cash flow of $1.2 billion, a decline of roughly 95% from the prior year, as quarterly capital expenditures reached $44.2 billion . Meta generated $12.4 billion in free cash flow during the same quarter against $19 billion in capital expenditures, and one model projects its free cash flow turning negative by 2027 .

Financing Leases Add to the Burden

The report emphasizes that traditional capital expenditure figures understate the true scale of investment. Financing leases, which are essentially debt-financed asset acquisitions, have become a major component of hyperscaler spending. During the dot-com era, financing leases were rarely used, but cloud providers are now committing to data center lease contracts totaling hundreds of billions of dollars .

Morgan Stanley’s software analysts estimate that financing lease capital expenditures from Microsoft and Oracle alone are sufficient to push the overall hyperscaler capex-to-sales ratio to the elevated levels projected for 2026 through 2028 . The combined future lease commitments of the five hyperscalers already exceed $660 billion, including approximately $248 billion from Oracle, $155 billion from Microsoft, and $104 billion from Meta .

Impact on the Broader Market

In 2025, hyperscalers accounted for more than 150% of the capital expenditure growth within the Russell 1000 Index, meaning the remaining constituents saw a net contraction in their investments. While hyperscaler capital expenditures grew approximately 70% year over year, spending by the rest of the index declined by 6% .

Morgan Stanley expects hyperscalers to represent roughly 40% of the Russell 1000’s total capital expenditure by 2026, doubling from 2024 levels, and potentially rising to 49% by 2028 . This concentration of spending reflects the scale of the AI infrastructure build-out relative to the broader corporate sector.

Semiconductor Suppliers Benefit

While hyperscalers face declining free cash flow, semiconductor companies enabling AI infrastructure have seen significant upward revisions to revenue expectations. Consensus estimates for semiconductor AI-enabling companies’ 2026 sales have risen approximately 60% over the past two years, far outpacing the roughly 8% increase for hyperscalers .

TrendForce reported on August 3, 2026, that combined capital expenditures for the world’s nine largest cloud service providers are expected to grow approximately 90% year over year in 2026, exceeding $886.7 billion. The five North American hyperscalers account for nearly 90% of that total .

TrendForce also raised its 2026 AI server shipment forecast from 28% to nearly 31% year-over-year growth, citing strengthened procurement demand for NVIDIA GB/VR rack-scale platforms and the ramp of next-generation ASIC platforms from Google and AWS in the second half of 2026 .

Depreciation and Debt Concerns

A depreciation wall is expected to impact earnings between 2027 and 2029 as the current wave of capital expenditures flows onto income statements. Servers depreciate over five to six years, while data center buildings are depreciated over twenty-five to forty years .

Trailing depreciation and amortization of approximately $149 billion sits against $433.9 billion of trailing capital expenditure, leaving a substantial gap that will be recognized in future periods . Meta’s infrastructure depreciation alone rose from $7.32 billion in 2023 to $13.36 billion in 2025, an 83% increase ahead of the largest spending waves .

To fund the build-out, Meta, Alphabet, and Amazon together raised well over $100 billion of debt across late 2025 and 2026. Sell-side desks project the sector may need $1.5 trillion of new debt over the coming years . Morgan Stanley estimates the sector’s off-balance-sheet obligations at close to $1.65 trillion .

Goldman Sachs estimates cumulative hyperscaler capital expenditures could reach as much as $7.6 trillion between 2025 and 2030, while Citigroup has raised its five-year forecast to $2.8 trillion .

nn

Image Via: Yahoo