Morgan Stanley projects that capital expenditure by hyperscale cloud providers in support of artificial intelligence will mark a distinct new phase - with capital-intensity metrics set to exceed levels seen during the dot-com boom.
Analyst Todd Castagno, in a note to investors, observed a rapid acceleration in spending plans that has not been matched by an equivalent rise in revenue expectations. Castagno quantified the shift by pointing to hyperscalers cash capex-to-sales ratios. Where the dot-com era registered roughly 32%, Morgan Stanley baseline forecasts show those ratios reaching 34%, 39%, and 37% in 2026, 2027 and 2028 respectively.
The bank also highlighted lease financing as a factor that further elevates headline capex measures. When lease-adjusted figures are included, Morgan Stanley estimates the capex-to-sales ratios could climb to 38%, 44%, and 45% across 2026-28.
Morgan Stanley expects hyperscalers to account for approximately 40% of total Russell 1000 cash capex in 2026-28, a share that represents more than $2 trillion in aggregate spending. Extending beyond hyperscalers, the firm said broader AI-related investment may exceed half of all Russell 1000 capital expenditures, underscoring what it described as the growing concentration of index-level investment within AI.
The bank characterized recent capex revisions as "unprecedented." Its estimates for 2026-27 spending have been revised higher by $630 billion compared with figures from six months earlier. By contrast, revenue revisions have lagged and free cash flow estimates have trended lower, a pattern Morgan Stanley attributes to the multiyear horizon required to monetise AI infrastructure.
Castagno cautioned that the magnitude of the spending shift has created material challenges for financial modelling. He said the revisions for 2026-28 reflect step-changes rather than incremental increases, complicating forecasts and scenario analysis.
As fixed-cost bases rise at companies making these investments, the bank warned that future earnings and free cash flow will become more sensitive to changes in revenue expectations. Semiconductor suppliers were identified as the clearest beneficiaries of the wave of investment, with Morgan Stanley citing sales revisions for that supply chain up by roughly 60% for 2026.
What this means
- Hyperscalers are preparing to deploy capital at a scale that pushes capex-to-sales metrics above a historical peak - increasing the financial commitment to AI infrastructure.
- Lease accounting and financing amplify reported capex intensity, and a large portion of Russell 1000 capex is being concentrated in AI-related spending.
- Semiconductor suppliers stand to gain from elevated demand, as reflected in significant positive sales revisions for 2026.
Analytical note
The bank's findings point to a reallocation of corporate investment toward AI infrastructure that is both large in absolute terms and concentrated within major public-market capital budgets. At the same time, Morgan Stanley flagged the uneven timing between spending and revenue realisation, which is already affecting free cash flow projections and will increase earnings volatility if revenue outcomes diverge from current expectations.