Morgan Stanley told clients on Wednesday that investors may be underestimating how large Microsoft's future AI-derived revenues could be as the company expands its AI infrastructure.
The bank said a rapid rise in capital expenditure forecasts - driven by demand for generative AI - appears to be outstripping consensus revenue expectations tied to that capacity. Analyst Keith Weiss highlighted the mismatch between spending and monetization in the note.
GenAI demand well ahead of supply has yielded swiftly ramping Capex forecasts
According to Morgan Stanley, revenue models have not kept pace with the swift ramp in datacenter investment. The firm writes that its specific metrics - which it describes as "Monetization per Megawatt and Capex-Implied Azure forecasts" - indicate revenue estimates may be lagging and could be revised upward.
Weiss points to hyperscaler spending patterns and Microsoft's ongoing datacenter expansion as the basis for the firm's more optimistic view on potential revenue upside. The note says the company appears to be building AI datacenter capacity ahead of the immediate ability to monetize it.
Our analysis suggests Microsoft is deploying AI datacenter capacity well ahead of near-term monetization,
The bank provided a numerical framework for the argument, estimating Microsoft's installed datacenter footprint could expand from roughly 5GW in fiscal 2024 to nearly 20GW by fiscal 2028 - roughly a fourfold increase. Morgan Stanley also expects that revenue per megawatt will decline from about $20-30 million today to the high teens by FY28, but said that even with that decline the scale of additional capacity implies continued potential for upside to revenue estimates.
The analysts further emphasized that Microsoft's AI-capable infrastructure supports more than just Azure. Morgan Stanley underscored that the buildout underpins a unified cloud and AI platform that spans Microsoft 365, Dynamics 365, and LinkedIn, suggesting monetization opportunities across multiple product lines.
Weiss summarized the bank's view by noting that the company's existing AI infrastructure footprint could back materially higher long-term revenue expectations than those currently reflected in consensus estimates.
Context and implications
Morgan Stanley's note centers on a capacity-versus-monetization gap: heavy near-term investment in AI datacenter capacity, paired with revenue forecasts that the bank believes may not fully capture the revenue that capacity could generate over time. The firm quantifies both capacity growth and an expected decline in monetization per megawatt, and concludes the overall outcome still points to potential upward revision of revenue estimates.