Morgan Stanley says big tech’s AI spending could still deliver 25% to 50% returns on invested capital

Artificial intelligence spending by large technology companies has become a magnet for investor anxiety, but Morgan Stanley’s latest bullish framing is that the expense spree may ultimately translate into measurable financial returns, not just promises of future growth.

The bank’s view, as reported by Yahoo Finance, centers on returns on invested capital, or ROIC, a profitability metric that compares operating profit to the amount of capital tied up in a business. In the note highlighted by the outlet, analysts project that major firms investing heavily in AI could achieve ROIC in a range of roughly 25% to 50% from their very large AI-related capital expenditures.

The specific companies referenced in the report are Amazon, Alphabet, and Microsoft, with the central argument being that their spending is not expected to be purely incremental cost. Instead, Morgan Stanley frames AI investment as capital deployment that can strengthen unit economics and earnings power over time, even if the buildout requires substantial upfront spending.

While the report points investors to the ROIC range as the key takeaway, it does not, in the cited Yahoo Finance item, provide a detailed breakdown of the assumptions behind that range. That includes what portion of spending is treated as capitalized versus expensed, how quickly new AI-driven revenue is assumed to ramp, or what margins the analysts model for the relevant products and services.

The note also appears aimed at addressing a common market question, namely whether the scale of AI infrastructure spending will compress returns by saturating demand, raising costs faster than pricing, or forcing companies into unfavorable capital structures. Morgan Stanley’s stance, as summarized by Yahoo Finance, is that the end result should still support strong returns on the capital invested.

The companies themselves are operating in a sector where AI spending has already become a defining competitive battleground. Cloud providers and platform companies are racing to build and supply AI compute, data tooling, and model-serving capabilities, while integrating them into customer-facing products. In such an environment, the timing of monetization matters, and so does the durability of margins once AI features move from experimentation to broader adoption.

Morgan Stanley’s specific ROIC framing matters to investors because it reframes AI spending from a spending headline into an expected outcome. If ROIC does land in the projected band, it would suggest companies can turn a capital-heavy AI cycle into efficient profitability rather than long-term value dilution.

That said, the Yahoo Finance item highlighted here does not disclose the detailed methodology or company-by-company mechanics, at least not in the text available through the posted market item. As a result, key uncertainties remain around the assumptions for revenue growth from AI initiatives, the sustainability of cost reductions from infrastructure learning curves, and the degree to which AI investment is expected to cannibalize or complement existing revenue streams.

Investors watching the next set of results may focus less on isolated AI announcements and more on operational indicates that support the ROIC argument, such as the trajectory of margins in cloud and software segments, the pace of AI-enabled revenue recognition, and the direction of capital intensity as companies expand AI capacity. Over time, the market will likely seek to validate whether the projected 25% to 50% ROIC window holds up against reported financial performance and updated guidance.