2026.02.04

Microsoft's Ai Capital Allocation

Microsoft’s earnings report on January 28th beat market expectations yet the stock fell more than 15% soon after. Wall Street was spooked by two revelations in the earnings. First, 45% of Azure’s Remaining Performance Obligations come from OpenAI, a company with no clear path to profitability. Second, Microsoft is allocating more compute to various first party Copilot providers and internal R&D, at the expense of Azure customers.

The OpenAi concentration risk does not seem so troubling in a context where compute is supply-constrained. (Presumably Azure could find another buyer, though it is a systemic risk in so far as OpenAI’s failure to meet its obligations likely correlates with the bursting of the AI bubble, and the evaporation of the supply constraint.) The second disclosure is more risky, as none of the internal economics governing the massive capital outlays are disclosed or easily modelled.

First, there is little evidence that most Copilot products have product-market fit. According to Mary Jo Foley “only 3.33 percent of M365/O365 commercial customers have bought Microsoft 365 Copilot licenses for $30 per user per month”. An underwhelming performance compared to Anthropic, OpenAI, or Google (even allowing for different go-to-market models), especially considering Microsoft’s massive distribution advantage and access to the much vaunted Microsoft Graph, that could provide rich, relevant context for the AI. With the possible exception of GitHub Copilot, Microsoft seems to be pushing market demand with a string, rather than being pulled along by it.

Second, the deeper risk is not that Copilot lacks product-market fit (we are still in the early innings) but that Microsoft’s internal incentives make finding it unlikely. As I discussed in a previous post ("The hidden costs of free") one fundamental concern is how Microsoft is pricing this new capital allocation internally, if at all. A key insight of Econ 101 is that mispriced (free) resources often get misallocated. The lack of discipline and focus brought about by free resources can severely undermine competitiveness, as Copilot teams focus on cheap funnel optimization and short-term conversion, over customer pain points and product market fit.

Investors have compressed Microsoft’s multiples because of the opacity of the new capital allocation strategy. The $440bn question is this: Will Microsoft allocate Azure capacity and enterprise distribution channels via rationing, as in the former USSR, or via the price system? If capacity is rationed then I am less hopeful for the future than if the Copilot teams face metered market prices. Presently, investors ignore how capital is allocated internally, as well as the one metric that matters for product market fit: quarter-over-quarter paid Copilot seat expansion per existing enterprise tenant. Hence, they are pricing in the uncertainty brought about by this new capital allocation scenario.

Enjoyed this post? Subscribe via RSS to get new articles delivered to your feed reader.