Microsoft's $3.2B Anthropic Gain Dwarfs OpenAI Markdown in One Quarter

The numbers buried in Microsoft’s latest earnings tell a fascinating story about the diverging fortunes of AI’s two biggest labs. While the company posted a monster fiscal year—$331.8 billion in revenue and $133.7 billion in net income—the real intrigue is how it’s valuing its competing bets on OpenAI and Anthropic.

In the fourth quarter alone, Microsoft recorded a $3.2 billion paper gain on its Anthropic investment, a stake it took last November for $5 billion as part of a circular deal where Anthropic agreed to spend $30 billion on Azure cloud services. That single-quarter boost added 33 cents to diluted earnings per share. The gain is nearly as much as its OpenAI investment generated over an entire year.

OpenAI was a different story this quarter. Microsoft marked down its roughly 27% stake by about $600 million, shaving 7 cents off EPS. It’s not a panic-inducing number—for a company that just reported $35.8 billion in quarterly net income, it’s closer to a rounding error. But the direction matters. Microsoft now discloses its Anthropic position quarterly, a change from past practice, precisely because the contrast has become so stark.

Zoom out, and OpenAI’s full-year performance still looks solid: a $5 billion gain that contributed 67 cents to annual EPS. But the fact that Anthropic nearly matched that in just three months is genuinely surprising—so surprising, Microsoft felt compelled to disclose it. The revenue-share payments from OpenAI remain a black box, meaning we’re only seeing half the picture. Still, the trendline suggests the market is starting to price one AI lab’s prospects very differently from the other’s.