In its fourth-quarter results for fiscal 2026 (period ended June 30, 2026), Microsoft disclosed how its stakes in the two largest rival AI labs performed. For the quarter, the company recorded a $3.2 billion gain on its investment in Anthropic, which increased diluted earnings per share (EPS) by about $0.33.
Microsoft invested $5 billion in Anthropic in November 2025 under a reciprocal arrangement in which Anthropic agreed to purchase $30 billion of Azure services. The company does not routinely revalue its Anthropic investment every quarter, but it did report the gain this period.
OpenAI holding marked down by roughly $600M
By contrast, Microsoft wrote down the value of its OpenAI investment by approximately $600 million in the quarter, reducing diluted EPS by about $0.07. Microsoft owns roughly 27% of OpenAI. Although Microsoft receives revenue-share payments from OpenAI, it does not disclose the amounts paid; instead it accounts for the investment’s carrying value.
The $600 million markdown was small relative to Microsoft’s overall results for the period: the company reported $90 billion in revenue and $35.8 billion in net income for the quarter. For the full fiscal year, Microsoft reported $331.8 billion in revenue and $133.7 billion in net income.
Full-year picture: OpenAI still contributed to annual gains
On an annual basis, Microsoft said its OpenAI investment produced a $5 billion gain for the fiscal year, adding $0.67 to diluted EPS. Microsoft’s diluted EPS for the fiscal year was $17.95.
Notably, the single-quarter gain on Anthropic was nearly as large as the full-year gain from OpenAI, which likely explains why Microsoft highlighted both positions in its quarterly disclosures.
Why this matters
Holding sizable stakes in multiple, competing AI developers exposes Microsoft to divergent short-term swings in valuation even while year-to-date results remain positive. The quarter’s figures illustrate that individual investments can move meaningfully in opposite directions within a short timeframe, affecting quarterly EPS differently from annual outcomes.
