Microsoft just made $3.2B on Anthropic. Then it pitched its own models to replace them.

Microsoft made $3.2 billion on its Anthropic investment last quarter.

One quarter. One bet. More than the full-year gain on OpenAI.

Then Satya Nadella walked into the earnings call and told enterprises to stop trusting frontier labs.

"Every customer wants the right model for each task based on quality, latency, cost, and compliance," he said.

Then he pitched Microsoft's own MAI models running on Microsoft's own Maya chips.

This is not diversification. This is a hedge against the companies Microsoft owns stakes in.

Azure just crossed $100 billion in annual revenue. Copilot hit 30 million paid seats.

Microsoft is no longer just the platform that sells OpenAI and Anthropic to enterprises.

It is now the competitor that sells cheaper alternatives — and profits when either lab stumbles.

If your AI strategy assumes Microsoft will always resell what OpenAI and Anthropic build, you are negotiating against your own vendor.

The entity selling you your AI models is also the entity betting against them.

Audit your multi-vendor AI contracts today.

The supplier with the most leverage is the one with a position on both sides of the table.

SOURCE: https://www.microsoft.com/en-us/Investor/earnings/FY-2026-Q4/press-release-webcast
VERIFIED: Microsoft Investor Relations FY26 Q4 press release, TechCrunch (July 29, 2026)
SIGNAL: Microsoft's dual position as investor in and competitor to the top AI labs creates a structural conflict that will reshape enterprise AI procurement negotiations.