Microsoft’s fourth-quarter earnings call on Wednesday revealed a striking contradiction at the heart of its AI strategy: the company logged a $3.2 billion gain on its investment in Anthropic while simultaneously pitching its own AI models in direct competition with the startup.

The gain came from marking up the value of Microsoft’s stake in Anthropic, according to the company’s earnings report for its fiscal year ending June 30. It’s a massive paper return on what has been a relatively modest investment compared to Microsoft’s roughly $13 billion commitment to OpenAI.

But the real story wasn’t in the numbers. It was in how openly Microsoft is now competing with the companies it has backed.

During the earnings call, CEO Satya Nadella and other executives detailed Microsoft’s plans to push its own homegrown AI models, tools, and even a competitor to Anthropic’s Claude-powered products. The company is building what Nadella called a “super app” that will combine Copilot’s chat, coding, and agentic capabilities into a single interface launching later this year.

“Copilot is evolving rapidly from chat to Cowork to Autopilots,” Nadella said. “This quarter, we are bringing these Copilot experiences together, including code, in one super app.”

That super app will directly compete with standalone offerings from both OpenAI and Anthropic. Microsoft isn’t just providing infrastructure anymore. It’s building the product layer too.

The OpenAI situation is messier. Microsoft reported mixed results from its OpenAI investment in the same earnings report where it celebrated the Anthropic gain. The company didn’t disclose specific numbers for OpenAI, but the characterization as “mixed” suggests the return hasn’t matched Anthropic’s performance, at least on paper.

This makes sense given the trajectory of both companies. Anthropic has raised capital at increasingly higher valuations throughout 2025 and 2026, most recently closing a massive round that reportedly valued the company above $60 billion. OpenAI, meanwhile, has faced internal turmoil, leadership changes, and questions about its path to profitability despite generating billions in revenue.

Microsoft’s dual role as investor and competitor isn’t new in tech. Amazon Web Services hosts Netflix while competing with Prime Video. Google Cloud serves Spotify while running YouTube Music. But the AI situation feels different because the technology is moving so fast and the stakes are so high.

When Microsoft first invested in OpenAI, the deal positioned Microsoft as an enabler. It would provide the compute infrastructure while OpenAI built the models. That arrangement has clearly evolved. Microsoft now builds its own models, offers its own AI products, and competes directly for enterprise customers.

The Anthropic investment was always more arms-length than the OpenAI partnership. Microsoft doesn’t have the same level of access to Anthropic’s technology or the exclusive cloud provider relationship. But even that investment now looks like it might have been as much about hedging as it was about collaboration.

What’s striking is how comfortable Microsoft seems with the arrangement. There was no attempt during the earnings call to downplay the competitive dynamics. Nadella didn’t hedge when describing Microsoft’s own AI ambitions or try to position the company’s models as complementary to those of its portfolio companies.

The subtext: Microsoft has decided it can’t rely on any single partner for its AI future, even ones it has invested billions in. It needs to own the full stack, from chips to models to applications.

This creates an awkward dynamic for both OpenAI and Anthropic. Microsoft remains their biggest cloud provider and a major investor. But it’s also building products that will directly compete for the same enterprise customers. The companies need Microsoft’s infrastructure, but they’re also feeding a competitor every time they deploy a new model or sign a new customer on Azure.

For Microsoft, the strategy makes sense. The company is hedging across multiple AI approaches while ensuring it captures value regardless of which models or applications win. If OpenAI succeeds, Microsoft wins as an investor and cloud provider. If Anthropic succeeds, same thing. And if Microsoft’s own models win, it keeps even more of the economics.

The $3.2 billion gain on Anthropic is nice. But the real prize is making sure Microsoft doesn’t get cut out of the AI value chain the way it largely missed mobile. Investing in potential winners while building your own alternatives is expensive and creates conflicts. It’s also probably the smartest move available.