On Wednesday, Satya Nadella took the stage on Wall Street and delivered a speech no investor expected to hear. It wasn’t the usual refrain about the cloud and Azure margins. It was a declaration of war. Before analysts, Microsoft unveiled its own artificial intelligence models, deployment toolkits, and a direct competitor to Mythos, Anthropic’s flagship platform. Hours later, fiscal results confirmed what nobody in Redmond wanted to admit out loud: the investment in OpenAI had been a bust, while its bet on Anthropic—the company founded by former OpenAI employees—had paid off. The news isn’t that Microsoft is making money from artificial intelligence. The news is that Microsoft has decided it no longer needs either OpenAI or Anthropic to reign.

The Strategic Pivot: From Partner to AI Rival

For years, Microsoft played the patient investor card. It put money on the table, integrated OpenAI’s models into Azure, and sold artificial intelligence like selling Office subscriptions: with the ease of someone who doesn’t have to invent anything, just package it. But the fourth quarter of fiscal year 2026, which ended on June 30, has changed the rules. According to documents filed with the Securities and Exchange Commission, Microsoft acknowledged that its investment in OpenAI has not generated the expected returns. The Russian state agency TASS, always keen to spot cracks in the American giant, spread the news, emphasizing that Microsoft’s business model depended on someone else’s technology. Though the claim must be taken with a grain of salt—TASS is hardly an impartial source—the financial data is public and verifiable: OpenAI has not delivered the returns Microsoft anticipated.

Against that relative failure, the investment in Anthropic shined. The profits earned are no small figure: they represent a significant return on capital invested, according to TechCrunch calculations. But the money isn’t what matters; it’s what Microsoft did with it. Instead of doubling down on its Anthropic bet, the Redmond-based company unveiled its own artificial intelligence models, deployment toolkits, and a direct competitor to Mythos, the platform Anthropic had launched just six months earlier as its big enterprise push.

Cannibalization as Strategy

What happened on Wednesday was not a routine product launch. It was an act of deliberate cannibalization. Microsoft isn’t just competing with OpenAI and Anthropic: it’s devouring them. The company has understood that relying on third parties for the core of its artificial intelligence offering is an unsustainable strategic risk. If OpenAI or Anthropic decide tomorrow to change their licensing terms, raise their prices, or, worse, ally with an Azure competitor, Microsoft would be left without its main selling point.

That’s why the new models unveiled aren’t a lab experiment. They are market-ready products, with deployment toolkits that directly compete with the solutions OpenAI and Anthropic offer to businesses. The Mythos competitor, in particular, targets the heart of Anthropic’s business: large corporations needing customized, secure artificial intelligence deployed in their own environments. Microsoft already has the Azure infrastructure, the enterprise customer base, and now its own models. Anthropic, which until yesterday was its ally, has become its direct competitor.

The Geopolitical Context: Sovereignty as an Excuse

This move doesn’t happen in a vacuum. The geopolitics of artificial intelligence is reshaping business alliances. China, with labs like Qwen, DeepSeek, and Ernie, is advancing at a pace that unsettles the West. The U.S. administration has tightened chip export controls and pressured big tech companies to keep the most advanced artificial intelligence under national control. In this context, relying on California startups for critical artificial intelligence technology is, for Microsoft, an untenable position.

The Redmond company has decided to take control of its own destiny. But the decision comes with a cost: OpenAI and Anthropic, which until now saw Microsoft as a financial and commercial partner, now face a giant with more resources, more infrastructure, and, above all, more patience than they have. The question hanging in the air is whether the startups can survive without the backing of their now-biggest competitor.

The Future: A Three-Player Board

Enterprise artificial intelligence is being reshaped into a three-player board: Microsoft, which has decided to build its own ecosystem; OpenAI, which is desperately seeking new investors after the blow from Microsoft’s fiscal results; and Anthropic, which has just lost its main ally and faces a competitor that knows its weaknesses better than anyone.

Microsoft’s move is bold, but also risky. Building proprietary artificial intelligence models isn’t like launching a new version of Windows. It requires talent, data, infrastructure, and, above all, time. Meanwhile, OpenAI and Anthropic won’t sit idly by. The war for enterprise artificial intelligence has just begun, and the first blow has been struck by the one who, until yesterday, was footing the bills.

The final reflection is uncomfortable for those who believed artificial intelligence would be a market of stable alliances. Microsoft has shown that, in technology, loyalty lasts as long as a funding cycle. And that true power lies not in having the best model, but in having the ability to build one yourself.