When Dario Amodei left OpenAI in 2021 and founded Anthropic, he took with him a handful of researchers and a fundamental disagreement about how to build safe artificial intelligence. Four years later, the two companies are locked in a rivalry that has drawn in the biggest names in tech and caught the attention of federal regulators. At the center of it all sits Microsoft, which bet $13 billion on OpenAI and is now quietly hedging that bet by partnering with the very company OpenAI’s former leaders built to compete with it.
Microsoft’s $13 billion anchor
The scale of Microsoft’s commitment to OpenAI is a matter of public record. In its quarterly SEC filing for the period ending September 30, 2024, Microsoft disclosed $13 billion in total funding commitments to OpenAI, accumulated across multiple investment rounds. That figure, filed under penalty of securities fraud liability, represents the largest single corporate wager on generative AI and has shaped how competitors, investors, and regulators evaluate the entire sector.
The investment gave Microsoft deep integration rights. OpenAI’s GPT models power Copilot across Microsoft 365, GitHub, and Bing. Azure became OpenAI’s exclusive cloud provider. For a stretch, the arrangement looked like a lock: Microsoft got the most advanced AI models in the world, and OpenAI got the computing infrastructure it needed to train them.
But exclusivity carries risk, and by late 2024, cracks in the arrangement had become visible.
The Anthropic hedge
Microsoft announced a partnership with Anthropic and Nvidia that brought Anthropic’s Claude models onto the Azure cloud platform, paired with Nvidia’s hardware. The Associated Press reported that the deal signaled a strategic shift away from sole reliance on OpenAI for Microsoft’s AI product roadmap.
The financial terms of the Microsoft-Anthropic arrangement have not been publicly disclosed, making a direct comparison to the $13 billion OpenAI commitment impossible for now. The gap could be vast or narrower than expected. What is clear is the intent: Microsoft wants enterprise customers on Azure to have access to AI models built outside the OpenAI ecosystem, and it chose OpenAI’s most direct competitor to provide them.
Anthropic, for its part, was already backed by Amazon, which invested up to $4 billion starting in 2023, and by Google, which committed $2 billion. The Microsoft deal added a third major cloud patron, giving Anthropic a presence across all three dominant platforms: AWS, Google Cloud, and Azure. That positioning makes Anthropic both a competitor to OpenAI and a supplier courted by the same tech giants regulators are watching most closely.
A rivalry rooted in a shared origin
The tension between OpenAI and Anthropic is not purely commercial. Amodei, along with his sister Daniela Amodei and several other early OpenAI researchers, departed over disagreements about the pace of development and the rigor of safety testing. Anthropic was founded with an explicit emphasis on AI safety research, a framing that implicitly criticized OpenAI’s direction under Sam Altman.
Since then, the two companies have competed for talent, for enterprise contracts, and for the trust of policymakers in Washington. Neither organization has issued public statements describing the other in openly adversarial terms, but the competition is visible in product launches, hiring patterns, and lobbying efforts. Industry observers have characterized the dynamic as a feud, though the word carries more weight than either company has put on the record.
OpenAI’s own governance upheaval in late 2023, when its board briefly fired and then reinstated Altman, added another layer. The episode raised questions about organizational stability that Microsoft, as the largest outside backer, could not ignore. OpenAI has since moved toward restructuring as a for-profit entity, a shift that may clarify its corporate governance but also changes the nature of Microsoft’s investment.
The FTC steps in
Federal regulators decided these intertwined relationships warranted formal scrutiny. In January 2024, the Federal Trade Commission issued compulsory 6(b) orders to five companies: Alphabet, Amazon, Anthropic, Microsoft, and OpenAI. The FTC’s announcement described an inquiry into generative AI investments and partnerships, requiring the named companies to produce detailed records on the structure, terms, and competitive effects of their deals.
A 6(b) order is not a suggestion. It is a formal investigative tool that compels document production, and noncompliance carries legal consequences. By targeting both the investors (Microsoft, Alphabet, Amazon) and the AI labs they fund (OpenAI, Anthropic), the FTC constructed a framework for examining how capital flows through the generative AI sector and whether those flows reduce competition.
In September 2024, the FTC published a preliminary staff report based on its initial findings, though the commission has not announced formal enforcement actions or complaints as of early 2025. Whether the inquiry leads to structural remedies, consent decrees, or a detailed public report remains an open question. But the investigation’s existence alone raises the compliance stakes for any new or revised partnership among the five named companies.
What the money trail reveals and what it doesn’t
Three tiers of evidence shape this story, and they carry different weights.
The strongest is the SEC filing. When Microsoft reports $13 billion in funding commitments in a quarterly report, that number reflects binding financial obligations, not aspirational press releases. It is as reliable as corporate disclosure gets.
The FTC press release sits on similar ground. A compulsory order naming specific companies and specific information demands is a matter of public record, not a leaked memo or anonymous tip.
The AP’s reporting on the Microsoft-Anthropic-Nvidia partnership is credible institutional journalism, but it remains one step removed from primary documents. The AP did not publish the contract, and its characterization of strategic intent is informed interpretation rather than a direct quote from a Microsoft executive about distancing from OpenAI.
What this layered evidence supports is a clear directional conclusion: Microsoft is diversifying its AI supplier base, Anthropic is expanding its footprint across competing cloud platforms, and federal regulators view the resulting web of investments as significant enough to investigate. The motivations behind these moves, whether Microsoft is hedging against OpenAI’s governance instability, seeking better pricing, or responding preemptively to regulatory pressure, are plausible but not directly attributed in any public filing or statement.
An alliance structure still being rewritten
Microsoft is not abandoning OpenAI. GPT models remain central to Copilot, Bing, and a range of Azure AI services. The $13 billion commitment is not being unwound. But the addition of Anthropic to Azure, combined with regulatory pressure and OpenAI’s own corporate restructuring, means the relationship that defined the first phase of the generative AI era is no longer the only one that matters.
Anthropic is not simply a backup option. It is an increasingly important counterweight whose fortunes are now tied to the same cloud and chip ecosystems that sustain OpenAI. And the FTC, by placing both labs and their largest backers under the same investigative umbrella, has made clear that the government considers these alliances a competition policy concern.
As of April 2026, the structural shift is best understood as a live process. Contracts are being renegotiated, regulatory findings are pending, and the balance of power between OpenAI, Anthropic, and their corporate patrons has not settled into a stable arrangement. The confirmed facts outline the boundaries of that change. The eventual outcome will be determined as much in regulatory filings and infrastructure build-outs as in model benchmarks or product launches.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
