Imagine investing $13 billion in someone, giving them your house, your car, your cloud servers and then watching them sign a five-year lease with a competitor. That, more or less, is what happened between Microsoft and OpenAI.
SurveyAlso read: Jensen Huang says ‘AGI is now’: Truth behind viral clip explained
It started quite nicely. In 2019, Satya Nadella made a prescient bet, investing an initial $1 billion in a then-niche research lab. The logic was that Sam Altman and his engineers needed Nadella’s money and Microsoft’s Azure data centers to build world-changing models, and Microsoft needed OpenAI’s technology to finally threaten Google. Two tech entities that shared the same goal. The honeymoon period was glorious. GitHub Copilot launched, ChatGPT exploded, and Microsoft, for the first time in years, looked genuinely cool. Bing, I would find myself occasionally using Bing voluntarily.
Then, the cracks started to appear. The first tremor came in June 2023 when Microsoft rushed GPT-4 into Bing against OpenAI’s warnings. Then came November 2023’s boardroom drama, the OpenAI board dramatically fired Sam Altman, blindsiding Microsoft entirely. Microsoft scrambled to hire him and the board reversed. Altman came back. Nobody likes discovering their partner makes major life decisions without so much as a text. From there, it evolved to a transactional relationship to quietly adversarial.
Also read: What is tokenmaxxing: A game employees are playing to show how much AI they use
The financial disputes were meaty. OpenAI wanted to slash Microsoft’s revenue share from 20% down to 10% by 2030, cap its equity stake at 33%, and in exchange have Microsoft forfeit rights to future profits. Microsoft, sitting on $13 billion invested and exclusive Azure cloud rights locked in until 2030, looked at those terms and said, “Absolutely not.” Talks stalled deep into mid-2025 with neither side blinking.
OpenAI, impatient with the standoff, simply started looking around. In March 2025 came an $11.9 billion five-year pact with CoreWeave, plus a $350 million equity stake, routing major workloads away from Azure. Then Oracle capacity for the Stargate project. Then, spectacularly, a June 2025 Google Cloud deal cleverly routed through CoreWeave, giving it plausible deniability while bypassing Azure exclusivity almost entirely. And then the one that really stung was a $50 billion deal with Amazon Web Services in 2026, prompting Microsoft’s lawyers to reach for their highlighters. For context, 45% of Azure’s entire contracted backlog is tied to OpenAI. This wasn’t just rivalry at that point, it was a hostage situation in reverse.
The IP and antitrust battles added a nastier edge. By June 2025, OpenAI executives were internally discussing accusing Microsoft of anticompetitive behavior and seeking a federal review of their contract. Microsoft’s attempted $3 billion acquisition of coding assistant Windsurf, where Microsoft reportedly demanded tech rights for GitHub Copilot as a condition, directly threatening OpenAI’s own developer turf. Microsoft, meanwhile, had already quietly hedged in 2024 by hiring Mustafa Suleyman from Inflection AI for $650 million, standing up an internal AI division that whispered a clear message: we can build our own if we have to.
Then there’s the AGI wildcard which, to me, was the strangest dispute of all. Buried in the original partnership structure is a clause granting OpenAI’s board the power to materially alter the partnership once AGI is achieved. OpenAI has been quietly suggesting that milestone might be approaching. Microsoft publicly called that progress “nonsensical,” almost certainly because agreeing would trigger concessions they have zero interest in making. So now the two companies are debating, with straight faces, what counts as the birth of artificial general intelligence while their lawyers take notes.
OpenAI’s conversion to a for-profit structure further complicated things, delaying Microsoft approvals and fueling standoffs throughout 2025. Microsoft has responded with a mix of selective support – backing OpenAI’s private equity fundraising rounds at 17.5% return targets – and quiet hedging, including reported overtures toward Elon Musk’s xAI. It seemed like Microsoft was saying that we still believe in you, but we’re also keeping options open.
And now OpenAI’s pre-IPO filing in March 2026 listed Microsoft explicitly as a top business risk. Not a partner, a risk. That’s the kind of language that ends up causing billion dollar divorces which they haven’t formally done. Contractually, they can’t, atleast not until 2030. But what was once the most consequential partnership in tech history now resembles two people sharing a lease they both regret, each quietly moving their valuables to storage. Microsoft wants control more than dependency. OpenAI wants independence more than loyalty. The relationship that lit the AI revolution on fire is now, fittingly, fuelled almost entirely by legal fees.
Also read: Anthropic’s Claude AI can now use your computer to complete tasks, but there’s a catch
Follow Us
Vyom Ramani
A journalist with a soft spot for tech, games, and things that go beep. While waiting for a delayed metro or rebooting his brain, you’ll find him solving Rubik’s Cubes, bingeing F1, or hunting for the next great snack. View Full Profile
