Unwelcome transparency arrived for Microsoft this week as a trove of internal emails from 2017 to 2019 spilled into public view, courtesy of discovery in Elon Musk’s lawsuit against OpenAI. The messages, unsealed by a federal judge, lay bare a company grappling with its own AI shortcomings and a visceral fear that OpenAI—then a comparatively obscure research lab—might ink a deal with Amazon Web Services. That dread, the emails show, was the accelerant that transformed Microsoft’s tentative curiosity about generative AI into a multibillion-dollar bet that would rock the tech world.

The emails emerged as exhibits in Musk’s ongoing legal crusade against OpenAI and its CEO, Sam Altman. Musk, a co-founder of OpenAI who parted ways in 2018, alleges that the company abandoned its non-profit, open-source founding principles in favor of a closed, for-profit model beholden to Microsoft. While the legal merits remain unresolved, the discovery process has wrenched back the curtain on how Microsoft decided to pour $1 billion into OpenAI in 2019—a partnership that gave Redmond an early lock on the technology powering ChatGPT.

Early Curiosities and Gates’ Skepticism

The paper trail begins in 2017, when Microsoft CEO Satya Nadella forwarded a Wired article about OpenAI to Chief Technology Officer Kevin Scott. Scott, who had joined Microsoft from LinkedIn the same year, quickly became the internal champion for a deeper alliance. Yet his enthusiasm wasn’t universally shared.

Bill Gates, still an influential voice inside Microsoft even after stepping away from day-to-day duties, peppered Scott with technical questions in late 2017. In one email chain, Gates expressed doubt that large language models could scale usefully, asking Scott to explain why he believed the technology was more than a research novelty. Scott responded with a detailed memo, but the exchange underscored the executive skepticism that Nadella would have to navigate.

“We Are Years Behind the Competition”

A pivotal moment arrived in June 2018, when Scott fired off a lengthy email to Nadella and CFO Amy Hood. The subject line was stark: “Thoughts on OpenAI.” In the message, Scott admitted that Microsoft’s own machine learning infrastructure was “years behind the competition” of Google and AWS. He argued that investing in OpenAI would not only give Microsoft access to cutting-edge models but also serve as a forcing function to modernize Azure’s AI stack.

Scott warned that OpenAI had already held preliminary talks with Amazon, and that the lab’s leadership was seriously evaluating AWS as its primary cloud provider. If Microsoft didn’t act, Scott wrote, “we risk losing this entire platform shift to our biggest rival.”

The Amazon Specter Looms

That fear—that OpenAI would defect to AWS—became the throughline of the internal debate. In multiple exchanges, executives fretted that Amazon’s superior cloud infrastructure and readiness to offer massive compute credits could woo Altman and his team. One email from Scott to Hood in early 2019 put it bluntly: “I’m very, very worried about Amazon. If we don’t land this, they will.”

AWS was already the dominant cloud provider, and its AI services like SageMaker had a deep head start. For Azure, still fighting for enterprise credibility, a high-profile AI partnership was a chance to leapfrog. But the risk was asymmetric: if OpenAI went to AWS, it would reinforce Amazon’s lead and leave Microsoft frozen out of what could become a foundational technology. If Microsoft won the deal and OpenAI failed, it would be a costly misadventure, but that was a risk Nadella was increasingly willing to take.

The Billion-Dollar Handshake

In July 2019, Microsoft announced a $1 billion investment in OpenAI, alongside a tight technical partnership that made Azure the lab’s exclusive cloud provider. The deal included complex provisions: part of the investment came as Azure credits, and Microsoft would eventually receive the lion’s share of OpenAI’s profits once the startup’s non-profit parent recouped its costs. At the time, many analysts viewed the move as a curious bet on a long-shot technology. The emails now reveal the desperation that fueled it.

Scott’s internal advocacy was crucial, but the emails show that Nadella ultimately made the call, overriding the lingering doubts from Gates and others. Hood signed off on the financial structure, while Scott was tasked with integrating OpenAI’s technology into Azure’s platform.

How the Wager Paid Off

The bet paid dividends faster than anyone imagined. OpenAI’s launch of ChatGPT in November 2022 ignited a global generative AI frenzy. Microsoft quickly built on that momentum by integrating OpenAI’s models into Bing, Office, and the Azure cloud itself. In 2023, Microsoft deepened its commitment with a reported $10 billion investment, and its market cap soared as investors bought into the AI narrative.

Crucially, the partnership forced Azure to accelerate its AI infrastructure. Microsoft built custom supercomputers for OpenAI’s training runs, and that expertise fed back into products like Azure AI and Copilot. Today, Azure is the second-largest cloud provider and has become the preferred destination for generative AI workloads, partly because of the OpenAI connection.

The Musk Factor and Lingering Tensions

The unsealed emails also reveal that Musk was on Microsoft’s radar early on. In a 2018 exchange, Kevin Scott noted that Musk’s involvement with OpenAI was a “wild card” but that aligning with Microsoft’s commercial cloud could give the lab stability absent from Musk’s mercurial influence. Musk departed OpenAI’s board later that year, and by 2024, he was suing the company, alleging it had become a “de facto subsidiary of Microsoft.”

The lawsuit has thrown Microsoft’s relationship with OpenAI into an uncomfortable spotlight. Regulators in the EU and the U.S. are already examining the deal’s competitive implications. And Musk himself has launched xAI, raising $6 billion to build a rival, with plans to use—you guessed it—AWS.

The Cloud Wars’ AI Chapter

For those watching the cloud market, the emails confirm what many suspected: AI has become the new theater of the AWS-Azure rivalry. Amazon has responded with its own major AI investments, including a $4 billion stake in Anthropic, a competitor to OpenAI. Google, too, has poured billions into its own Gemini models. But Microsoft’s early, fear-driven move gave it a two-year head start that the others are still trying to close.

The revelations also highlight how deeply competitive dynamics shape corporate strategy. Microsoft wasn’t merely excited about AI’s potential; it was terrified of missing out and ceding ground to Amazon. That terror, more than any technical curiosity, sealed the deal.

What Comes Next

As the Musk litigation grinds on, more emails could surface, potentially exposing further details about the OpenAI partnership’s financial terms and Microsoft’s internal calculations. For now, however, the picture is clear: the generative AI boom was not an inevitable product of academic research or entrepreneurial vision alone. It was also, in significant part, the result of a defensive maneuver by a cloud giant with something to prove—and everything to lose to a rival named AWS.