Internal emails reveal Microsoft invested in OpenAI out of fear of Google

Back in 2019, Microsoft was not the confident AI powerhouse running circles around Google with ChatGPT. Behind closed doors, its leadership was panicking.

A newly unsealed internal email, released as part of the US Justice Department’s antitrust trial against Google, reveals just how terrified Microsoft was about falling behind in artificial intelligence.

The email, titled simply Thoughts on OpenAI, was sent on June 12, 2019, by Microsoft CTO Kevin Scott to CEO Satya Nadella and co-founder Bill Gates. Scott’s verdict was blunt: Microsoft was multiple years behind Google in machine learning scale.

When copying Google took six months

Early on, Scott wasn't impressed by the AI showmanship from OpenAI or Google DeepMind. He viewed their early public feats, like AlphaGo, as game-playing stunts designed for PR rather than practical utility.

That changed rapidly when natural language processing models entered the scene. Once Scott dug into the actual model training capabilities at Google, he wrote that he became very, very worried.

The technical gap was glaring. Scott revealed that it took Microsoft engineers six grueling months just to replicate Google’s BERT language model and get it trained. The problem was not the talent, but Microsoft’s own infrastructure, which was fundamentally not up to the task.

Google was already weaponizing these models across its products. Scott noted that early AI advantages were hurting Bing in search, while small everyday features like Gmail’s Smart Compose were getting scarily good.

Nadella read Scott’s assessment and forwarded the thread directly to CFO Amy Hood, who controls the company purse strings, adding a single sentence: this is why I want to do this.

Shortly after, Microsoft announced its initial $1 billion investment into OpenAI. According to Business Insider, Bill Gates had been regularly meeting with OpenAI leadership since 2016 and helped broker the relationship.

Why it matters

This paper trail proves that Microsoft’s AI leadership today was not born out of foresight, but pure fear.

We think this might be one of the smartest strategic pivots in modern tech history. Nadella swallowed corporate pride. Instead of spending five years trying to fix internal infrastructure from scratch, Microsoft outsourced the frontier model research to OpenAI and retooled Azure specifically to meet their massive compute demands.

That gamble paid off massively. What began as a $1 billion defensive hedge turned into an investment exceeding $13 billion, effectively letting Microsoft hijack the generative AI boom and force Google onto its back foot for the first time in two decades.

Yet, it also highlights Microsoft’s lingering vulnerability. By relying so heavily on OpenAI instead of building the core foundation in-house, Microsoft tethered its flagship enterprise future to a notoriously volatile startup. Fear bought Microsoft a commanding lead, but managing that partnership is proving to be a whole new challenge.