AI Business
Microsoft Feared OpenAI Would Storm Off to Amazon and Shit-Talk Azure. It Happened Anyway.
Court documents from Musk v. Altman reveal Microsoft execs in 2018 feared OpenAI would storm off to Amazon in a huff and shit‑talk Azure. Eight years later, those fears came true within 24 hours.
The Smoking‑Gun Email
January 2018. Microsoft CTO Kevin Scott wrote an internal email weighing the risks of not funding OpenAI's compute demands. His conclusion was blunt and prophetic. "I guess the other thing to think about here is the PR downside of us not funding them, and having them storm off to Amazon in a huff and shit‑talk us and Azure on the way out," Scott wrote, The Verge reported from documents revealed during the Musk v. Altman trial. That email — now public — captures the moment Microsoft saw OpenAI as a reputational risk worth paying to neutralize.
The $300M Ask Microsoft Almost Walked Away From
The court filings — drawn from over 200 internal documents reviewed by GeekWire — trace the partnership's origins. In summer 2017, Sam Altman told Satya Nadella that OpenAI needed $300 million at Azure list prices. Azure chief Jason Zander pushed back: "For those numbers to make sense we'd have to be generating significant incremental revenue directly due to the deal ($500M+)," GeekWire reported. Microsoft didn't want to be what Scott later called "a bucket of undifferentiated GPUs," GeekWire confirmed.
Dota 2 Bots and Nadella's Congratulations
OpenAI's Dota 2 bot beating a professional player changed the tone. Nadella personally congratulated Altman. Altman pitched an even larger partnership. Scott acknowledged OpenAI's pulling power: "They are building credibility in the AI community very fast, recruiting well, and are going to be an influential voice," The Verge reported.
"Highly Dismissive" — The Mea Culpa
By 2019, Scott admitted he'd been "highly dismissive" of game‑playing AI, GeekWire reported. What changed? NLP breakthroughs. The models were reading, writing, and reasoning. Microsoft also grew alarmed about falling behind Google. One month after Scott's internal memo, which GeekWire reviewed, Microsoft announced its $1 billion investment, GeekWire confirmed.
2026: The Fears Come True — Within 24 Hours
On April 27, 2026, Microsoft and OpenAI ended their exclusive cloud agreement. The next day, OpenAI went live on Amazon Bedrock with GPT‑5.5, Codex, and Bedrock Managed Agents. AWS confirmed OpenAI usage now counts toward existing AWS cloud commitments. OpenAI told employees the Microsoft deal "limited our ability to meet enterprises where they are — for many that's Amazon Bedrock."
The Multi‑Cloud Marketplace Era
AWS customers can now access OpenAI's frontier models through the same Bedrock APIs they use for Claude, Llama, and Mistral. Meanwhile, Microsoft deepened its Anthropic relationship. The Verge noted both clouds now carry both labs. The era of one cloud, one model is over — cloud providers are distribution layers and the AI labs compete for shelf space.
Related News
May 9, 2026
OpenAI Ships GPT-5.5-Cyber, a Near-Mythos Model for Vetted Defenders
OpenAI launched GPT-5.5-Cyber, a specialized model for cybersecurity defenders that scored 81.9% on the CyberGym benchmark and completed simulated corporate cyberattacks. The UK AISI found it nearly as capable as Anthropic's Claude Mythos — 20% vs 30% success on a 32-step attack simulation. But the strategy diverges: Anthropic locks Mythos to ~40 orgs, while OpenAI offers tiered access through its Trusted Access for Cyber program.
May 9, 2026
Anthropic Inks $1.8B Cloud Deal With Akamai, Its Biggest Compute Bet Yet
Anthropic signed a $1.8 billion, seven-year cloud infrastructure deal with Akamai — the largest contract in Akamai's history and the latest in a series of massive compute commitments from the Claude maker. Combined with its SpaceX deal and 80x annualized revenue growth, Anthropic is building the most diversified AI compute backbone in the industry.
May 9, 2026
OpenAI Ships GPT-Realtime-2 — A Voice Model That Reasons Inside the Audio Loop
OpenAI launched GPT-Realtime-2 and two companion voice models on May 7, 2026. The flagship brings GPT-5-class reasoning to live voice with 128K context window.