AI's IPO Race: Anthropic Files as Microsoft Drops Its Own Coding Models

Anthropic filed confidentially for an IPO on June 1. One day later, at Microsoft Build 2026, the company unveiled MAI-Code-1-Flash and MAI-Thinking-1 — its own inference-efficient AI models built to run on Azure and bypass the competition. The timing wasn’t accidental.

The moves are two sides of the same story: the AI industry’s sprint toward public markets and the infrastructure layer beneath it.

The Revenue Nobody Saw Coming

Anthropic’s annualized revenue crossed $47 billion in May 2026. At the end of 2025, that figure was $9 billion. That’s over 5x growth in roughly five months, driven almost entirely by enterprise adoption — KPMG deploying Claude across 276,000 employees, PwC rolling it out to hundreds of thousands of professionals. When Big Four firms move that fast, the market has already decided.

The company raised $65 billion in its Series H just days before the IPO filing, valuing it at $965 billion. It now eclipses OpenAI, which was valued at $852 billion in March. Both companies have hired bankers — Anthropic filed under seal, OpenAI reportedly engaged Goldman Sachs and Morgan Stanley. The race is on.

Microsoft Gets Serious About Playing the Layer

Microsoft’s Build announcements are notable precisely because Microsoft has been on both sides of this market — $13 billion invested in OpenAI, $5 billion in Anthropic, Azure hosting both. Now it’s pushing its own models.

MAI-Code-1-Flash is the flagship: a coding model that Microsoft says is “inference ultra-efficient,” available now in GitHub Copilot and Visual Studio Code. MAI-Thinking-1 is a reasoning model described as medium-sized, built for high performance at low token cost. Both run on Azure infrastructure — meaning Microsoft keeps the margin instead of paying a third party.

The number that matters from the announcement: after fine-tuning for consulting firm McKinsey, Microsoft claims its model outperformed OpenAI’s GPT-5.5 with 10x better cost efficiency. That’s the pitch — not just competitive models, but economics that OpenAI and Anthropic can’t match while running on their infrastructure.

Also on Tuesday: small Aion models that run on Windows PCs, updated cloud models for speech recognition and image generation, and a stated goal of moving “every company from consuming a frontier model to fully participating at the frontier.”

xAI’s IPO Will Make Musk First Trillionaire

Elon Musk’s xAI is gearing up to go public next week in what sources describe as the largest IPO in history. SpaceX filed the paperwork in May. A successful listing would make Musk the first trillionaire, according to multiple estimates. xAI operates Grok, which competes directly with Claude, ChatGPT, and Gemini — and has access to SpaceX’s GPU clusters (Colossus 1 and 2) as part of Anthropic’s infrastructure agreements.

The Takeaway

The IPO filings from Anthropic and OpenAI aren’t just financial events — they’re signals that the model’s layer is consolidating. When companies that control the inference infrastructure start filing to go public, it’s because they believe the window for private growth is closing and the extraction point is at market scale.

Microsoft’s counter-move is also a bet: that enterprises don’t want to be dependent on a single model’s economics forever, and that Azure-native models with 10x cost advantages will pull some of that spend back. Whether that’s realistic or spin is still unknown. What is known: the next twelve months will test whether AI valuations at this scale have a floor or just a ceiling.

For operators, the signal hasn’t changed: the window to build meaningful vertical AI businesses before the layer solidifies is measurable in months, not years.

Sources: