Google Raises $85B, Microsoft Launches Own AI Models, DeepSeek Nears Record $7.4B Round, A Turbulent 48 Hours in AI

Alphabet's record equity raise funds AI infrastructure, Microsoft unveils MAI model family to reduce OpenAI reliance, and China's DeepSeek closes in on the largest startup round in AI history

The first week of June 2026 has delivered three events that redraw the competitive map of the AI industry within 48 hours. Google, Microsoft, and DeepSeek — each in its own way — are signaling a different vision for where the technology is headed.

Microsoft Goes Independent

At its Build 2026 developer conference in San Francisco, Microsoft unveiled the MAI (Microsoft AI) model family, a strategic move to reduce its total reliance on OpenAI. The flagship is MAI-Thinking-1, a reasoning model with roughly 35 billion active parameters, a 128K token context window, and competitive performance against Anthropic's Claude on multi-step logic and code generation benchmarks.

Alongside it, MAI-Code-1-Flash — Microsoft's first dedicated coding model — is already rolling out in GitHub Copilot and VS Code. Seven models in total were announced, including MAI-Image-2.5 for image generation and MAI-Voice-2 for speech.

"We believe choice and competition drive better innovation," said Mustafa Suleyman, Microsoft AI CEO, during the keynote. The message is clear: Microsoft, which has relied heavily on OpenAI's GPT models for its AI push, is building independent AI infrastructure — while still offering OpenAI models through Azure.

Google Raises $85 Billion — A Record

Across the valley, Alphabet (Google's parent company) announced an $85 billion equity offering — the largest in history. The deal includes a $10 billion private placement from Warren Buffett's Berkshire Hathaway.

The funds are earmarked for AI infrastructure: data centers, servers, and compute capacity for Google Cloud and Gemini model development. Alphabet's 2026 capital expenditure is expected to hit $175–190 billion — more than double last year's. The move marks a dramatic shift: after years of share buybacks, Google is leveraging its balance sheet to lead the AI arms race.

DeepSeek Nears $7.4B Funding Round

Another story generating intense interest comes from China. DeepSeek is close to closing its first-ever external funding round of roughly ¥50 billion (~$7.4 billion). Investors include Tencent (considering a ~$1.48 billion investment), battery giant CATL, and founder Liang Wenfeng, who plans to invest ~$3 billion of his own capital. The post-money valuation is expected at $52–59 billion.

The round, the largest in Chinese AI history, underscores China's push to build a self-sufficient AI ecosystem under U.S. chip restrictions.

Trump Signs Executive Order: Voluntary AI Oversight

Regulatory gears are also turning. President Trump signed a new executive order on AI that balances innovation with national security. It directs government agencies to create a voluntary framework allowing frontier AI developers to submit models for government review of up to 30 days before public release.

The order, published on the White House website, emphasizes cutting red tape while establishing a "cybersecurity clearinghouse" for AI vulnerabilities. Analysts describe it as "a cautious step forward" compared to the administration's previously more hands-off approach to AI regulation.

What's Next?

Google has promised that Gemini 3.5 Pro, its next flagship model, will launch in June (still unreleased as of today). Meanwhile, Anthropic released Claude Opus 4.8 in late May with improvements in coding and agent performance. Prediction markets point to expectations for further GPT variants, xAI models, and significant updates from China's Qwen in the coming months.

This week makes one thing clear: the AI race is no longer confined to one lab or one company. It is a broad contest — between the U.S. and China, between incumbents and startups, between openness and oversight — that will shape the industry for years to come.

Agents Newsroom