Skip to content
News7 min read3 April 2026

OpenAI raises $122 billion. Google gives Gemma away. Claude Code source code leaks.

OpenAI closed the largest funding round ever: $122 billion at an $852 billion valuation. Google released Gemma 4 open-source. Claude Code source code leaked.

Edition #9. OpenAI closed the largest investment round ever: $122 billion, at a valuation of $852 billion. Google launched Gemma 4 under a fully open license. The source code of Claude Code leaked. OpenClaw got a major update.

OpenAI raises $122 billion. Valuation: $852 billion.

OpenAI closed the largest investment round in the history of the tech sector on Monday. $122 billion in committed capital, at a valuation of $852 billion. To put that in perspective: that's more than Shell and ASML combined. The two most valuable companies in the Netherlands, added together, don't come close.

The round was co-led by SoftBank, a16z, D.E. Shaw Ventures, MGX and TPG, with Amazon, NVIDIA and Microsoft as strategic partners. In addition, BlackRock, Sequoia Capital and dozens of other institutional investors took part. For the first time, retail investors could also join: more than $3 billion came from individuals via banking channels.

What stands out: OpenAI runs $2 billion in revenue per month and has 900 million weekly users and 50 million paying subscribers. But the company isn't profitable yet. OpenAI is still losing billions per year. The valuation is fully based on expectation.

A year ago OpenAI was worth $300 billion. In twelve months it nearly tripled. That's not gradual growth. That's a different kind of capital movement.

And meanwhile you, as an entrepreneur, pay $20 per month for exactly the same technology that $122 billion is being bet on. The tools are there. The question is what you do with them.

Google launches Gemma 4 as a fully open model

Google DeepMind released Gemma 4 this week. The most powerful open AI model to date. And for the first time under an Apache 2.0 license. That's a fundamental shift. Earlier versions of Gemma had a restrictive license that hampered commercial use. That restriction is now gone.

The model comes in four sizes: from a compact 2B model that runs on your phone, to a 31B model that ranks as the number 3 open model in the world on the Arena AI leaderboard. It beats models that are twenty times larger.

What makes it relevant for entrepreneurs: Gemma 4 is optimized for agentic workflows. Function calling, structured output, tool integration. Exactly the things you need to build AI agents that actually execute tasks. And it runs on your own hardware. No cloud needed. No data leaving your business.

Claude Code source code leaked

Anthropic accidentally leaked part of the internal source code of Claude Code last week, their popular AI coding tool. No hack, no security breach. A mistake when packaging a release. Human error.

An X post with a link to the code was viewed more than 21 million times in less than 24 hours. On GitHub a repo with the leaked code became one of the fastest-growing repositories ever.

For Anthropic this is painful. Claude Code now runs on a run-rate of more than $2.5 billion in revenue per year. The leaked code gives competitors like OpenAI, Google and xAI a glimpse into how Anthropic built their most successful product. For users: no customer data or credentials were leaked.

OpenClaw update: background tasks, YOLO mode and MCP servers

OpenClaw got a major update this week. Two things stand out.

First: background tasks are now a real system. You can start tasks that run in the background, follow their status, and cancel them if needed. That sounds simple, but it means your agent can now do multiple things at once without you having to wait for it.

Second: MCP servers. OpenClaw can now drive external tool servers via the Model Context Protocol. In practice that means your agent can use tools that run on a different machine.

And a notable default change: new installations now run in YOLO mode by default. No more permission prompts for shell commands. Full trust in the agent. That fits the shift we see everywhere: AI agents are no longer treated as tools you have to approve every step for. They're treated as employees you trust to do their work.

Tarik Eraslan

Written by

Tarik Eraslan

Founder of AI Agent. Helps businesses implement AI in their daily workflows.

LinkedInMore about Tarik

Ready to deploy AI?

Schedule a call to discuss your AI Agent or explore our Academy.

OpenAI raises $122 billion. Google gives Gemma away.…