Moonshot AI just banked another $2 billion, pushing its valuation to $20 billion, and the message is unmistakable: China’s open source AI strategy is a brutal, effective land grab. Backed by Meituan’s venture arm, Tsinghua Capital, and others, the Beijing lab is riding a wave of global demand for cheap, capable open weight models. Its Kimi K2.6 is now the second most popular LLM on OpenRouter, a direct threat to closed source behemoths like OpenAI. This isn’t just a funding round; it’s a declaration that open source isn’t just catching up, it’s becoming the default for inference.
The Open Source Trap: Cheap Inference, Massive Leverage
Moonshot’s strategy is blunt: sacrifice a sliver of top line performance for massive cost advantages. The Kimi models, now with K2.6 leading the charge, prove that developers will trade a few benchmark points for drastically lower API costs. This is the same playbook DeepSeek is using, reportedly now chasing a $45 billion valuation. The result is a classic tech ecosystem play. Moonshot is building a massive user base and a $200 million annual recurring revenue stream, not by selling a superior product in a vacuum, but by making the economics of AI inference irresistible. This commoditizes the entire industry, pressuring Western labs to justify their premium pricing.
Rewriting the Valuation Playbook in Beijing
The numbers are staggering. Moonshot’s valuation rocketed from $4.3 billion at the end of 2025 to $20 billion in six months, fueled by a total of $3.9 billion raised. This isn’t venture capital; it’s state backed industrial policy moving at hyperspeed. Competitors like Zhipu AI and MiniMax are now publicly traded, with massive market caps. The message to Silicon Valley is clear: the Chinese AI sector is not just surviving the export controls, it is innovating around them and building a self sustaining economic engine. For investors, the bet is that the open source wave, led by labs like Moonshot, will define the next era of AI, and you don’t want to be left on the wrong side of history.
Source: Techcrunch
