# Kimi K3 Just Vaporized $314 Billion From OpenAI and Anthropic's IPO Dreams > Source: > Published: 2026-07-23 13:11:00+00:00 # Kimi K3 Just Vaporized $314 Billion From OpenAI and Anthropic's IPO Dreams # Kimi K3 Just Vaporized $314 Billion From [OpenAI](/glossary/openai) and [Anthropic](/glossary/anthropic)'s IPO Dreams Moonshot AI launched its new flagship model, Kimi K3, and the pre-IPO market didn't flinch. It screamed. By the time the dust settled on July 23, Anthropic's implied valuation had dropped $232 billion, a 7.31% haircut. OpenAI shed $82 billion, down 5.62%. Combined, $314 billion in paper wealth gone. One model. One launch. Here's the playbook and it's the same one [DeepSeek](/compare/llama-4-vs-deepseek-r1) ran in January 2025. Build a Chinese open-[weight](/glossary/weight) model that matches or beats the Western frontier on key benchmarks, price it at a fraction of the cost, and let gravity do the rest. Kimi K3 doesn't need to be the best model ever made. It just needs to be good enough to make investors ask the question nobody in San Francisco wants to answer: why would anyone pay $30/M tokens for Sol or $25/M for Opus 4.8 when an open-weight alternative does 90% of the job at 5% of the cost? The market answered that question already. Moonshot had to stop accepting new subscriptions on July 23, not because the model failed but because demand crashed their infrastructure. They're adding capacity now. Existing users stay online. New users wait. The open weights drop on July 27. That's four days from now. Once AWS, Azure, and GCP can host K3 on US infrastructure directly, the China National Intelligence Law data residency concern disappears. You'll be able to run this thing in Virginia. For enterprise compliance teams that have been blocking Chinese-hosted models by policy, the door swings open July 27. What does this mean for the IPO calendar? Anthropic has been signalling late 2026, maybe early 2027. OpenAI's timing is less clear but the same window applies. Both companies are now going to market against a backdrop where the pricing power narrative that justified trillion-dollar valuations is under live-fire challenge from a company most Americans can't pronounce. Moonshot isn't DeepSeek. It's a different team, a different model, a different threat. But the pattern is the same and the market recognized it instantly. The hard truth that frontier labs are now facing: the model layer is commoditizing faster than anyone priced in. The moat isn't the weights anymore. It's distribution, infrastructure, and the application layer on top. [Claude](/glossary/claude) Cowork, [ChatGPT](/compare/chatgpt-vs-claude) Work, [Copilot](/compare/github-copilot-vs-cursor), these are where the value lives. The model underneath is becoming table stakes. When an open-weight model matches your $50/M token flagship on [reasoning](/glossary/reasoning) benchmarks, the product experience has to justify a 10x price premium. Good luck with that. OpenAI and Anthropic aren't going to zero. Their products are best-in-class, their distribution is unmatched, and their enterprise relationships are deep. But a trillion-dollar valuation doesn't require best-in-class. It requires best-and-nothing-else-even-close. That's not the world we're living in anymore. The open weights drop July 27. Watch the pre-IPO secondary market that day. If K3 runs clean on US cloud infrastructure, the $314 billion might just be the opening bid. Get AI news in your inbox Daily digest of what matters in AI. ## Key Terms Explained [Anthropic](/glossary/anthropic) An AI safety company founded in 2021 by former OpenAI researchers, including Dario and Daniela Amodei. [Claude](/glossary/claude) Anthropic's family of AI assistants, including Claude Haiku, Sonnet, and Opus. [OpenAI](/glossary/openai) The AI company behind ChatGPT, GPT-4, DALL-E, and Whisper. [Reasoning](/glossary/reasoning) The ability of AI models to draw conclusions, solve problems logically, and work through multi-step challenges.