$852 billion is a big number to defend when you're losing money faster than you're making it. The Financial Times reports that OpenAI's valuation is under pressure as the company changes strategy and investors ask harder questions about the company's valuation. Tech watchers on Hacker News see a "valuation crunch" coming for standalone AI companies. The structural problem is real. OpenAI and Anthropic depend entirely on AI revenue. Google, Meta, and Microsoft have existing businesses to fall back on if AI monetization stalls. Claude Code's rapid adoption among developers proves AI tools can find product-market fit. But that's one tool for one audience. Enterprise AI spending at scale remains unproven.
Competition isn't helping. Chinese AI company DeepSeek has emerged as a legitimate challenger, disproving assumptions that US companies would own this market. The open-source and local-model movements add pressure. If powerful models can run on consumer hardware, the moat around proprietary LLMs gets shallow fast. And hardware cycles could make today's massive data center investments look wasteful when next-gen GPUs deliver far better performance per dollar.
OpenAI isn't waiting around. The company brought on Richard Ho, a former lead engineer on Google's TPU chips, to build custom silicon. There's also the Rain Neuromorphics acquisition, plus ongoing work with Broadcom on design and TSMC on manufacturing. The goal is to cut what some call the Nvidia tax and bring down compute costs. Whether that's enough to justify $852 billion is what investors are trying to figure out.