🧠 Food for thought

Implications, context, and why it matters.

Beneath the rally, the clock is ticking

• Stock prices are up, yet Zhipu and MiniMax spend about $10 for each $1 in revenue 1.
• Compute bills drive the gap; in the first half of 2025, Zhipu’s research and development costs ran at more than 800% of revenue 1.
• Cash from their January 2026 initial public offerings covers about 12 to 18 months at today’s pace, so the search for profits has little time 1.

Hong Kong’s listing rules and China’s push for domestic tech champions

• Investors are backing domestically listed Chinese foundation-model developers (companies that build large, general-purpose AI models) instead of focusing only on US rivals such as OpenAI and Anthropic 1.
• Hong Kong rules let unprofitable tech companies go public; Chapter 18C, introduced in 2023 for specialized technology companies, cleared a path as private funding struggled with AI’s huge cash needs 2.
• A domestic price war cut API costs by about 92% and pushed firms onto different paths 1. Zhipu favors higher-margin on-premise deployments (customer-run software installed in a company’s own data centers) for state-owned enterprises, banks, or government agencies. MiniMax targets overseas consumer apps to dodge API commoditization 1