🧠 Food for thought

Implications, context, and why it matters.

A small team and disputed third-party estimates of Zhipu’s rapid financial growth

• Public discussions of Zhipu AI’s revenue and losses often skip day-to-day operating scale. A third-party profile says the company had 57 employees as of late 2024 1.
• The same profile lists 2024 revenue of $168 million, yet the page also gives other totals. That mix would put revenue per employee near $3 million, though it does not align with the company’s 2024–2025 results figures cited elsewhere 1.
• This small headcount fits Zhipu’s start as a research spinoff from Tsinghua University. Work has centered on foundational models such as ChatGLM and GLM-130B (large language models trained on broad datasets) 1.

Zhipu’s revenue model leans on on-premise deployments

• Zhipu AI’s on-premise deployment revenue is almost three times larger than its cloud API sales.
• The split differs from the API-first approach common in the West. It also fits demand from Chinese state-affiliated groups and large enterprises that want tighter data security and control.
• For leading Chinese AI firms, profits may come fastest through high-value contracts for private, customized installations. That route shapes a different competitive landscape than mass-market API access.