Zhipu AI and MiniMax: Clashing All-Hands Emails Reveal Who's Exposed

Photo by Cheng Shi Song / Pexels

---

Zhipu (02513.HK) rose 12% on its first day of listing but pulled back the next day; three months later, its stock price is down 18% from the IPO price. MiniMax (00100.HK) broke issue on day one and is currently trading about 22% below its offering price. Both companies' 2024 financial reports show revenues under 1.5 billion RMB, yet their combined net losses exceed 5 billion RMB. Their Price-to-Sales ratios remain at 45x and 38x respectively—numbers that look like a bubble in any mature industry, but in the large model race, capital markets are still betting on the "AGI" card.

---

Short-term view: The collision of all-staff emails is essentially a scramble for the same capital narrative

On April 10, Zhipu sent an all-staff email with the core message: "Focus on commercialization, achieve break-even this year." On April 11, MiniMax sent its own all-staff email, with keywords being "Scale first, persist in long-term investment." Three days later, both companies coincidentally added the same AGI formula to their financial reports: AGI = Base Model Capability × Scenario Penetration Rate × Ecosystem Leverage.

This isn't tacit understanding; it's anxiety.

From a business model perspective, both companies are currently stuck at the same bottleneck:

• Highly similar revenue structures: API call revenue accounts for over 70%, government/enterprise customized projects less than 20%, and C-end products have almost no presence.
• Overlapping customer profiles: AI SaaS companies, small-to-medium developers, and some local government projects. Neither has secured a core enterprise order at the BAT level.
• Worrying gross margins: According to public financial reports, Zhipu's gross margin is around 35%, and MiniMax's is around 32%, lower than the industry average of 40%, indicating they don't hold pricing power.

```

// Let's do some simple math

Single inference cost (using a 70B model as an example): 0.008 RMB per 1k tokens

External selling price: 0.012 RMB per 1k tokens

Gross margin = (0.012 - 0.008) / 0.012 = 33%

Factoring in training cost amortization, GPU depreciation, and ops team costs...

Original link: https://www.tmtpost.com/8063081.html