Core Summary
ZeroOneAllThings was once one of the “Six Little Tigers” in the AI large-model competition, but it fell into difficulty due to lagging behind its competitors (other companies either went public or had high valuations, while ZeroOne cut its model team). Kai-Fu Lee has crafted a new narrative for the company: abandoning the pursuit of general-purpose large models and shifting towards B2B enterprise services, aiming to compete with American firm Palantir, which helps businesses integrate data for decision-making. ZeroOne relies on Li’s personal network as its primary sales force and focuses on the “Number One Leader Project” (where the company’s top executive personally drives the AI transformation). Currently, orders are growing rapidly, but the revenue conversion rate is low, and there are risks associated with a project-based model. Going public in the future will require proving that the company can operate independently of Li’s personal reputation.
1. Why Stop Competing with Large Models? The Forced Shift from “Six Little Tigers” to “Leopard”
ZeroOne was never among the leading players in the “Big Model Six Little Tigers”: Zhipu and Minimax have already gone public, with Moon’s Dark Side valued at 330 billion yuan (investors even consider it a French company), and Jieyue Xingchen is still in the model development phase. Even more problematic for ZeroOne was its early claim of having a “model + AI infrastructure + application” integrated defense strategy; last year, it handed over most of its model team to Alibaba—like a bun shop saying it would no longer make its own fillings, it would be embarrassing to still be called the “Bun Tiger.”
Li Kai-Fu’s response was to change the company’s focus: from competing on the quality of models to seeing who can make money first, comparing ZeroOne to a “leopard” (the prefix is more important than the animal itself). He argues that the “foundation model war” is unsustainable and suggests adopting the narrative of being China’s version of Palantir—after all, the previous technical expertise can still be utilized, just with a different story.
2. Li Kai-Fu as the Chief Salesperson: Is the “Kai-Fu Universe” a Shortcut or a Hidden Danger?
ZeroOne’s current strategy revolves around Li Kai-Fu’s personal brand: he acts as the primary salesperson, and the AI they developed is called “Kai-Fu AI.” The B2B approach involves Li meeting with the company’s top executives. The advantages are clear: no one will refuse an invitation from Li Kai-Fu; orders reached 500 million in 2025 and exceeded 1.5 billion by May 2026.
However, there are significant risks: customers are buying into Li Kai-Fu’s credibility, not ZeroOne’s products. It’s similar to how MCNs rely on influencers to attract traffic; if the influencer leaves, so does the customer. Li doesn’t have time to build a strong organizational structure—ZeroOne lacks the computing power, funding, and cloud ecosystem necessary for a solid foundation model. His personal network is their most valuable asset for generating revenue.
3. What Exactly Can ZeroOne Learn from Palantir? The Combination of “Number One Leader Project” and “Sovereign AI”
Palantir helps businesses organize complex data and processes into systems understandable by AI, such as streamlining the production of the A350 aircraft by integrating data from four countries and 5 million parts, resulting in a 33% increase in delivery speed. ZeroOne is learning two key aspects from Palantir:
- Number One Leader Project: AI transformation cannot be left to the technical department alone; it must be driven by the company’s top executive (otherwise, partial changes will be hindered by inefficient departments). Li Kai-Fu also emphasizes the urgency, saying, “Either you use AI to outperform your competitors, or they will do the same to you—there are no other options.”
- Sovereign AI: Companies fear being controlled by external platforms; ZeroOne offers to help them build their own AI systems, expanding the market from businesses to governments and industries (for example, local governments concerned about losing control of their AI capabilities can turn to ZeroOne for solutions).
ZeroOne’s “Wan Ce AI” platform aims to do just this: it deploys models within companies to create usable AI assistants, but the key is to first gain the trust of the top executive before implementing the solution.
4. Concerns: Will ZeroOne Repeat the Mistakes of the AI Four Dragons? How to Overcome Project-Based Models?
The AI Four Dragons (Shangtang, Yuncong) faced issues where they had many orders but no profits; Shangtang’s revenue in 2024 was 3.77 billion yuan, yet it lost 4.3 billion yuan, with over 3.3 billion yuan in accounts receivable. ZeroOne is facing the same problem: although it has 500 million in orders for 2025, only 250 million in revenue has been confirmed, and deliveries rely on personalized services (e.g., engineers working directly within customer companies to modify processes). This project-based model makes it difficult to scale up into a platform company.
Li Kai-Fu claims they want to avoid repeating these mistakes, but as long as deliveries still depend on human intervention, there’s a risk of slipping back into the same pitfalls. For instance, ZeroOne’s “Number One Leader AI” applications show a five-fold increase in orders, but it’s unclear whether these are one-time projects or long-term subscriptions (subscriptions are a hallmark of a platform company).
5. Li Kai-Fu’s Final Battle: How Long Can His Personal Reputation Last? A Critical Question Before Going Public
Li Kai-Fu is 65 years old, and this is likely his last entrepreneurial venture. He values his reputation and doesn’t want ZeroOne to end up poorly. However, the company is too dependent on him: customers are attracted by his presence, the narrative is crafted by him, and even the product names include his surname.
Going public (planned for 2027 in Hong Kong) will expose ZeroOne to public scrutiny: does its revenue come from products or projects? Are there customers who can continue using their services without relying on Li Kai-Fu’s personal influence? These questions cannot be answered by simply relying on his name. The founder’s credibility is a starting point, but ultimately, the company must prove itself with its own systems and products.
In essence, this news story highlights a once-promising AI company that, after failing in its original field, is trying to survive by changing its focus using the founder’s personal brand. However, whether it can transition from being driven by an individual to being driven by a systematic approach will be the real test of its survival. Li Kai-Fu’s story is compelling, but the market ultimately cares about sustainable, tangible revenue.
---
Please note that this translation attempts to maintain the original structure and tone of the Chinese text while adapting it to English financial journalism. Some phrases have been adjusted to fit the target audience and language context.