Liang Wenfeng's 'no overtime, 6x profit' claim is a more dangerous signal than DeepSeek itself

Within 24 hours, the discussion around "Open source doesn't affect 6x profits" generated over 30,000 interactions in the tech community. 70% of the comments focused on the same question: How exactly did he pull it off?

This isn't just another "leaked internal sharing session." In my three years doing short video reviews, I've seen too many companies brag about their technology and teams, but few dare to put "profit" and "ease" together and say it so casually. Liang Wenfeng's original quote was, "We are very relaxed; we don't even need to work overtime." The impact of this statement is greater than any funding news.

Let's break down the three layers of logic behind this signal.

Layer 1: Open source isn't charity; it's the most ruthless monetization strategy

Traditional business logic suggests that open source means giving up direct revenue and relying on the ecosystem to sustain itself. But Liang Wenfeng's team used a counter-intuitive move: open-sourcing core capabilities while maximizing commercialization. Where does the 6x profit come from? It's not from selling API interfaces at high prices, but from the traffic bonus and brand premium brought by "open source," which leads enterprise clients to actively seek out big contracts.

I've tested DeepSeek's models. To be honest, on specific tasks, it's not as smooth as GPT-4o, but its cost-performance ratio and customization capabilities are crushing. When your model is open-sourced so that all developers can use it for free and modify it as they please, you essentially turn the entire developer community into your free testing team and word-of-mouth channel. This approach only works for teams that truly maximize efficiency.

A deeper question is: If the open-source model proves that "you can earn 6x profits without working overtime," are other AI companies still relying on 996 (work culture) to pile up manpower going down the wrong path?

Layer 2: Kimi employees' irony exposes industry anxiety

The news mentioned that the US accused K3 of distilling large American models, and Kimi employees ironically called this a "Guinness World Record achievement." This reaction is interesting. It indicates two things: first, distillation has become an open secret in the industry; second, everyone is using the name of "open source" to do "closed source" things.

Liang Wenfeng's "open source doesn't affect profits" hits exactly this pain point. When your model is open source, if others want to distill it, you don't need to accuse them because your customers simply don't care whether your model is distilled—they only care if you can provide cheaper and more stable services. It's like a knife seller who wouldn't sue someone for infringement just because they used his knife to cut vegetables, because his business model is selling the knife itself, not selling a "vegetable-cutting license."

Layer 3: Honor's new logo looking similar and Duan Yongping's "not selling Pop Mart"

Looking at these two news items together, they actually tell the same story: the misalignment between brand premium and asset pricing. Honor's new logo being pointed out as looking like Honor of Kings and Grok shows that visual "look-alikes" will become increasingly frequent in the AI era, as AI-generated design templates are leveling the creative barrier. Meanwhile, Duan Yongping saying "I probably won't sell Pop Mart within 10 years" tells us from an investor's perspective: an IP that seems "unprofitable" might be worth more than any tech company.

Returning to Liang Wenfeng's remarks, I suddenly realized that the real trend might not be "how AI changes the world," but "how AI companies change business logic." When a company can achieve "no overtime, no involution, 6x profits," it is actually redefining what "efficiency" means. And those of us creating content also need to start thinking: When AI can help you generate scripts, edit, and dub, what else can you rely on to maintain your "irreplaceability"?

[!note] I predict that within the next three years, more than 50% of AI startups will shift to an "open source + high-profit service" model, because Liang Wenfeng has proven that this path works and can be walked extremely easily. Those companies still struggling with "closed source protecting core technology" will find themselves increasingly like Nokia back then—holding the best technology but losing to the business model.

Original link: https://www.leiphone.com/category/zaobao/H3hlbyekoDsIhJld.html