DeepSeek’s bold gamble: AGI over profit.

Liang Wenfeng rejects Silicon Valley’s playbook, betting everything on artificial general intelligence.

In a rare public revelation, Liang Wenfeng, the CEO of Chinese AI lab DeepSeek, has laid out his uncompromising vision for the future of artificial intelligence. Unlike Western counterparts such as Sam Altman of OpenAI or Dario Amodei of Anthropic, Liang is not chasing IPOs or trillion‑dollar valuations. His singular focus is artificial general intelligence (AGI), machines that can think and learn like humans.

Liang’s stance is striking in an industry dominated by profit‑driven strategies. He dismissed the idea of taking DeepSeek public, describing revenue as mere “sesame seeds” compared to the “watermelons” of long‑term breakthroughs. For him, financial gain is secondary to creating technology that genuinely benefits humanity. “Vision is not a slogan on the wall,” he emphasized, “it is how you operate.”

At the heart of DeepSeek’s mission is the belief that AI development is a staircase, each step building on the last. Liang argues that continuous learning is the missing piece preventing AI from reaching singularity- the point at which systems can autonomously improve themselves. DeepSeek’s models are designed first to serve the company’s own research, accelerating the creation of more advanced versions.

But the path to AGI is not just about algorithms- it is also about hardware. With U.S. restrictions limiting access to Nvidia’s cutting‑edge GPUs, Liang sees opportunity rather than setback. He claims Nvidia is “digging its own grave” by pushing Chinese firms to innovate. Huawei, with its supernode chips, is already partnering with DeepSeek to build alternatives. While these domestic chips may lag behind Nvidia’s by a couple of years, Liang believes they will eventually match performance while offering lower costs.

This strategy mirrors China’s broader industrial approach: producing competitive technology at lower prices. Just as Chinese manufacturing reshaped global markets, Liang predicts Chinese AI models will be powerful yet affordable. Unlike closed‑source Western systems that charge per token, DeepSeek and other Chinese labs such as Moonshot AI and Zai are releasing open‑weight models, making them accessible and cost‑effective.

Liang’s declaration is more than a technical roadmap- it is a philosophical challenge to Silicon Valley. By rejecting short‑term profit and embracing open collaboration, DeepSeek positions itself as a lab driven by goodwill and long‑term vision. Whether this gamble pays off remains to be seen, but one thing is clear: Liang Wenfeng has thrown down the gauntlet, and the race to AGI has entered a new phase.