At a recent DeepSeek investor exchange, Liang Wenfeng shared core views on the company's strategy, its AGI vision, and commercialization path, sparking widespread industry discussion.
Liang first clarified the company's long-term objective, stating that the ultimate goal is to achieve Artificial General Intelligence (AGI). However, he contrasted this with current industry trends, noting that last year the focus was on building chatbots and capturing consumer traffic, while this year the scramble is for enterprise revenue, driven by a fear of being left out. In Liang's view, these pursuits are not of primary importance.
He pointed out that internally, DeepSeek is genuinely focused on the AGI roadmap and how to achieve the next technological breakthrough.
He acknowledged an interesting paradox: "The things you most desire often remain out of reach, while those you pay less attention to can be obtained quite easily." This, he explained, is the source of DeepSeek's strategic advantage. "With our minds set on AGI and actively working towards it, when we turn to consumer or enterprise applications, it requires very little deliberate effort. We can execute them well with minimal focus. Operating from a position of advanced technology to tackle relatively lower-level technical challenges—this is what constitutes a 'dimensional dominance'."
Liang cited last year's consumer-side experience as evidence: "We didn't invest much effort on the consumer side. At one point, we even considered not maintaining those users, but they wouldn't leave. This year, the growth in enterprise revenue is also quite optimistic, likely better than our peers—but again, we haven't dedicated significant effort to it; it was completely incidental." He elaborated: "On the path to AGI, these technologies are inherently part of the journey. Offering them via API to serve users doesn't involve doing anything extra. Both consumer and enterprise applications are byproducts, intermediate outputs, on our road to AGI."
He further depicted the lightweight nature of this model: "We only need a few people to maintain the API. We don't even have customer service or sales teams; users come on their own." Contrasting this with other companies, he noted, "They build models specifically to serve consumer or enterprise clients, but our original intent is the pursuit of AGI. Commercialization is merely a coincidental output. This vision attracts and unites more top talent, creating an organizational advantage—this is another form of dimensional dominance."
Liang emphasized that current model technology is the most critical factor. "Other companies may have product or traffic advantages, but they lack a technological edge." Recalling the unexpected surge in popularity during last year's Lunar New Year period, he stated frankly: "That was completely outside our script; we never anticipated it. Our focus was simply on perfecting the technology."