DeepSeek Founder Prioritizes AGI, Likely to Keep Top Models Open-Source

Chinese AI startup DeepSeek is prioritizing the long-term development of artificial general intelligence (AGI) over immediate profit maximization, according to its founder, Liang Wenfeng. In remarks to investors, Liang stated that the company is likely to maintain its most advanced models as open-source, believing that open development and commercialization are not mutually exclusive.

DeepSeek gained international recognition early last year with its V3 and R1 models, challenging the dominance of U.S. AI leaders despite export restrictions on advanced chips. The company recently concluded its initial fundraising round, achieving a valuation of approximately $52 billion. This move marked a shift from its previous strategy of rejecting external capital, driven by increasing competition from domestic tech giants like ByteDance and well-funded startups such as Moonshot and Z.ai.

Liang identified computing power as the most significant constraint, suggesting that the gap with U.S. AI capabilities lies more in resource access than in technical expertise. DeepSeek intends to establish its own large AI computing clusters in the future, though a decision on developing proprietary AI chips has not yet been made. Liang expressed a preference for purchasing chips at reasonable prices rather than undertaking in-house chip development.