DeepSeek prioritizes AGI research over products and commercial growth

Chinese AI developer DeepSeek is reportedly prioritizing artificial general intelligence (AGI) research over building consumer platforms or maximizing near-term revenue. The company is focusing on AGI development, with its open-source solutions, such as the R1 reasoning model, seen as byproducts of this long-term research objective.

Founder Liang Wenfeng is said to have clarified to investors that DeepSeek is not focused on competing for consumer traffic or near-term revenue. This approach explains the company’s emphasis on open-source models, lower API prices, and improvements in training and inference efficiency, setting it apart from AI companies focused on standalone products and enterprise sales.

Liang reportedly views continual learning as the next major technical challenge after AI agents. He described a development path from chain-of-thought reasoning to agents, continual learning, and eventually embodied intelligence. While current AI systems generally require users to re-provide context for each task, future models are expected to learn continuously from experience.

DeepSeek is not prioritizing 3D generation, video generation, or world models, which are seen as useful applications but not central to the development of intelligence. Coding agents are a priority as they can improve both the company's products and its own research process. Multimodal capabilities will still be developed for consumer and enterprise products.

The API pricing is reportedly designed to recover hardware costs in about 10 months, rather than maximize profit. DeepSeek also plans to keep its strongest models open source and use the same models internally and externally, rather than releasing weaker versions to the public. This strategy positions the company primarily as an AI research organization.