Hangzhou DeepSeek Artificial Intelligence Basic Technology Research Co., Ltd. did not just release a new AI model on January 20, 2025. It dropped a bomb on the economics of artificial intelligence. DeepSeek-R1, an open reasoning model, performs at a level comparable to proprietary systems like OpenAI’s GPT-4 and o1. The company’s training cost was significantly lower. That is the story. Not the benchmark scores, but the bill.
DeepSeek was founded in July 2023 by Liang Wenfeng, who also co-founded High-Flyer, a Chinese hedge fund. He runs both. That matters. High-Flyer provides the capital. DeepSeek provides the science. The hedge fund connection gives DeepSeek a financial runway most AI startups lack. They can take risks. They can train models without begging for venture capital.
The open nature of DeepSeek-R1 changes the competitive landscape. Proprietary models are black boxes. You pay for access. You never see the weights. DeepSeek-R1 is open. That means researchers, startups, and competitors can inspect it, modify it, and build on it. The company is betting that collaboration will outpace secrecy. That is a gamble, but the early evidence suggests it might pay off.
DeepSeek’s trajectory is steep. The company has existed for roughly 18 months. In that time, it moved from a founding idea to a model that competes with systems built by organizations with years of head start and billions in funding. Liang Wenfeng set out to create a company that could develop innovative AI solutions. He appears to have done exactly that.
The release of an eponymous chatbot alongside the model shows DeepSeek is not just a research lab. It wants users. It wants applications. The chatbot puts the reasoning model in front of ordinary people, not just engineers. That is a direct challenge to OpenAI’s ChatGPT and other consumer-facing products.
Where does this lead? First, expect the cost of training capable models to drop further. If DeepSeek can achieve comparable performance for significantly less money, other companies will follow. The barrier to entry in AI research just got lower. Second, expect the open vs. proprietary debate to intensify. DeepSeek-R1 is a concrete example of open models matching closed ones. That will pressure companies like OpenAI to justify their closed approach. Third, expect more Chinese companies to enter the global AI race. DeepSeek is not the only player in China, but it is now the one that proved it can compete at the highest level.
The hedge fund backing is a wild card. High-Flyer is a quantitative trading firm. It understands risk, data, and models. But it also operates in a different regulatory environment than Silicon Valley. How Chinese authorities view open-source AI models remains an open question. DeepSeek is pushing the boundaries of what is possible within that system.
DeepSeek-R1 is not just another model release. It is a signal. The signal says the center of gravity in AI development is shifting. The signal says open models are viable. The signal says a company founded 18 months ago can stand toe-to-toe with the giants.
