Bismarck Brief

Bismarck Brief

AI 2026: China’s DeepSeek Cuts Compute Needed For Artificial Intelligence

Offering a superior option to China’s stifling and grind-focused universities, Liang Wenfeng has harnessed the country’s mathematics talent and founded an AI lab capable of advancing technology.

Samo Burja's avatar
Samo Burja
Sep 03, 2025
∙ Paid
8
Share
Mobile interface of the DeepSeek app in January 2025. Photo by Solen Feyissa. Source.

This is the first Bismarck Brief of our upcoming AI 2026 report, which will feature in-depth analysis of the state of the artificial intelligence sector and technology, together with a comprehensive profile of key players. Subscribe now to never miss a report.

DeepSeek is one of the world’s top artificial intelligence research laboratories and is a wholly-owned subsidiary of the Chinese company High-Flyer, a quantitative hedge fund that specializes in machine learning-derived strategies. The January 2025 release of DeepSeek’s R1 large language model (LLM) brought the company global prominence: R1 performed roughly on par with the then-leading o1 reasoning model offered by the U.S. company OpenAI, but at a much lower cost of inference. DeepSeek surprised many observers who had assumed large U.S. companies had a substantial lead in the field due to access to greater quantities of more powerful, cutting-edge computer chips, and could hire what were assumed to be better software engineers and AI researchers. Through new technical breakthroughs, this relatively small Chinese company has however also now set new standards for performance in the rapidly growing AI sector that has attracted trillions of dollars in investment worldwide.

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 Samo Burja
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture