Close

DeepSeek: The Chinese Startup Turning the AI World Upside Down

January 27, 2025

1 min 40 sec read
4 Facebook Twitter Whatsapp Facebook Twitter LinkedIn Pinterest WhatsApp Copy Link Your browser does not support automatic copying, please select and copy the link in the text box, then paste it where you need it.
Move over, Silicon Valley—there's a new player shaking up the artificial intelligence game, and it's not where you might expect. DeepSeek, a Hangzhou-based startup, has burst onto the global scene with a bold claim: its AI models are as good as, if not better than, industry leaders like OpenAI and Meta, but at a fraction of the cost. And, by all accounts, they might be onto something.

AI Circuit Board
DeepSeek's breakthrough comes in the form of its latest AI models, DeepSeek-V3 and DeepSeek-R1. According to the company, these models rival the best in the business while offering a significant edge in cost efficiency. DeepSeek-V3, for instance, reportedly required less than $6 million in computing power to train—thanks to Nvidia's H800 chips—an impressive feat when you consider the billions being spent by U.S. tech giants on similar projects.

The startup's AI assistant, powered by DeepSeek-V3, has already outpaced OpenAI's ChatGPT as the top-rated free app on Apple's App Store in the U.S. This rapid ascent is causing waves across the tech industry, with some questioning whether the billions of dollars pledged by American companies toward AI development are yielding the expected results. Shares of major players like Nvidia have even taken a hit as the news spreads.

But it's not all applause and high-fives for DeepSeek. Skeptics, including Scale AI CEO Alexandr Wang, have raised questions about how the company managed to access the advanced Nvidia H100 chips despite U.S. export controls on such technology. Meanwhile, analysts from Bernstein have pointed out discrepancies in the reported training costs of DeepSeek's models, suggesting they may be higher than the startup claims.

So, who's behind this disruptor? DeepSeek was founded by Liang Wenfeng, a co-founder of the quantitative hedge fund High-Flyer. The startup is part of a larger vision Liang announced in 2023, aiming to dive deep into artificial general intelligence (AGI). High-Flyer itself appears to be deeply intertwined with DeepSeek, sharing patents and even office space.

DeepSeek's rise hasn't gone unnoticed in China's political circles, either. Founder Liang was recently spotted at a closed-door symposium hosted by Chinese Premier Li Qiang, signaling Beijing's interest in the startup's potential to boost the nation's tech independence. This aligns with China's broader strategy to sidestep U.S. restrictions and achieve self-sufficiency in critical industries like AI.

Whether DeepSeek is a genuine game-changer or just another overhyped startup remains to be seen. What's clear, though, is that it's already forced the AI world to sit up and take notice—and that's no small feat in a sector dominated by giants.

Want to read this in Spanish? Spanish Version >>

4 Facebook Twitter Whatsapp Facebook Twitter LinkedIn Pinterest WhatsApp Copy Link Your browser does not support automatic copying, please select and copy the link in the text box, then paste it where you need it.