Link: China’s DeepSeek AI is hitting Nvidia where it hurts

A Chinese AI startup, DeepSeek, has overtaken OpenAI's ChatGPT as the top free app on the US App Store with its innovative chatbot. DeepSeek's chatbot uses models that are cheaper to train, using fewer resources than current leading AI models.

After unveiling the new R1 reasoning model on January 20, designed for intricate problem-solving, downloads soared. This model builds on their earlier V3 LLM, which claims to match the performance of industry leaders like GPT-4o and Claude 3.5 Sonnet at a lower cost.

DeepSeek reports that it required only 2,000 Nvidia chips to train its V3 model, a stark contrast to the 16,000 chips typically needed for similar AI projects. This efficiency sparked a significant drop in Nvidia’s stock price and raised questions about the sustainability of the current AI development models.

These claims are causing a stir among developers and investors, questioning whether the heavy investments in AI by giants like Nvidia and OpenAI are justifiable. If DeepSeek's methods are proven, it could reshape industry practices and funding allocations.

Currently, leading tech firms like Nvidia, Microsoft, and Meta are massively investing in AI, with projects like Stargate commanding budgets in the billions. With market trends showing a decline following DeepSeek’s announcement, the financial strategy of these tech behemoths is under scrutiny.

The implications extend beyond economics; they challenge the technological dominance and investment strategies of the biggest players in AI. This scenario highlights the potential shift in the landscape of AI development, driven by more cost-effective innovations.

 #

--

Yoooo, this is a quick note on a link that made me go, WTF? Find all past links here.