A Chinese startup just showed every American tech company how quickly it's catching up in AI

[Grant Anderson]

out of nowhere, a small Chinese startup called DeepSeek released one of the world's most advanced AI models, rivaling those of OpenAI, Google, and other American tech leaders. Chinese engineers accomplished this feat at a fraction of the cost, using none of the expensive, cutting-edge and prohibited US chips.

[stainedmonster]

This is why we should be skeptical of Altman’s push for more international AI and data regulation. It’s not about keeping the technology in check—it’s about making it harder for startups to compete by forcing them through hurdles that established companies never had to face. For example, if new laws prevented data scraping for AI training, it would only hurt new models, while existing ones would keep their advantage.

While I wish stronger protections had been in place from the start, our best bet now is to encourage a diverse AI ecosystem, ensuring no single company dominates the space.

[Polly Ranton]

The title should have been: A Chinese startup just exposed how inefficient and overhyped American VCs and startups are.

Deepseek was just a side project for a few bored hedge fund quants—no deep LLM experience, no cutting-edge hardware. Yet, with only $5.5 million and off-the-shelf tech, they’ve reached OpenAI’s level. Let that sink in. 😂

[Dray hand]

Is there any evidence that Deepseek really was trained for only $5.5 million on commodity hardware? Personally I have no idea, but considering how disruptive Deep Seek R1 release has been, I am really curious to know.