DeepSeek & the Energy Question No One is Asking

On DeepSeek, AI’s energy reckoning, and the unsustainable data center gold rush

Over the past few days, the AI industry has been in an uproar over a Chinese AI lab called DeepSeek. And for good reason. The company has just released a state-of-the-art Large Language Model “trained” for just $6 million, with just 2,000 Nvidia chips (the “R1”). That’s a fraction of the billions spent by hyperscalers like OpenAI, Google DeepMind, and Anthropic.

That low cost is coupled with the reveal that, since going live on January 20, DeepSeek-R1 has earned good marks for its performance… and rivals its larger competitors. There are also numbers flying around claiming it’s 27x times cheaper to operate than ChatGPT (per million token).

In short? We’re not just seeing yet another LLM launch. It’s a paradigm shift.

If DeepSeek can achieve GPT-4o-level performance at 1/100th of the cost, the implications extend far beyond model training. They force us to rethink everything about AI economics, infrastructure, and sustainability. That’s why the Nasdaq fell by 3% on Monday, driven by losses of chip maker Nvidia of nearly 17%.

But, amid all the debates over model quality, valuations, and the geopolitical AI race, some crucial…