Cade Metz on NVIDIA's Challenging Day:
When DeepSeek introduced its DeepSeek-V3 model the day after Christmas, it matched the abilities of the best chatbots from U.S. companies like OpenAI and Google. That alone would have been impressive.
But the team behind the new system also revealed a bigger step forward. In a research paper explaining how it built the technology, DeepSeek said it used only a fraction of the computer chips that leading A.I. companies relied on to train their systems.
The world’s top companies typically train their chatbots with supercomputers that use as many as 16,000 chips or more. DeepSeek’s engineers said they needed only about 2,000 Nvidia chips.
In today's stock market, the most significant worries centered around NVIDIA. The concern here being that their chips are probably overvalued. I have no idea if this might be correct, either way I am more interested in the macro impact, and I think Satya is onto something.
In the field of economics, the Jevons paradox refers to a counterintuitive situation that arises when we find ever more efficient ways to use a resource. Jevons thesis considered English consumption of coal in the 1860s.
Initially, any efficiency improvement reduces the amount of the resource required for a single application, this is great! However, as the cost of utilizing the resource decreases due to the improved efficiency, there is a tendency for more and more applications and scenarios to use the resource. This surge in demand ultimately results in higher total consumption of the resource, even though each individual application requires less of it. Historically, economies have tended to assume that efficiency gains would naturally lead to a reduction in overall resource consumption (Nvidia chips in today’s example). They have, however, often overlooked the possibility that the Jevons paradox could cause resource consumption to rise instead.
Food for thought!
