GenAI.Financial

Chinese tech dynamo makes Nvidia irrelevant! Beats OpenAI on a sliver of the budget! ?? Sensational headlines get all of the attention (and clicks), but is there substance to the hype over DeepSeek’s AI models and discoveries? In this case: yes.?? First things first—what is DeepSeek and why are we hearing about it now? DeepSeek is a Chinese AI research organization created by a Chinese hedge fund. And it’s been making waves for releasing AI models competitive with the top models from OpenAI on a far tighter budget and much more limited resources. If the story stopped there it would already be interesting. But DeepSeek made larger waves by making its top models completely available to anyone with powerful-enough systems to run it. It went even a step further and released the high-level details on how it was able to build such powerful models so cheaply. In doing so it gave up any semblance of moat or meaningful competitive advantage by allowing direct access its models and telling key secrets used to create them. This in itself renders billions of dollars in AI training costs from OpenAI, Anthropic, Google, etc... almost irrelevant in one fell swoop. The knock-on effects for AI hardware titans such as Nvidia are in my opinion overdone, but it's likewise clear DeepSeek has quickly leveled the playing field for true innovation in AI models. Now comes the fun part--think of all of the fascinating applications made possible via super cheap inference costs. A topic for another post certainly, but in the meantime know that the hype and shock is real.