I have to be honest, until I read today’s headlines I wasn’t very familiar with DeepSeek, but I’m learning quickly. DeepSeek seems to be giving the AI industry fits, and has shaken stocks of those that power the traditional AI models powered by Nvidia. Investors have also taken notice Nvidia is down 11.2 percent as of typing this.
Why such concern? DeepSeek is a large language model that is said to be as good as current models already in production at a fraction of the cost. We don’t have a full grasp of how this model was built and what chips might have been used to build it, given the current chip restriction in China. It has shot to the top of the Apple App Store today, surpassing the likes of ChatGPT and Copilot. Does this model continue to fight with the existing models? Maybe what it does do differently is consider alternate approaches before picking the best solution and then offer an explanation for its solution This approach is yet to be seen in the existing models.
DeepSeek is powered by the DeekSeek V3 model founded by Liang Wengeng is 2023. DeepSeek was created and released as an entire open source project for about 6 million dollars in training cost. That doesn’t even compare to the billions already spent on exiting models.
If this model sticks or it doesn’t, the concept is already in place. We’ll have to see how investors handle this news or if they tighten their wallets on the current models.
