Nvidia shares falling again today as the AI trade looks for some new equilibrium following the release of Deep Seek.

As comes as Alibaba now is touting their new AI model, which it claims surpasses Deep Seek V3.

Adir DeBos of course has more on this rapidly changing AI race. Now we got Baba to play with Dee.

You know, rapidly might be an understatement. And really, this is

the takeaway. In the AI arms race, there are before and after moments. DeepSeek created one,

not unlike OpenAI's chat GPT moment a few years ago, a breakthrough so significant, it redraws

the battle lines. Now, complaints from Microsoft and OpenAI and President Trump's AI SAR, David Sacks, accusing

DeepSeek of using Open AI outputs, it's just an attempt that won't go anywhere to put

the cat back in the bag.

Because just like OpenAI trained ChatGBTGPT on Internet data, Deepseek trained its model

on updated generative AI internet data.

It copied the playbook, but far more efficiently,

and then it open sourced it. The new buzzword that Wall Street should take note of is distillation,

a way to shrink a big AI model into a faster, smaller one while keeping what it learned.

Here's what Databricks CEO, Ali Godzi, told me earlier this week.

So we're going to just see distillation happening left and right. It's already happening.

Like there's so many versions of DeepSeek that have been reproduced and redone just in the last

week as we speak. So this distillation is going to just create so much competition at the

LLM or the AI layer.

And so Alibaba's new Kwen model, which it claims surpasses deepseeks V3, that's just part of a new frenzy post deep seek.

One that meta, the American open source leader, is no doubt already engaged in.

Here's perplexity CEO, Arvin Srinivasa, a few weeks ago.

...