Speaker 1: One Chinese startup just launched a new AI model to rival OpenAI, and it is called DeepSeek. It's raising questions about U.S. dominance in artificial intelligence. Madison, always been looking at this. First of all, what exactly does DeepSeek do?

Speaker 2: Yeah, absolutely, Stu. So it's an AI model and has a corresponding chatbot, like ChatGPT, of OpenAI. And it has significantly narrowed the gap between the U.S. and China when it comes to artificial intelligence. DeepSeek is the brainchild of a small group of researchers working for a Chinese hedge fund manager that have been able to produce technology that is on par with OpenAI and Google, even though those companies are sinking billions of dollars and years into development. Meanwhile, DeepSeek did it for cheaper in a couple of months with fewer chips. So on January 20th, DeepSeek introduced R1. This is an AI model that reasons, so it can do complex problem solving. It thinks through the problem, essentially. Silicon Valley advisor to President Trump, Marc Andreessen, said, DeepSeek said training one of its latest models costs just $5.6 million. Compare that to the $100 million that Anthropic used last year. As a result, tech stocks dipping today. You see them all in the red there. DeepSeek's AI model has raised serious questions. About American dominance in the space. It has rattled chip stocks because DeepSeek claims to have used just around 2,000 NVIDIA chips to train its V3 model. Compare that to the tens of thousands of chips used for U.S. models. But there are also questions around whether China is underreporting how much money and how much computing power they're using. Skepticism from Elon Musk as well. He says that there's no way they're able to do it with this amount of chips. DeepSeek's latest flagship model, V3, also. Won't answer political questions about China or Xi Jinping. And this is likely a similar setup to TikTok, right, when we talk about that issue. The concern there is that information used is sent to their servers in China. But this is an important distinction I want to get to. Because DeepSeek is open source, essentially it means that developers in the U.S. can use the code and programming locally on their computers if they have enough computing power. So that means that you can run it here without this concern of American dominance. And that's why it's so important to have American data being sent to China. So the biggest concern is the eroding American share of the AI market. We are unseating U.S. dominance in the space. And I think we saw quickly a full screen. It's risen in the charts in terms of U.S. use. So much so that people are having issues accessing DeepSeek today because of how many downloads it has.

Speaker 1: It's like a pivotal moment, isn't it?

Speaker 2: Incredibly. There you have it. Incredibly pivotal. We are so far ahead of China before this news that we were just miles. And now we're not. And then there's, as you've been mentioning throughout the morning, the concerns around how much money does this actually take? Because we've been investing so much money, infrastructure, power. And China, DeepSeek is claiming that they've really slashed that. So how does that work?

Speaker 1: It's fascinating. I don't know how it works. I really don't know how it works. Good stuff, though.