Nvidia Buys Hugging Face, AI Outages, and Data Center Dissent 9/3/26

**SPEAKER_1** (0:01)
Never bet against American grit or American energy. Through innovation, Venture Global is not only building some of the largest energy facilities in the world right here in the United States, but delivering American energy at a fraction of the cost and a fraction of the time. So while others are busy talking, we're busy building.
That's Venture Global. That's unstoppable energy.

**SPEAKER_2** (0:30)
Trading at Schwab is now powered by Ameritrade, bringing you an expanding library of education with even more ways to sharpen your trading skills. Access new online courses, insightful webcasts, articles, engaging videos, and more. All curated just for traders. Plus, guided learning paths with content designed to fit your unique interests. And no sifting to find exactly what you need, so you can spend your time learning to trade brilliantly. Learn more at schwab.com/trad And that's...

**Kelly Evans** (1:00)
Thank you very much, Scott. Maybe they won't hike after all. And what's going on with a lot of the large language models going down this morning? I'm Kelli Evans and welcome to The Exchange. Let's start with the markets. Stocks are up. Dow's up 600 points. S&P up 1%. NASDAQ still leading the way. After the Fed's Waller told Reuters he'd be inclined to support holding rates steady this month. The NASDAQ, the best performer, as you can see, up 1.4%.
Software names trouncing the hardware trade today. The Crypto Complex also outperforming Tesla up again, though still in the red this year, ahead of its expected CyberCap launch, plus a blockbuster deal that transforms Nvidia. The company buying Hugging Face, yes, that Hugging Face that was hacked by OpenAI bots in July, will tell you why it's such a big deal for Nvidia, which is morphing closer to a competitor with Anthropic and OpenAI.
An AI is where we begin today with this widespread outage affecting multiple different names, OpenAI's ChatGPT, Anthropic's Claude, and XAI's Grok all down simultaneously. Kate Rooney joins us with the latest. Kate, what do we know?

**Kate Rooney** (2:06)
Kelly, that's what we're seeing. We are seeing reports of these widespread outages with some of the leading AI platforms out there. This is according to the website DownDetector, which tracks all of these issues. Some of the outages include ChatGPT from OpenAI. There are also reported issues with Claude, the chatbot from Anthropic, as well as Grok, which is the AI model from Elon Musk's SpaceX.
Finally, Google's Gemini and some of these outages had actually fallen off around 1130 Eastern, so it does appear that some of them are coming back online. But we also had a big outage from Microsoft and Outlook this week, so we are seeing more of these and we'll keep you posted, but that is the latest here, Kel.

**Kelly Evans** (2:43)
Yeah, I wonder, we'll talk to David Kennedy in just a moment.
I do wonder if the timing of this is just a coincidence. Not just because the models are all coming online, but also what that flurry of activity might be doing. Like I said, we'll get into that in just a moment. Speaking of Anthropic and OpenAI, Kate, you have some exclusive new reporting on both companies.

**Kate Rooney** (3:01)
That's right, Kelly. Well, speaking of those companies, Google, OpenAI, and Anthropic have poured billions of dollars into building the world's most advanced AI platforms, but there is a new threat to those investments. Experts now tell us that foreign companies are trying to illegally access those systems and try to use them to train their own cheaper AI technology. For the first time, Anthropic is speaking on camera about some of these issues and these attacks that they allege what it could mean for its business as well as the risk to national security. According to that AI lab, companies are doing this by ripping off American technology.
Cybersecurity experts say foreign criminals are getting stolen credit card information and AI accounts from the dark web to illegally access platforms like Claude, Gemini and ChatGPT in places where these platforms restrict access like Iran, China, North Korea and Russia. But it's not just so they can check out the latest technology. Experts tell CNBC they're trying to steal it.
It's a hard-to-trace tactic called distillation. To do this, hackers will ask AI models thousands of questions. Then they use those answers to train their own smaller AI platforms, which are often offered at a fraction of the price. Travis Lanham is the chief technology officer of cybersecurity firm Armadin. He says not only are cyber criminals using individual consumer accounts, but larger business accounts that allow them to ask many more questions and allow them to execute this at scale.

**Travis Lanham** (4:37)
Those types of accounts that can be used to ask a lot of questions or do a lot of prompts and then capture that data versus the smaller limits that are applied to the individual user accounts.