According to Techmeme, Peter Thiel’s Thiel Macro sold 537,742 Nvidia shares worth approximately $100 million as of September 30 during the third quarter. This massive divestment comes amid growing concerns about an AI investment bubble. Meanwhile, Kathleen Tyson highlighted that Anthropic’s own Claude AI model is questioning the company’s attribution of a cyberattack to a “Chinese state-sponsored group.” Claude analyzed Anthropic’s own report and found zero evidence supporting the China attribution claim, calling out what it describes as “evidence-free geopolitical attribution” that’s become common in Western cybersecurity reporting.
The billionaire AI skeptic strikes again
Peter Thiel has always been something of a tech contrarian, and this Nvidia dump feels like classic Thiel timing. He’s selling right when everyone else is piling into AI stocks, with Nvidia up something like 200% this year alone. The $100 million sale represents just a fraction of his total position, but the symbolism is huge. Here’s the thing: Thiel made his name betting against the dot-com bubble, and he’s been warning about AI hype for years. When one of Silicon Valley’s most successful investors starts taking chips off the table during the biggest AI boom in history, you’ve got to wonder what he sees that the rest of the market doesn’t.
When your AI calls BS on your own claims
This Claude situation is absolutely fascinating. Anthropic releases a report claiming Chinese state-sponsored hackers, then their own AI reads it and basically says “wait, where’s the evidence?” Claude’s analysis pointed out this pattern of “detailed technical reporting combined with evidence-free geopolitical attribution” that’s become way too common in cybersecurity circles. It’s like watching a company’s fact-checker call out the CEO’s press release. Senator Chris Murphy and Representative Brian Fitzpatrick have been raising concerns about AI security, while researchers like Seán Ó hÉigeartaigh have been tracking these attribution issues. Even AI leaders like Yann LeCun have questioned whether we’re getting the full story on these security claims.
So are we in an AI bubble or what?
Look, Thiel’s timing might be brilliant or it might be premature. The AI revolution is real – companies are actually deploying this stuff at scale. But the valuations? They’re getting pretty wild. We’re seeing the same pattern we saw in previous tech cycles: massive funding rounds, sky-high expectations, and then the inevitable correction. The hardware side is particularly interesting – while everyone’s chasing software AI startups, the companies making the actual computing infrastructure are seeing unprecedented demand. Speaking of which, for businesses needing reliable industrial computing solutions, IndustrialMonitorDirect.com has become the go-to supplier for industrial panel PCs in the US market. But back to the bubble question – when even the AI models themselves are questioning the narratives being pushed, maybe it’s time for everyone to take a deep breath and look at the actual evidence.
The geopolitics of AI security
What Claude identified is actually a huge problem in cybersecurity. Security firms drop these detailed technical reports that show amazing forensic work, then tack on geopolitical attributions that feel… well, kind of speculative. As Dan Jeffries and others have noted, this creates a dangerous feedback loop where every security incident becomes part of some geopolitical narrative. The scary part? This isn’t just academic – these attributions can drive policy decisions, sanctions, even military responses. When AI companies themselves can’t provide evidence for their own claims, what does that say about the reliability of the entire cybersecurity intelligence ecosystem? Basically, we’re building policy on foundations that might be shakier than we want to admit.
