You know, I've been thinking about something that's been bugging me about Warren Buffett's track record. Here's a guy who built Berkshire Hathaway into a trillion-dollar behemoth by sticking to a pretty straightforward playbook: buy quality businesses at reasonable prices, hold them forever, and let compounding do its thing. Over six decades, his Class A shares delivered nearly 6,100,000% in cumulative gains. That's not luck—that's discipline.

But here's where it gets interesting. Back in Q3 2022, when the market was getting absolutely hammered, Buffett and his team made a move that looked brilliant at the time. They picked up 60 million shares of Taiwan Semiconductor Manufacturing—TSMC—for around 4.12 billion dollars. The logic was sound: TSMC was the foundational chip maker for the entire AI revolution, making the advanced processors that power everything from Nvidia's GPUs to Apple's devices. The company had this cutting-edge chip-on-wafer technology stacking GPUs with high-bandwidth memory. When everyone else was panicking, Buffett was buying a genuine industry leader at a discount. Classic move.

Except it wasn't classic. Not for Buffett, anyway.

By Q4 2022, just a few months later, Berkshire had already bailed out of 86% of the position. And by Q1 2023, they were completely gone. We're talking about a five to nine month holding period for what should have been a decade-plus conviction play. Buffett's explanation? "I don't like its location, and I've reevaluated that." He was worried about geopolitical risks, export restrictions to China, and the implications of the CHIPS Act.

Here's the thing though—his timing to exit was absolutely brutal. Right after Berkshire sold, demand for AI chips exploded. TSMC's CoWoS capacity got crushed with orders. The stock just kept climbing. By July 2025, TSMC joined the trillion-dollar club. If Buffett had just stuck with his original thesis and held that initial 60 million share stake, it would be worth close to 20 billion dollars today. Instead, that decision to sell has cost Berkshire roughly 16 billion dollars in unrealized gains.

It's a rare miss for someone whose entire philosophy is built on long-term thinking and trusting in high-quality businesses to compound over time. This is the guy who's supposed to sit on his hands and wait for the market to give him opportunities, not panic-sell a foundational AI infrastructure play because of regulatory concerns that, in hindsight, didn't materially impact TSMC's ability to execute.

What makes this so notable is that it breaks one of Buffett's core principles—the one about patience and conviction in quality businesses. It's a humbling reminder that even the best investors can second-guess themselves when markets get uncertain. His successor, Greg Abel, is probably going to study this case pretty carefully. The lesson here seems pretty clear: stick to the principles that got you here, even when the noise gets loud.