Sometimes the best investment move is knowing when to walk away, even after a massive win. That's exactly what Meta Platforms Inc. (META) founder Brad Gerstner did with his sizable stake in Meta, and his reasoning cuts right to the heart of where value is flowing in today's AI revolution.

Why Altimeter Walked Away From a Winner

Speaking on CNBC's "Halftime Report" Tuesday, Gerstner explained why Meta was nowhere to be found on his top holdings list for 2026. Altimeter Capital originally jumped into the stock in late 2022, perfectly timing the company's efficiency transformation that Gerstner himself had publicly urged. That bet paid off spectacularly, with shares climbing from around $90 to over $700.

So why leave now? Because in Gerstner's view, Meta is hitting a wall when it comes to AI leadership. Sure, the company has been making moves, including rebooting its AI lab, but that doesn't change the fundamental problem.

"The fact of the matter is, they're not market leaders today in AI," Gerstner said plainly. He acknowledged Meta will be a "huge beneficiary" of AI technology, but in what he calls a "stock pickers market," he'd rather put capital into companies with lower multiples and more obvious near-term growth catalysts. Translation: Why settle for a company playing catch-up when you can own the ones already winning?

Following the Money to Infrastructure

Gerstner isn't just selling Meta—he's making a concentrated bet on the infrastructure powering the entire AI boom. His largest positions now sit squarely in the "picks and shovels" category: Nvidia Corp. (NVDA), Taiwan Semiconductor Manufacturing Co. Ltd. (TSM), and cloud provider CoreWeave Inc. (CRWV).

The thesis here is simple but powerful. Gerstner believes most investors are still "underestimating" how massive the AI supercycle will be. Consider the numbers: major tech companies dropped $150 billion on capital expenditures in 2023 to build data centers. By 2026, that figure is expected to blast past $500 billion.

"That's not speculative. That is purchase orders. Those are buildings. That's power," Gerstner emphasized. In other words, this isn't some pie-in-the-sky projection—it's concrete spending already in motion.

Why Nvidia Still Has Room to Run

When you're holding Nvidia as your top position after the stock has had a monster run, you better have a good answer for the "hasn't it gone too far?" question. Gerstner does.

With Wall Street consensus expecting 65% earnings growth this year, he argues the stock's current valuation of roughly 25 times earnings isn't remotely demanding. Altimeter doesn't need the multiple to expand to hit its return targets—they just need Nvidia to deliver the earnings analysts are already forecasting. That's a much less risky bet than hoping for sentiment shifts or multiple expansion.

How Meta Has Performed Lately

The timing of Gerstner's exit looks increasingly smart when you look at recent performance. Meta (META) shares have dropped 8.04% over the past six months, though they're still up 6.92% for the year. On Tuesday, the stock inched up 0.28% to close at $660.62, then slipped 0.21% in after-hours trading.

The technical picture shows weakness in medium and long-term price trends, though short-term momentum remains solid. For a company facing what Gerstner calls a "tough transition" to leverage AI effectively, that chart pattern tells a story that aligns with his fundamental concerns about leadership in the space that matters most right now.

The bigger picture here isn't just about one stock sale. It's about where the real value creation is happening in AI. Gerstner is betting that infrastructure companies with actual purchase orders and physical buildouts will outperform application companies still figuring out their AI strategy. Given the trajectory of capital spending and who controls the essential technology, it's hard to argue with that logic.