How David Tepper's Latest 13F Reveals a Major Shift in AI Stock Strategy

When Wall Street’s top hedge fund managers file their quarterly 13Fs with the Securities and Exchange Commission, they’re essentially leaving breadcrumbs for other investors to follow. David Tepper, the billionaire founder of Appaloosa Management, is no exception—and his recent filings tell a fascinating story about where he believes AI opportunity truly lies. After more than three decades of finding hidden gems for his fund, Tepper’s latest moves reveal both significant losses and a strategic repositioning that challenges conventional wisdom about which AI stocks matter most.

The Form 13F filings, submitted up to 45 days after each quarter ends, offer a window into what sophisticated money managers are actually buying and selling. For David Tepper specifically, the patterns across five consecutive quarters paint a portrait of an investor exiting one AI narrative while doubling down on another—and the reasons behind these moves tell us plenty about the evolving AI landscape.

The Unexpected Exit: Why David Tepper Retreated from Oracle

When March 2024 rolled around, David Tepper’s Appaloosa fund held 2.3 million shares of cloud infrastructure giant Oracle. That position seemed solid—Oracle was riding the AI wave, and the company’s cloud services appeared perfectly positioned to benefit from the boom. But what followed was a methodical, quarter-after-quarter retreat.

The scale of the divestment is striking. Between late March 2024 and mid-2025, Tepper reduced Appaloosa’s Oracle stake by 93%:

• Q2 2024: 300,000 shares sold
• Q3 2024: 426,606 shares sold
• Q4 2024: 173,394 shares sold
• Q1 2025: 700,000 shares sold
• Q2 2025: 550,000 shares sold (150,000 shares remaining)

From David Tepper’s perspective, this likely looked like prudent profit-taking at first. Between March 2024 and November 2024, Oracle’s stock had rallied roughly 50%—a significant move for a company of its size. Locking in gains on that magnitude is what disciplined fund managers do.

But there’s more to the story. Tepper may have been concerned about the broader market context. By September 2024, the S&P 500’s Shiller price-to-earnings ratio had approached 40—a threshold that’s only been breached twice in 154 years during uninterrupted bull markets. When valuation extremes emerge, even high-quality growth stocks like Oracle become vulnerable to sell-offs. Additionally, Oracle had missed consensus earnings expectations in three of the last four quarters, suggesting the company might be losing momentum relative to expectations.

The irony? In recent months, Oracle shares have doubled, and the company dazzled investors with eye-popping guidance. Management projected that Oracle Cloud Infrastructure revenues would explode from approximately $10.3 billion in fiscal year 2025 to $144 billion by 2030—a 69.5% compound annual growth rate for its high-margin AI segment. CEO Safra Catz also revealed that remaining performance obligations (RPO), which represent future revenue from signed contracts, surged 359% to $455 billion. Those moves validated Oracle’s AI opportunity but came too late for David Tepper’s fund.

The Surprise Play: David Tepper’s Renewed Focus on TSMC

While David Tepper was systematically exiting Oracle, he was engineering a complete reversal on the other side of the AI hardware spectrum. Between March 2024 and December 2024, Tepper initially cut Appaloosa’s Taiwan Semiconductor Manufacturing (TSMC) position in half, reducing it from 500,000 shares to 250,000 shares. But rather than abandoning the company entirely, he pivoted—and hard.

In Q1 2025, Tepper purchased 20,000 shares. Then, in Q2 2025, he embarked on a significant accumulation, buying 755,000 shares in a single quarter. Appaloosa’s total TSMC position now stands at 1.025 million shares, representing a more than doubling of the fund’s exposure compared to where it stood before the buying spree began.

This move challenges the popular narrative about AI hardware dominance. Most investors instinctively point to Nvidia as the indispensable chip company for artificial intelligence. Nvidia’s graphics processing units (GPUs) are undeniably the brains behind data center decision-making and large language model training. That’s all true. But here’s what gets overlooked: Nvidia’s chips don’t exist without TSMC.

Taiwan Semiconductor Manufacturing is the company that physically produces many of the world’s most advanced semiconductors, including Nvidia’s cutting-edge AI chips. More specifically, TSMC’s chip-on-wafer-on-substrate (CoWoS) technology packages the high-bandwidth memory that data centers need to run complex AI workloads. Without TSMC’s manufacturing prowess, the AI revolution simply doesn’t happen at the scale we’re witnessing.

Advanced AI chip orders represent TSMC’s most explosive growth opportunity. The company has been expanding monthly CoWoS capacity to meet seemingly insatiable demand. Yet David Tepper’s appetite for TSMC extends beyond the AI thesis. TSMC manufactures wireless chips for smartphones, components for Internet of Things devices, and semiconductors for next-generation automobiles. These segments don’t grow as rapidly as AI chips, but they provide a stable revenue floor and generate predictable cash flow—exactly the kind of diversification that appeals to a sophisticated investor.

Finally, there’s valuation. When Tepper was accumulating TSMC shares in Q2 2025, the stock was trading at a forward price-to-earnings ratio between 12 and 18 times. Given that TSMC is expected to deliver solid double-digit sales growth, this represented an attractive entry point compared to many trillion-dollar technology companies trading at significant premiums. For David Tepper, the combination of AI exposure, business diversification, and reasonable valuation made TSMC too compelling to resist.

What David Tepper’s Portfolio Moves Signal for Investors

Studying the holdings of sophisticated fund managers like David Tepper won’t make you rich overnight. But the patterns reveal important clues about market sentiment and emerging opportunities. When Tepper systematically exits a narrative—no matter how bullish the headlines become—it suggests he’s already priced in the opportunity or spotted risks on the horizon. Conversely, when he aggressively accumulates a position like TSMC, he’s telegraphing conviction.

The broader insight from David Tepper’s recent activity is that the path to AI riches isn’t always the most obvious one. The companies that supply the picks and shovels—the manufacturers who enable others’ dreams—often deliver more reliable returns than the most celebrated players themselves. For investors trying to navigate the crowded AI landscape, watching where sophisticated capital flows can prove far more valuable than chasing momentum.