How Warren Buffett's TSMC Retreat Became a $16 Billion Lesson in Breaking His Own Rules

For decades, Warren Buffett built his legendary reputation by adhering to a deceptively simple set of investing principles. His journey with Berkshire Hathaway stands as perhaps the most compelling testament to this discipline—Class A shares have delivered nearly 6,100,000% in cumulative returns under his stewardship. Yet there exists one glaring exception to this otherwise pristine track record, a brief detour into Taiwan’s semiconductor industry that has cost his company close to $16 billion in lost opportunity (and counting).

What makes this particular misstep so instructive is not that it happened, but precisely how it contradicted everything Warren Buffett had spent six decades perfecting.

The Investing Principles That Built an Empire

Before examining what went wrong, it’s essential to understand what went right for more than 50 years. The foundation of Warren Buffett’s success rests on several interconnected principles that functioned almost like a financial immune system—protecting Berkshire from the worst excesses of market cycles while positioning it to capitalize on genuine value opportunities.

His most fundamental belief centered on patience and long-term orientation. Rather than treating stock ownership as a trading vehicle, Warren Buffett approached every investment as if he were buying an entire business to hold indefinitely. This perspective shifted his entire analytical framework. Price fluctuations that torment short-term traders become irrelevant noise when your investment horizon spans decades. Economic downturns that temporarily crater valuations transform into opportunities for the patient capital deployer.

Closely intertwined with this time horizon philosophy sat an uncompromising commitment to value. Warren Buffett famously distinguished between a wonderful company at a fair price versus a mediocre company at a cheap price. He recognized that overpaying for quality is far less damaging than underpaying for mediocrity. This meant he could sit idle for extended periods, hands metaphorically folded, waiting for the market to present genuine dislocations where high-quality assets traded at significant discounts.

Competitive advantages—or “economic moats” in his terminology—represented another crucial pillar. He gravitated toward businesses with durable competitive positions: companies whose brand commanded customer loyalty, whose operational efficiency created structural cost advantages, or whose network effects made them increasingly valuable over time. These were the businesses that didn’t just survive difficult environments; they actually strengthened their competitive positions during downturns.

The final element involved capital discipline. Warren Buffett favored companies with robust return-of-capital programs through dividends and share buybacks. These mechanisms ensured that even if he misjudged a company’s growth trajectory, he’d still receive tangible returns for his capital deployment.

The Geopolitical Miscalculation That Broke the Mold

During the third quarter of 2022, with equity markets in disarray and genuine price dislocations becoming apparent, Warren Buffett made his move. Berkshire Hathaway initiated a substantial position in Taiwan Semiconductor Manufacturing (TSMC), purchasing over 60 million shares for approximately $4.12 billion. On the surface, this decision aligned perfectly with his established playbook: a market downturn, a world-leading company trading at depressed valuations, a business with unquestionable competitive moats.

TSMC’s positioning appeared especially compelling because the company sat at the epicenter of the emerging artificial intelligence revolution. Its advanced manufacturing technology, particularly its proprietary chip-on-wafer-on-substrate process, stacked graphics processing units alongside high-bandwidth memory—the precise architecture demanded by exploding AI infrastructure buildouts. Major customers including Apple, Nvidia, Broadcom, Intel, and Advanced Micro Devices depended on TSMC’s fabrication capacity.

Yet what would become the defining mistake happened disturbingly quickly. Form 13F filings reveal that Berkshire Hathaway liquidated 86% of its TSMC stake during the subsequent quarter, completely exiting the position by the first quarter of 2023. A five-to-nine-month holding period. For a purported long-term investor, this qualifies as heresy.

When discussing the reversal with Wall Street analysts in May 2023, Warren Buffett’s explanation centered on a single phrase: “I don’t like its location.” His reasoning referenced the geopolitical complexities introduced by the CHIPS and Science Act, signed into law in 2022. The legislation aimed to bolster domestic American semiconductor manufacturing, and subsequent Biden administration export restrictions on high-powered AI chips bound for China raised legitimate questions about Taiwan’s long-term positioning.

Warren Buffett apparently concluded that similar export challenges would inevitably confront Taiwan or that broader geopolitical tensions would undermine the investment thesis. It was, by his own admission, a reevaluation.

The AI Boom That Reshaped Everything

The cruel irony lies in the timing of this exit. Almost immediately after Berkshire’s departure, demand for Nvidia GPUs reached apocalyptic levels. Supply constraints became severe, with production backlogs stretching into six-month territory. In response, TSMC aggressively expanded its manufacturing capacity, particularly for the advanced processes underpinning artificial intelligence. The company’s growth rate didn’t merely accelerate—it experienced something approaching exponential expansion.

TSMC’s stock price reflected this dramatic shift in circumstances. The company achieved trillion-dollar status during 2025, joining an exclusive roster of publicly traded companies that have reached that valuation milestone.

Had Warren Buffett retained Berkshire Hathaway’s original stake without selling a single share, the position would be worth approximately $20 billion as of early 2026. The math is unforgiving: a potential $20 billion value against the $4.12 billion purchase price, minus what Berkshire received from selling most of the position, yields roughly $16 billion in foregone gains.

Why This Matters Beyond the Dollars

This episode presents an uncomfortable truth about even legendary investors: geopolitical prognostication represents one of the most treacherous arenas for predicting future investment returns. Warren Buffett allowed himself to be diverted from his core competency—identifying great businesses trading at fair prices—by attempting to handicap complex international relations.

His geopolitical concerns, while not unreasonable in 2022, proved subordinate to the overwhelming technological demand driving the AI revolution. The market ultimately selected Nvidia’s insatiable demand for chips as the dominant force, overwhelming any export restriction headwinds that might face Taiwan.

More fundamentally, the TSMC decision violated Warren Buffett’s cardinal rule: maintain conviction in excellent businesses identified during market dislocations. The temporary exit from his time-tested investing framework cost Berkshire Hathaway approximately $16 billion. Whether viewed as a cautionary tale about straying from principles or as evidence that even the greatest investors periodically misjudge complex scenarios, the lesson remains potent.

Warren Buffett’s successor, Greg Abel, inherits an organization built on precisely the opposite approach—steadfast adherence to long-term orientation, rigorous value discipline, and resistance to the siren song of short-term trading. The oracle’s misstep has become his legacy’s most expensive tuition payment.