The Rhyme: TSMC's Perfect Quarter Meets a Selloff & Its 2000 Echo

"History doesn't repeat… but it rhymes." — Mark Twain

TSMC delivered the best quarter in the history of the semiconductor industry yesterday morning, and the stock dropped three point four percent before lunch. Revenue hit $40.2 billion, up 34 percent year over year. Profit surged 77 percent to a record $22 billion. Then management raised the 2026 capex plan by eight billion dollars to a midpoint of $62 billion, and the market decided that was the problem. The SOX semiconductor index has now fallen eighteen percent from its June 22 peak, erasing roughly $2.7 trillion in four weeks. The last time the most important infrastructure company on earth crushed earnings and watched its stock crater into a falling sector, the year was 2000, and the company was Cisco Systems.

TSM EPS: $4.31 vs $3.80 est | Revenue: $40.2B (+34% YoY) | 2026 Capex raised to $60–64B | SOX: −18% from June peak | Sector: −$2.7T in 4 weeks

In the summer of 2000, every data packet that crossed the internet passed through a Cisco router or switch. Every telecom company building out the fiber-optic backbone bought Cisco gear. The phone companies, the cable operators, the dot-com startups running server farms in Northern Virginia — all of them needed the same equipment, and there was really only one place to get it. Cisco's stock had touched $82 a share in late March, giving it a market cap of $546 billion and making it, briefly, the most valuable corporation on the planet.

On the afternoon of August 8, John Chambers delivered the results from San Jose. Revenue for the fourth fiscal quarter came in at $5.72 billion, up 61 percent from the year before. For the full year, Cisco had brought in $18.93 billion — five years earlier, that number had been $2 billion. Chambers told analysts that "the second Industrial Revolution is just beginning" and that governments worldwide were turning to Cisco to transform themselves. It was the company's thirteenth consecutive beat, and the consensus on Wall Street was that Cisco was the one tech stock you could own safely.

But the Nasdaq had already peaked five months earlier, on March 10, at 5,048. By August it was below 3,800, down twenty-five percent, and nobody at Cisco seemed to notice because the orders kept coming. SBC, WorldCom, Sprint, Global Crossing — they were all laying fiber as fast as they could buy it. What Chambers didn't know was that the telecom companies were ordering gear with money they'd borrowed against future demand that would never arrive at the scale they'd assumed. The internet was real. The buildout had just gotten five years ahead of the need.

It took eight months for the music to stop. Sales plunged thirty percent in a single quarter. Cisco announced a $2.25 billion inventory writedown — the largest in corporate history — and laid off 8,500 workers. Chambers cut his salary to a dollar a year. On April 6, 2001, thirteen months after the stock had been at $82, it closed at $13.63. It bottomed at $8.60 in October 2002, down eighty-nine percent from its peak. The shares didn't fully recover for twenty-five years.

In both cases, the most important company in the cycle delivered its best numbers ever — and the market sold the news. The index was already falling, and the earnings beat couldn't stop it.

After Cisco's August 8 blowout, here is what happened. The stock bounced for a few days, touching the low $70s. Analysts raised their targets. The consensus was that Cisco was the safe port in the tech storm because the orders were real and the internet wasn't optional.

Then September came. The Nasdaq started sliding again. Cisco drifted to the mid-$50s by October. In November, Chambers delivered another strong quarter — revenue up 66 percent — and the stock barely moved. That was the tell. When the best earnings in the sector can't lift the stock, the cycle is turning.

By December, the first cancellations trickled in. Telecom companies that had been buying aggressively began pushing deliveries out by a quarter, then two quarters, then indefinitely. By February the trickle was a flood. In April, Cisco announced the writedown and the layoffs. The stock was already at $15, and it still had another $5 to go.

The pattern was clean. The stock peaked five months before the earnings peaked. The orders looked strong right until the moment they weren't. And the CEO was saying all the right things about structural demand the entire way down. For TSMC today, the question is whether we're in August 2000 — the start of the unraveling — or still in 1999, where the selloff is just a bump in a cycle with years left to run.

The edge: Alphabet reports Wednesday evening — the first hyperscaler to show its cards. The number that matters isn't revenue. It's forward AI capex guidance. If Alphabet trims spending plans, the Cisco parallel tightens fast. If it raises again, the rhyme breaks. The answer arrives in five days.

The VIX has barely moved during this semiconductor rout, sitting near 16 while $2.7 trillion evaporates from chip stocks. The last time the volatility index stayed calm while a sector crash was building underneath was the summer of 2000 — the VIX didn't spike until well after the damage was done.