SemiconductorsLOT 261122 · 4 min read

Chip Stocks Shed $1.5 Trillion as SOX Logs Fifth-Worst Day

The PHLX Semiconductor Index fell 10.73% on Nov. 20, ending its worst week since 2001, while 62 chip stocks lost an estimated $1.5 trillion in one session.

By Rebecca Stone851 words

Key takeaways

  1. SOX closed at 8,408.59 on Nov. 20, 2026, down 10.73% in one session and 17.13% for the week.
  2. Jared Blikre counted about $1.5 trillion in losses across 62 chip stocks on Nov. 20.
  3. The weekly decline was SOX's fourth-worst since 1994 and its worst since the week ended Sept. 21, 2001.
  4. NVIDIA reported fiscal Q3 2027 revenue of $115.84 billion, up 103.2%, and net income of $65.29 billion; Q4 revenue guidance was $130.0 billion plus or minus 3%.
  5. SMH lost 17.8% for the week and holds more than 19% of its assets in NVIDIA, while XSD's largest positions each weigh about 3%.
Chip Stocks are Now Repeating What Happened in 2001
Chip Stocks are Now Repeating What Happened in 2001 — Buffik / pixabay

By Rebecca Stone

The PHLX Semiconductor Index closed at 8,408.59 on Friday, Nov. 20, 2026, after a 10.73% plunge; market analyst Jared Blikre counted about $1.5 trillion in market-value losses across 62 chip stocks that day. The index finished the week 17.13% below its Nov. 13 close of 10,146.32, its worst week since the week ended Sept. 21, 2001.

VanEck Semiconductor ETF, or SMH, closed at $400.06 after losing 17.8% for the week, its worst week since the same week in 2001. NVIDIA, the fund’s largest holding, closed at $125.53 after dropping 14.01% on Friday. SMH still places more than 19% of its assets in that single company.

The price action echoes 2001. The earnings do not. The sector’s leading companies now generate profits far beyond anything chipmakers earned during the dot-com cycle, though that cushion has not answered whether AI-chip demand can hold.

How does this compare with the 2001 collapse?

This was SOX’s fourth-worst week in the index’s daily history back to 1994. Each of the three worse stretches arrived as the dot-com collapse unfolded:

  • Week ended April 14, 2000: -27.1%, opening the collapse.
  • Week ended December 1, 2000: -20.2%, compounding it.
  • Week ended September 21, 2001: -25.2%, ending it.

U.S. markets stayed closed Sept. 11 through 14, 2001, so that weekly result compares the Sept. 21 close with the Sept. 10 close of 509.15. Friday’s decline ranked fifth among the index’s worst days and was its worst since a 12.25% fall on Nov. 5, 2026. Selling did not let up.

SOX peaked at 1,332.73 on March 10, 2000, and stood 71.4% below that high on Sept. 21, 2001. It then reached 354.63 on Oct. 2 before rallying 47% to 522.16 by Dec. 31. That rebound failed. The index’s final bottom came on Oct. 9, 2002, at 214.06, 83.9% beneath its 2000 peak.

How much faster is the current decline?

The current fall has outrun its dot-com predecessor. SOX took 151 days to shed 42.5% from its June 22 record of 14,634.72; the comparable 2000 decline took 214 days. That faster pace suggests traders rapidly unwound borrowed money and crowded positions, which likely uses up selling sooner unless forced sellers remain.

The harder question is when to step away from a mania before it breaks.

How did losses spread across the chip group?

Blikre also found five AI hardware and cloud names, NVIDIA among them, that each fell more than 20% for the week. The pullback dragged the S&P 500 down 3.70% to 6,910.39, while the Nasdaq Composite fell 5.28% to 22,617.73.

Do today’s leaders have stronger fundamentals?

Matt Orton, a strategist at Raymond James, sees a sharp break with 2000. He said: “In 2000, many chipmakers traded on hope, while today’s biggest names earn large profits.” That distinction supports the sector’s earnings base, but it does not settle whether AI-chip demand will hold.

NVIDIA’s third-quarter fiscal 2027 revenue reached $115.84 billion, up 103.2%, and net income reached $65.29 billion. For the fourth quarter, the company guided to revenue of $130.0 billion, plus or minus 3%.

Demand must also absorb customer concentration. Three direct customers represented 14%, 11% and 11% of NVIDIA’s quarterly revenue. Jensen Huang has resigned as CEO, while SEC enforcement staff are reportedly examining the company’s accounting practices; no misconduct has been established.

Even after the selloff, SOX remained 18.7% higher for 2026 than its 7,083.13 close at the end of 2025. The retreat came after a long advance, not from a sector already posting 2001-style losses.

How should ETF investors weigh concentration?

SMH’s structure concentrates that risk. More than 19% of the fund’s assets sit in NVIDIA, a company searching for a CEO and facing a reported SEC review. The 2001 record also shows that the sharp rally following a crash can fail, even when industry earnings exceed those of the previous cycle.

State Street SPDR S&P Semiconductor ETF, or XSD, tracks a modified equal-weighted index and offers a different distribution. Its latest fact sheet put each of its largest positions at about 3% of assets, limiting the damage one company can inflict on the portfolio.

24/7 Wall St.’s analysis sets a specific test for the bearish view: “The view is wrong if the SOX closes back above 10,146.32 before NVIDIA’s next quarterly filing explains the SEC matter.” For investors holding concentrated chip exposure, that filing—and whether AI demand supports the current earnings base—matters more than the 2001 analogy alone.

Going forward, stock pricing will turn on whether AI-chip demand can sustain NVIDIA’s $130.0 billion quarterly revenue guide; a miss would make its concentration in SMH and its accounting scrutiny more consequential than the earnings cushion alone suggests.

247wallst.com (Source)

Filed under

Share this article:

More from Rebecca Stone

Rebecca Stone

Show full bio

Correspondent covering technology and product news in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.

1,245 articles

Read also

« Previous article