Advertisement|Remove ads.

Advertisement|Remove ads.
A rough quarter for the iShares Semiconductor ETF (SOXX) has brought “The Big Short” investor Michael Burry’s criticism of AI spending into focus, including his warning that the industry’s reliance on power-hungry chips has created a “parameter trap.”
SOXX fell 11.3% in the third quarter, its worst quarterly performance since March 2025, while the bearish Direxion Daily Semiconductor Bear 3X Shares ETF (SOXS) gained 2.1%, snapping five consecutive quarters of losses.
Burry examined an 1880 newspaper account of Melville Ballard, a deaf man who recalled questioning the origins of humanity and the universe before learning written language. Professor Samuel Porter presented the case at the Smithsonian Institution. Burry interpreted it as evidence that reasoning can precede words.
Advertisement|Remove ads.
“Today, well into the 21st century, by putting language before the capacity for reason, we are not building intelligence; we are building an increasingly sophisticated mirror,” Burry said. He argued that scaling language processing cannot establish genuine understanding, proposing “Ballard’s Test”: an entity must demonstrate reason without language before it can be said to possess understanding.
Burry’s main concern is that developers are pursuing larger models and greater computing resources without resolving fundamental limitations. “This ‘bad start’ has led to a ‘parameter trap,’ where brute-force language processing powered by zillions of power-hungry chips has become an incredibly ironic bottleneck,” he said.
He pointed to compression and reasoning-first approaches as potential alternatives, citing discussions with Klarna CEO Sebastian Siemiatkowski and research including Google’s AlphaGeometry and Meta’s Coconut.
Advertisement|Remove ads.
“This multi-trillion dollar scaling myth may be rendered redundant by a return to the silence of pre-linguistic reason,” he said. His investment concern is that more efficient AI could reduce computing requirements, leaving chip suppliers and infrastructure investors with weaker returns than expected.
This week, Burry said he had covered direct shorts in Micron Technology, Nebius Group, Caterpillar, SOXX, CoreWeave, Nvidia and Palantir Technologies, replacing exposure with puts across several names.
His SOXX position moved into September 2027 puts with strikes in the low $400s. Nvidia puts carried September 2027 strikes in the mid $100s, while an enlarged Palantir put position targeted the same expiry at strikes in the low $100s.
Advertisement|Remove ads.
Micron exposure shifted into June-expiry puts around $500, and Nebius into June-expiry puts at double-digit strikes. Caterpillar and Oracle positions moved into December 2027 puts. He covered CoreWeave without immediately finding attractively priced replacement options.
Burry said research strengthened his belief that the AI bubble could burst “sooner than later.” The update followed his disclosure from last week that he had added “in some size” to shorts in Micron, Nebius, SOXX and Palantir amid concerns about expanding memory supply.
On Stocktwits, retail sentiment for SOXX improved to ‘neutral’ from ‘bearish’ a quarter ago as message volume rose 34% over the same period. Meanwhile, sentiment for SOXS flipped to ‘bearish’ from ‘bullish’ despite a 150% jump in quarterly message volume. Watcher counts increased 3% for SOXX and 1% for SOXS in a quarter.
Advertisement|Remove ads.
One user said, “$SOXX rip rip rip, buy the dip! If $MU doesn't deliver, they'll just announce a circular deal somewhere else and pump the market! It always comes from semis!!!”
Another user said, “$SOXS I found it so amusing. $AMD just hit ATH a few days ago. Some of the others like $NVDA TSM are inching close to ATH. If your rich dream is based on the bubble burst, it is DEAD wrong. The fundamentals justified at the current high valuation. When the AI bubble starts bursting, the SOXS must be in the 20s or even 10s. Aka, the timing is not there yet. The earliest you can short semi is when the leader NVDA TSM are showing peak earnings.”
Advertisement|Remove ads.
So far this year, SOXX has gained 89%, while SOXS has plunged 95%.
For updates and corrections, email newsroom[at]stocktwits[dot]com.
Advertisement|Remove ads.
Comments posted here will also appear on symbol pages.