When Steve Eisman makes a move, Wall Street pays attention. The former Neuberger Berman portfolio manager etched his name into financial history by famously shorting the 2008 subprime mortgage crisis a contrarian plays immortalized in The Big Short.
Now, Eisman is pivoting his gaze toward the current market darling: Artificial Intelligence.
In a sharp departure from the broader market’s unbridled tech optimism, Eisman recently liquidated his long-held stake in Alphabet and significantly reduced his broader AI exposure.
During a recent CNBC interview, Eisman didn’t mince words. He painted a picture of a tech sector dangerously over-reliant on the AI narrative, warning that equity markets have essentially morphed into a fragile, single-trade environment.
The relentless cheerleading for massive AI infrastructure spending, which fueled historic rallies for chipmakers and mega-cap tech over the past year, is facing a harsh reality check.
According to Eisman, the enthusiasm surrounding AI is actively shifting toward anxiety, and investors need to seriously re-evaluate the vulnerabilities lurking beneath these sky-high valuations.
Even sectors that appear entirely insulated from tech volatility, such as traditional investment banking, are deeply tethered to this dynamic because their recent earnings boosts rely heavily on AI-related financing.
The Threat of a CapEx Pullback
The current AI ecosystem is floating on an ocean of capital expenditure (CapEx). Tech hyperscalers are pouring hundreds of billions of dollars into data centers and Nvidia GPUs to build out their artificial intelligence infrastructure.
Up until recently, the market blindly rewarded every CapEx hike as a signal of future dominance.
Eisman sees this as the market’s Achilles’ heel. He points directly to Alphabet’s recent earnings report, where the tech giant beat revenue expectations but still saw its stock stumble after revising its CapEx guidance upward to a staggering $205 billion.
The narrative is growing far more complicated than it was a year ago. Institutional investors are no longer just asking how much these companies are spending; they are demanding to know exactly when that hardware spending will translate into sustainable software profits, becoming increasingly wary of the hidden cost of AI in business.
When pressed on what happens if these hyperscalers finally decide to cut their CapEx, Eisman was uncompromisingly blunt, stating the broader market would “go straight down.”
While a spending reduction might eventually prove healthy for long-term corporate balance sheets, the immediate shockwave would be catastrophic for semiconductor giants.
Nvidia, which recently posted 85% revenue growth on the back of this exact spending boom, would be caught in the crossfire of a severe market reaction. A sudden CapEx freeze would immediately deflate the single trade propping up current index valuations.
Questionable Moats and the Price War Reality
Beyond the sheer volume of infrastructure spending, Eisman harbors deep skepticism regarding the actual product layer of the AI boom: Large Language Models (LLMs).
The core issue is whether heavyweights like OpenAI, Anthropic, or even Google possess sustainable competitive advantages—or what the industry refers to as an “economic moat.”
Developing and training these frontier models is profoundly capital-intensive, yet the end products are increasingly looking like commodities. Eisman highlighted a looming threat that many Western investors are actively ignoring cheaper international alternatives.
As the global debate over open vs closed AI models intensifies, Chinese AI models, heavily leveraging open-source frameworks, are currently operating at roughly one-fifth the cost of their Western counterparts.
This dramatic cost disparity raises the specter of a brutal price war. If open-source models can achieve enterprise-grade performance for a fraction of the price, the multibillion-dollar valuations of proprietary LLM developers become incredibly difficult to justify.
You simply cannot maintain massive profit margins on a digital product that someone else is willing to give away or sell below your break-even cost.
Eisman’s decision to exit his Alphabet position isn’t a prediction that AI technology is a failure. Rather, it is a calculated assessment that the current risk-reward ratio in the stock market has completely decoupled from fundamental reality.
When a legendary contrarian quietly heads for the exits, it signals that the easy money in the AI trade has already been made and the hangover might be much closer than the market wants to admit.
Source: CNBC, "Steve Eisman Is Starting to Have Doubts About AI. The 'Big Short' Investor Just Sold a Key Tech Stock"
