Nov 9, 2026 · 40 min
Listen from 3:07
Listen at 3:07
AI’s infrastructure boom hides five systemic risks
Big Short Investor Explains How the AI Bubble Could Burst
Steve Eisman argues that AI may transform the economy while concentrated financing, power constraints, and fragile business models could still trigger a broad downturn.
- 1AI spending now links technology, utilities, industrials, financials, and corporate bonds to a small group of companies.
- 2Eisman identifies capital intensity, power shortages, concentration, OpenAI’s funding needs, and weak competitive moats as major risks.
- 3The technology can succeed even if individual AI stocks fail, making infrastructure exposure more attractive than speculative valuations.
Don't miss
Eisman’s five-risk framework shows how an OpenAI failure could spread through Oracle, hyperscalers, banks, industrial companies, and the broader U.S. economy.
The brief
Steve Eisman says AI has become the dominant force in markets and the U.S. economy, but its investment boom rests on a narrow group of companies and financing relationships.
The infrastructure build-out spans data centers, GPUs, power, lenders, and private capital, raising a basic question: who ultimately profits when spending becomes this capital-intensive?
Eisman weighs bullish, bearish, intermediate, and doomsday scenarios, arguing that transformative technology does not prevent overbuilding, Chinese competition, or a market downturn.
His five risks run from hyperscaler leverage and power constraints to concentration, OpenAI’s financial fragility, and AI models with few durable competitive moats.
The conclusion separates AI’s economic importance from individual stock success: Eisman favors picks-and-shovels infrastructure while warning against low-switching-cost businesses priced for extraordinary profits.
What was said on this episode
44 statements · 14 positive · 25 negative · 2 mixed · 3 neutral
AI currently drives markets and substantially influences the U.S. economy
“AI has become the dominant theme in investing today. It's driving the stock market, influencing the bond market, and to a remarkable degree, driving the entire US economy.”
Listen at 0:00
Agentic AI has substantial potential to change living and working
“I think it has enormous potential to change how we live and work.”
Listen at 3:07
A gigawatt-scale AI data center may cost approximately $30 billion
“A new large AI data center capable of supporting a gigawatt of power demand could cost around, get this, $30 billion to build and equip.”
Listen at 3:35
Hyperscalers are expected to spend roughly $850 billion on AI infrastructure in 2026
“In 2026 alone, hyperscalers, just the hyperscalers, are expected to spend roughly $850 billion on AI infrastructure and potentially even more next year.”
Listen at 4:19
AI will transform the world
“AI will change the world.”
Listen at 6:08
Anthropic and OpenAI could drive markets higher while NVIDIA sets records
“Anthropic and OpenAI will continue to drive the stock market to new and greater heights with NVIDIA setting endless records.”
Listen at 6:09
Eisman assigns the bullish AI scenario a 30% probability
“I give this narrative a 30% chance.”
Listen at 6:29
Chinese open-weight models could trigger AI price competition and weaken returns
“The Chinese open-weight models will put pressure on pricing, a price war will ensue, moats vanish, and the returns on investment don't measure up.”
Listen at 6:51
Eisman assigns the bearish AI scenario a 30% probability
“I give this narrative also a 30% chance.”
Listen at 7:27
Excess AI capital spending could erase gains and produce a bear market
“Result, the market unwinds, much of its AI gains, and we have a bear market for a time.”
Listen at 8:44
Eisman assigns the intermediate AI scenario a 40% probability
“I give this narrative a 40% chance.”
Listen at 8:51
Eisman assigns the AI doomsday scenario approximately zero probability
“I give this narrative about a 0% chance.”
Listen at 9:52
Market direction depends heavily on AI’s future health and success
“the direction of the market hinges on the future health and success of AI.”
Listen at 11:45
Hyperscalers have approximately $2 trillion in multi-year AI cloud backlog
“The hyperscalers have around 2 trillion of multi-year AI cloud-related backlog.”
Listen at 12:13
AI will transform software, medicine, and other sectors
“AI will change it.”
Listen at 12:38
AI may transform society without current spending earning adequate returns
“AI may be transformative, but that does not mean that all the money currently being spent will generate an adequate return.”
Listen at 13:26
ServiceNow and Salesforce likely retain moats through high enterprise switching costs
“some software companies are so embedded in the operations of enterprises that the switching costs are too prohibitive. I think that probably applies to companies like ServiceNow and Salesforce.”
Listen at 14:28
Most private-equity-held software companies have lost roughly half their value
“the value of most private equity-held software companies is down 50%, with cybersecurity software companies being a notable exception.”
Listen at 15:53
Private-equity software loan losses will become clearer during next year’s refinancing
“This will start to become clear when these loans start to be refinanced next year.”
Listen at 16:08
NVIDIA Blackwell GPUs can cost more than $5,000 each
“The cost to buy NVIDIA's latest GPU chip, Blackwell, can exceed $5,000 per chip.”
Listen at 17:34
Google raised $85 billion in equity capital for AI needs in 2026
“Google raised $85 billion in equity capital in 2026 to help fund its AI needs.”
Listen at 18:08
Oracle’s total debt rose from $94 billion to $155 billion over two years
“In the past two years, Oracle's total debt has increased from $94 billion to $155 billion.”
Listen at 18:39
Meta’s Beignet data-center debt remains off Meta’s balance sheet
“Meta is building the data center and has promised to rent the data center for Beignet for 20 years and bears the cost of any project delays and overruns, the debt will not sit on Meta's balance sheet.”
Listen at 20:37
The AI investment thesis requires construction of AI data centers
“The AI story cannot work unless AI data centers are built.”
Listen at 22:49
Permitting for AI data centers is delayed across many U.S. areas
“the process of permitting ai data centers is being delayed in many parts of the country”
Listen at 23:24
GE Vernova has benefited from demand for gas turbines
“That is why a stock like GE Vernova has done well, because GEV is a large builder of gas turbines.”
Listen at 23:54
Five customers accounted for 70% of NVIDIA’s accounts receivable
“70% of NVIDIA's account receivable, which is a proxy for revenue, was from five customers.”
Listen at 25:11
OpenAI accounts for approximately half of Oracle’s backlog
“around 50% of that backlog is from OpenAI alone.”
Listen at 25:59
Hyperscaler AI spending represents roughly half of projected 2026 GDP growth
“that spending accounts for around half of the 2% GDP growth projected for 2026.”
Listen at 26:13
U.S. economic health depends heavily on hyperscaler spending and leading AI labs
“the health of the U.S. economy is extremely dependent on hyperscaler CapEx. And hyperscaler CapEx is highly dependent on the health of Anthropic and OpenAI.”
Listen at 26:21
OpenAI’s underpriced consumer strategy proved flawed
“It chose to chase the business-to-consumer market, B2C. It did so by giving its product away or underpricing it with the hope that later on it would raise prices. That strategy proved to be flawed.”
Listen at 28:34
Eisman is not yet predicting OpenAI will fail
“I'm not predicting, at least not yet, that OpenAI will fail.”
Listen at 29:52
OpenAI’s failure could push the U.S. into immediate recession
“I'm starting to think that the demise of OpenAI would push the U.S. into an immediate recession.”
Listen at 31:37
Failure of Anthropic or OpenAI within a year would have severe consequences
“If either company fails within, I believe, a year, the ramifications would be severe.”
Listen at 31:53
Amazon, Google, Microsoft, and Oracle shares would decline after OpenAI failure
“These stocks, Amazon, Google, Microsoft, and Oracle, just to name a few, they'd all correct.”
Listen at 32:22
AI-beneficiary industrial stocks would decline after an AI-lab failure
“And these stocks would decline as well.”
Listen at 33:05
AI companies currently lack durable competitive moats
“There don't seem to be any moats around AI.”
Listen at 34:10
Switching costs between AI models are very low
“the switching cost between models seems to be perhaps zero, but certainly very low.”
Listen at 34:16
Chinese open-weight models cost approximately $0.20 per million tokens
“openweight models can cost around $0.20 per 1 million tokens, so much cheaper.”
Listen at 35:56
Anthropic and OpenAI cannot withstand a prolonged AI price war financially
“Anthropic and OpenAI do not have the balance sheets to withstand a price war.”
Listen at 36:22
AI will generate substantial profits for some participants
“I think an enormous amount of money will be made from it.”
Listen at 37:24
Eisman prefers AI infrastructure and picks-and-shovels companies
“I'm much more interested in the picks and shovels.”
Listen at 37:30
Investors should be cautious with low-switching-cost AI businesses priced for huge profits
“Where I'd be much more careful is in businesses where investors are assuming enormous future profits despite very low switching costs.”
Listen at 37:54
AI stocks can disappoint even if the technology succeeds when expectations are excessive
“when valuations assume an extraordinary future, you don't necessarily need the technology to fail. for the stocks to disappoint.”
Listen at 38:40
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.