Oct 23, 2026 · 1h 45m
George Noble warns AI spending could become the next shale bust
"Coreweave's a Zero" | George Noble on Why He's Short Neoclouds, Why AI is a Shale 2.0 Bubble Headed for Debt/Depreciation Brick Wall, and Why He's "Coming For The Momentum Bros"
The episode tests whether AI infrastructure can earn adequate returns before debt, depreciation, and excess capacity expose the boom’s weakest assumptions.
- 1AI infrastructure may resemble shale: sustaining growth could require continuous spending while leaving behind debt and excess capacity.
- 2Noble’s strongest bearish conviction targets CoreWeave and other neoclouds, whose expansion depends on costly financing and persistent investor enthusiasm.
- 3Gold and energy serve as Noble’s hedges against monetary debasement, inflation, and years of underinvestment in supply.
Don't miss
Noble’s comparison of AI infrastructure with shale reframes the boom as a potentially self-defeating cycle of spending, debt, depreciation, and excess capacity.
The brief
George Noble argues that AI began with a compelling idea but has become a capital-spending race increasingly detached from revenues, profits, and economic reality.
His sharpest short thesis targets CoreWeave and other neoclouds, where expensive financing and aggressive expansion depend on investors continuing to fund the story.
Noble compares AI infrastructure with shale: both can require relentless investment to sustain growth, leaving debt, depreciation, and excess capacity when demand normalizes.
He remains bullish on gold and energy as protection against monetary debasement and structural underinvestment, rather than as simple bets on short-term prices.
The episode’s investor-psychology lesson is the Magellan gap: even strong fund performance can produce poor realized returns when investors chase and mistime it.
Noble avoids promising a precise AI crash date, but distinguishes tactical rebounds from his high confidence that CoreWeave and similar neoclouds face serious risks.
What was said on this episode
41 statements · 7 positive · 34 negative
AI is currently in a bubble.
“I think what I'm troubled by with AI bubble, it is a bubble.”
Listen at 2:08
AI unit economics do not support current spending.
“if you boil it down to its unit economics, the numbers don't end up.”
Listen at 2:25
AI compute and token prices are rapidly declining.
“the pricing on compute and tokens is collapsing at a breathtaking rate”
Listen at 4:25
AI companies’ balance sheets are rapidly deteriorating.
“the balance sheets of these companies are deteriorating at a breathtaking rate”
Listen at 4:31
Chinese open models can provide AI capabilities at 95% lower cost.
“the Chinese can do it all for 95% less with their open models.”
Listen at 4:54
Hyperscaler earnings will become negative by 2030.
“hyperscaler earnings will turn negative in 2030”
Listen at 11:37
Hyperscaler AI depreciation will exceed $500 billion by 2030.
“by 2030, depreciation is going to be running over $500 billion.”
Listen at 11:50
AI’s addressable market is much smaller than commonly expected.
“the addressable market for this thing, in my opinion, is a fraction of what people think it is.”
Listen at 13:06
Most major enterprises will not become Anthropic or OpenAI customers.
“almost universally, none of them are going to become customers of Anthropic or OpenAI.”
Listen at 17:01
Major enterprises will avoid AI labs because of compliance and data-security risks.
“they're just not going to use it because it's just, you know, all of a sudden you're giving all your trade secrets to scam all in.”
Listen at 17:16
Questionable CoreWeave bond repayment would imply worthless equity.
“if there's any question about whether Corweave bonds are money good, then you can mark the equity at zero.”
Listen at 19:03
AI companies will sharply reduce expansion and the system will collapse.
“The whole thing just collapses.”
Listen at 20:45
CoreWeave equity is worth only a fraction of its current value.
“I think the equity is worth a fraction of where currently.”
Listen at 25:57
CoreWeave stock has very poor investment prospects.
“I think it looks pretty dreadful.”
Listen at 26:33
Other neocloud companies face similar severe problems to CoreWeave.
“I'd throw all the other neoclouds to varying degrees under the bus in a similar fashion.”
Listen at 26:34
Buying stocks at ten times revenue is generally a poor strategy.
“buying stocks on 10 times revenues is a very bad strategy. Awful strategy.”
Listen at 29:03
SpaceX and Tesla are extremely overvalued and attractive short candidates.
“I've never seen companies, it's such overvaluation, such prime short candidates at scale. as we have with SpaceX and Tesla.”
Listen at 30:21
Shorting SpaceX will eventually be highly profitable.
“the fullness of time shorting SpaceX is going to be hugely profitable.”
Listen at 30:45
SpaceX cannot grow into its current valuation.
“When you're starting from such overvaluation, you can't grow into the valuation.”
Listen at 31:03
Tesla has not earned its cost of capital, except for one year.
“Tesla never earns his cost of capital”
Listen at 35:40
Elon Musk destroys capital through his business decisions.
“Elon Musk destroys capital”
Listen at 35:43
Overvalued AI narratives will eventually confront financing costs.
“eventually, the lies and the fantasies meet the cost of capital.”
Listen at 40:00
Tesla trades at roughly fourteen times revenue despite being ex-growth.
“14 times revenues for an ex-growth auto company.”
Listen at 41:27
Nvidia will eventually become an attractive short.
“I suspect in the fullness of time, Nvidia will prove to be a great short.”
Listen at 44:30
Micron’s current 85% gross margin is abnormal and unsustainable.
“85% gross margin for Micron, that's not normal.”
Listen at 45:24
Micron’s earnings will normalize over time.
“I think what's going to happen over time earners are going to normalize”
Listen at 46:31
Semiconductor monopoly profits will disappear within a few years.
“I'm highly convicted that these monopoly profits will not be with us within a couple of years”
Listen at 50:03
Gold prices will rise.
“the price of gold is going to go up.”
Listen at 55:30
Fiat currencies have recently lost value at an accelerating rate.
“all fiat currencies have collapsed at a particularly accelerated rate the last couple of years.”
Listen at 56:03
Interest rates will rise materially.
“The rates will start going meaningfully higher.”
Listen at 59:45
Gold protects wealth against central-bank policy mistakes.
“Gold, more than anything, is an insurance policy against the stupidity of central bankers.”
Listen at 1:02:09
Current market conditions favor selective stock picking.
“I think we're in the golden age of stock picking right now.”
Listen at 1:04:36
Energy stocks will substantially outperform the S&P 500.
“I think energy is going to absolutely run rings around the S&P, not even close.”
Listen at 1:08:44
AI infrastructure requires continuous spending to maintain competitiveness.
“What if I told you AI is shale? You got to keep spending.”
Listen at 1:16:54
Competitive AI spending dynamics make a major investment crash likely.
“So it almost guarantees you a car crash. It's a prisoner's dilemma type thing.”
Listen at 1:17:07
Crude prices will remain sustainably higher than semiconductor prices.
“I think the opposite is true with crude. I think energy prices are sustainably higher.”
Listen at 1:20:09
Temporary semiconductor demand leaves permanent debt and depreciation burdens.
“the demand is temporary, but the debt and the depreciation is permanent.”
Listen at 1:24:58
Temporary high prices create lasting depreciation and interest expenses.
“the prices are temporary, the depreciation and the interest expense are permanent.”
Listen at 1:27:02
CoreWeave and many neoclouds will be worthless absent a bailout.
“I'm highly confident that CoreWeave, as short of a bailout, is a zero, as are many of the neoclouds.”
Listen at 1:27:47
A major deterioration in AI-related markets is highly likely.
“I think the likelihood that you know what hits the fan is extremely high.”
Listen at 1:28:08
CoreWeave’s stock will be substantially lower before the cycle ends.
“before the cycle's over, CoreWeave is going to be a lot lower. A lot lower.”
Listen at 1:36:43
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.