Nov 11, 2026 · 55 min
Listen from 36:32
Listen at 36:32
AI Boom Risks Flooding Markets With Unused Compute
Too Much Compute: Why This Top Hedge Fund Manager Believes AI is Heading for a Massive Oversupply | David Orr
If AI infrastructure outpaces real-world demand, excess capacity could drive prices lower and reshape technology investment, labor, and economic priorities.
- 1AI infrastructure may become oversupplied before the physical economy can absorb its expanding compute capacity.
- 2Google, Amazon, and Microsoft offer relatively protected AI exposure through infrastructure contracts, even if individual AI businesses struggle.
- 3Automation could make many office tasks nearly free while shifting opportunity toward trades, design, construction, and tangible production.
Don't miss
Orr connects AI’s potential compute glut to a post-office economy in which automation makes white-collar work cheap and raises the value of tangible production.
The brief
David Orr argues that AI infrastructure is arriving faster than the physical economy can absorb it, creating a future glut that could force compute prices sharply lower.
The thesis complicates the market sell-off: Orr is not buying the dip because timing matters, and excess capacity could still arrive after another rally.
He favors Google, Amazon, and Microsoft as relatively protected AI exposure, arguing their infrastructure contracts may remain attractive even if other AI businesses falter.
The conversation moves from drug discovery’s physical bottlenecks to office work, where AI could make many tasks nearly free and push workers toward tangible production.
Orr’s Militia Capital is keeping roughly neutral market exposure, reflecting a view that the AI trade’s eventual outcome is consequential but not yet fully timed.
What was said on this episode
24 statements · 12 positive · 11 negative · 1 neutral
Google, Amazon, and Microsoft are David Orr’s preferred AI investments.
“My favorite AI play would just be along the hyperscalers like Google, Amazon, and Microsoft.”
Listen at 9:42
- David Orron Hyperscaler contracted projectsPositive11:04
Hyperscalers’ contracted projects generate roughly 35% returns on capital.
“the way they structured the contracts, basically they're like 35% return on capital projects.”
Listen at 11:04
- David Orron Hyperscalers’ AI capacity investmentsPositive11:18
Hyperscalers have limited downside from their contracted AI capacity investments.
“their downside is pretty small.”
Listen at 11:18
- David Orron Amazon, Google, MicrosoftPositive11:28
Amazon, Google, and Microsoft would lose little in a severe AI bust.
“Amazon, Google, Microsoft won't lose much money at all in the bad outcome.”
Listen at 11:28
- David Orron Hyperscaler capital investmentsPositive11:41
Hyperscaler capital investments could earn 35% returns if compute remains scarce.
“Then you have the 35% return on capital investments, which are incredible.”
Listen at 11:41
- David Orron Democratic gains in 2028Negative12:43
Democratic gains in 2028 could lead to higher corporate tax rates.
“if let's say that they do really well in 28 they could increase corporate tax rates for example”
Listen at 12:43
- David Orron Tax-related market declineNegative13:39
A tax-related market decline might reasonably be six or seven percent.
“maybe it should be more like six percent or seven”
Listen at 13:39
An Anthropic IPO would negatively affect market liquidity or sentiment.
“if the anthropic comes out, that's just another blow.”
Listen at 14:35
- David Orron Hyperscaler AI investmentsPositive20:58
If intelligence remains scarce, hyperscalers’ investments could earn 35% returns over the next decade.
“if AI, if intelligence is in tight supply for the next decade, then they're investing these hundreds of billions of dollars at 35%. return on capital uh that's incredible”
Listen at 20:58
- David Orron Everyday software productsNeutral22:22
Everyday products such as Windows, Instagram, and Netflix need little superintelligence.
“you don't need a ton of uh super intelligence for this”
Listen at 22:22
- David Orron AI-enabled medical researchNegative22:35
Physical experimentation constraints will limit AI-enabled medical research sooner than compute.
“the physical bottleneck is going to hit way faster”
Listen at 22:35
- David Orron AI-driven experimentsNegative23:53
Scaling AI-driven experiments requires many years to build physical laboratory capacity.
“actually physically building all of those labs and everything, I mean, that's just going to take many, many years.”
Listen at 23:53
- David Orron Compute pricesPositive25:43
Much smarter computers could restart scientific progress and drive compute prices sharply higher.
“maybe if the computer gets even way smarter than they ever were, maybe this will resume. In which case, yeah, the price of compute could just go crazy high.”
Listen at 25:43
- David Orron White-collar office work costsNegative27:29
White-collar office work costs will approach zero within five years.
“all of that stuff is going to go to zero in five years”
Listen at 27:29
- David Orron Automated white-collar office workNegative27:38
Automated white-collar office work will require almost no compute.
“it will use almost no compute”
Listen at 27:38
- David Orron Automated tax preparationPositive28:36
Consumers will capture most benefits from automated tax preparation.
“the only people who win are kind of the consumer.”
Listen at 28:36
- David Orron White-collar jobsNegative29:35
Most white-collar jobs will disappear.
“I think that most of those white collar jobs are going away.”
Listen at 29:35
- David Orron Wide-moat companies adopting AIPositive29:56
Wide-moat companies could become highly profitable after replacing most workers with AI.
“because they got to fire almost all their workers, they just kind of have this cash printer.”
Listen at 29:56
- David Orron White-collar jobsNegative30:20
David’s base case is that white-collar jobs will disappear.
“My base case is that all those jobs are going away, though.”
Listen at 30:20
Meta may lack a durable moat if its platform becomes purely media.
“if it's just only media then maybe they don't have any moat at all”
Listen at 45:53
Meta’s platform could become commoditized within five years.
“maybe in five more years, their platform is just a commodity.”
Listen at 46:05
- David Orron AI-generated social-media contentPositive46:21
AI could improve social-media content and increase user engagement.
“it gets more and more interesting because ai makes the content actually genuinely better”
Listen at 46:21
- David Orron Many-to-many social networksPositive47:19
Many-to-many social networks usually have wide economic moats.
“So usually that's a pretty wide moat business.”
Listen at 47:19
- David Orron AI music recommendation assistantsPositive50:51
AI assistants may eventually recommend music better by learning individual preferences.
“maybe those will start being able to recommend, you know, if you start going down rabbit holes of, hey, you know, can you create like a music playlist for me?”
Listen at 50:51
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.