Nov 2, 2026 · 42 min
AI spending faces a trillion-dollar reality check
Where Do You Hide From an AI Bubble?
The episode connects fragile AI economics with recession risks, persistent inflation and the portfolio choices investors face when governments and companies compete for capital.
- 1Current AI infrastructure spending may require an implausibly large share of US GDP to generate acceptable returns.
- 2An AI investment slowdown could damage earnings and trigger a sharp recession even without becoming a full financial crisis.
- 3Capital Gearing Trust favors inflation-linked bonds and real yields while avoiding equities, long-dated bonds and gold.
Don't miss
Clothier says AI services may need to reach close to 15% of US GDP by 2030 for today’s infrastructure investment to earn acceptable returns.
The brief
Chris Clothier argues that the AI infrastructure boom rests on demanding assumptions: roughly a trillion dollars of annual spending, short-lived hardware and revenue growth sufficient to cover depreciation and financing costs.
The central test is stark: by 2030, AI services may need to absorb close to 15% of US GDP to justify current investment, leaving spending vulnerable to cheaper models and disruption.
As borrowing costs rise, cash-flow pressure is spreading from hyperscalers to neocloud providers and highly leveraged companies, raising the prospect of an abrupt slowdown and nasty recession.
The inflation outlook complicates the AI bear case: energy shocks, war, deglobalization, wage growth and defense spending could keep prices rising even as technology investment falters.
Clothier explains why Capital Gearing Trust prefers US TIPS, UK index-linked bonds and selective real assets while avoiding equities, long-dated bonds and gold.
What was said on this episode
30 statements · 9 positive · 19 negative · 2 neutral
AI services would require nearly 15% of US GDP by 2030 to justify investment
“by 2030, users will have to be spending close to 15% of US GDP on AI services. And that seems completely implausible”
Listen at 2:55
Hyperscalers need $2.75 trillion annual revenue by 2030
“you need revenues of $2.75 trillion a year in 2030”
Listen at 7:35
AI labs would need $5.5 trillion revenue at a 50% gross margin
“at a 50% gross margin, well, that points to $5.5 trillion of revenue”
Listen at 8:28
SLMs, Chinese models, or open-weight models could disrupt AI infrastructure spending
“what if all of this spend gets swiped off at the knees? by the SLMs or indeed the Chinese models or the lower cost open weight models”
Listen at 10:51
AI infrastructure spending is unlikely to become systemic for hyperscalers
“it's unlikely that all of this spend is likely to be systemic for the hyperscalers themselves”
Listen at 11:50
“I think it's highly unlikely that at that cost of capital that it can make any sense for them to continue investing.”
Listen at 12:44
Established AI-related businesses are beginning to show financial stress
“more established businesses are showing some signs of stress”
Listen at 13:16
Limited credit availability could slow AI infrastructure investment
“it is possible that if the availability of credit just isn't there, then all of this does slow down”
Listen at 15:12
A credit-driven AI slowdown would have relatively limited fallout
“the fallout would be relatively... limited”
Listen at 15:28
Unlimited AI investment appears economically irrational
“this sort of runway of kind of unlimited investment that strikes me as being kind of economically insane, frankly.”
Listen at 15:55
An AI-related slowdown could produce a severe recession
“it's entirely possible that we have a nasty recession”
Listen at 16:57
A credit-driven AI slowdown would significantly hurt chip companies
“any major slowdown as a result of availability of credit would hit them”
Listen at 17:26
A sharp AI-boom slowdown would reduce inflation
“a sharp slowdown in the AI boom would definitely reduce inflation”
Listen at 19:40
Global inflation is expected to persist over the medium and long term
“our core case remains medium, longer term inflation globally”
Listen at 19:45
Defense spending does not increase economic productive capacity
“the point about defence spending is that it is... economically non-productive”
Listen at 21:09
Gold had begun behaving like a speculative asset
“we saw that gold was starting to behave like a speculative asset”
Listen at 23:33
Selected infrastructure investments may return about 9% nominally annually
“we think that these things offer nominal returns of the order of 9% per annum”
Listen at 24:23
Selected infrastructure investments may deliver 5–6% real returns
“they're going to deliver something like five to six percent real”
Listen at 24:32
Long-dated bonds are in a secular bear market until fiscal conditions change
“we think we're in a secular bear market for long-dated bonds until this changes”
Listen at 26:39
No UK budget action would make 30-year gilts attractive
“there's nothing that the Chancellor could do that would make me want to own 30-year gilts”
Listen at 27:17
Global spending splurges are creating genuine competition for capital
“the competition for capital is real”
Listen at 29:54
The global spending surge could abruptly end
“this could all come to a shuddering halt”
Listen at 31:00
Real government-bond yields are attractive relative to history
“Real yields in government bond markets are fundamentally very attractive compared to history”
Listen at 31:20
Five- to six-year nominal and real bonds are attractive
“We are very constructive on the yield curve, both nominal and real, going out about as far as five, six years.”
Listen at 31:48
Investors can comfortably own five-year nominal or index-linked gilts
“you can own, say, five-year nominal or index-linked gilts with a huge amount of comfort”
Listen at 32:19
Chris Clothier would buy gold at 3.5
“Three and a half.”
Listen at 33:03
Central banks are unlikely to abandon gold
“I see no reason why central banks will give up on gold”
Listen at 33:52
Western fiscal consolidation would initially cause a severe slowdown
“if you saw that Western governments really trying to get a grip, of their fiscal situation, well, of course, that's going to start with a nasty slowdown”
Listen at 34:28
A halt in government spending would severely hurt equity markets
“if it does come to a crashing halt, then that's very nasty for equity markets”
Listen at 35:11
Babylonian sesame-seed prices rose tenfold over several years
“I think sesame seeds went up tenfold over a short few years”
Listen at 37:17
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.