Oct 27, 2026 · 1h 3m
Cheap intelligence shifts value toward physical moats
Cliff Sosin - Owning the Complements to Intelligence - [Invest Like the Best, EP.497]
The conversation tests whether AI’s biggest winners will be intelligence suppliers or businesses whose scale, density, and physical constraints make intelligence useful.
- 1Intelligence faces limits from complexity, human attention, physical systems, and demand even as models become cheaper and more capable.
- 2Businesses with local density, scale, and low-cost operations may capture more value than companies selling increasingly commoditized intelligence.
- 3AI could reshape investing and education, but human judgment remains important where experience, incentives, and information sit beyond the internet.
Don't miss
Sosin’s sharpest distinction is between enthusiasm for AI’s usefulness and the belief that every physical process can become infinitely fast.
The brief
Cliff Sosin rejects the idea that intelligence is an unlimited scalar racing toward godlike capability. Complex systems, human attention, physical constraints, and demand impose limits that better models cannot simply erase.
The conversation shifts from AI suppliers to its complements: businesses with scale, local density, and low-cost operations may gain more as agents optimize decisions and make physical services easier to use.
Nvidia and TSMC anchor a discussion of infrastructure and physical moats, while frontier-model economics remain uncertain if intelligence becomes cheaper faster than demand expands.
Sosin expects humans and models to work together before AI takes over longer-horizon investing, because experience and information outside the internet still matter.
The closing argument reaches education: students should use models on ambitious problems but still understand, defend, and own the work they produce.
What was said on this episode
22 statements · 10 positive · 6 negative · 2 mixed · 4 neutral
Everyone will have highly intelligent agents working for them.
“We're all going to have a team of 200 IQ people working full time in our pocket.”
Listen at 6:48
Investors should own assets complementing ubiquitous intelligence.
“What you want to own are the complements to intelligence.”
Listen at 8:24
Meta’s core business is personalized television.
“the core business at Meta is personalized TV.”
Listen at 9:32
Intelligence costs will fall roughly 100-fold within five years.
“the price of intelligence is falling by a factor of 100 or something over the next five years”
Listen at 10:40
Mature AI applications behind the frontier will commoditize.
“Those areas, they clearly commoditize.”
Listen at 14:25
Very few use cases will require super-elite intelligence.
“what we end up with is some version of there are very few use cases for super elite intelligence.”
Listen at 16:11
Very good AI intelligence will have many uses but become commoditized.
“There are lots of use cases for very good intelligence. That very good intelligence is highly competitive and commoditized.”
Listen at 16:16
A market’s size eventually falls to zero when its price reaches zero.
“eventually, of course, if you take the price to zero, then the size of the market goes to zero.”
Listen at 17:51
AI demand will be highly price-sensitive.
“AI demand is going to be really sensitive to the price”
Listen at 25:30
Price sensitivity will limit consumers’ total AI spending.
“which intrinsically, I think, does limit in some sense the amount in total consumers are going to be willing to spend on this.”
Listen at 25:32
AI infrastructure is a massive build-out for an important technology.
“This is a huge infrastructure build out of a huge and important technology.”
Listen at 26:20
TSMC’s physical, tacit processes are difficult for AI to replicate.
“much less amenable to an AI just because it is so physical and because it's all these little things, many of which aren't even legible to an AI.”
Listen at 29:41
AI agents will shift consumers toward low-cost providers.
“So the first order effect is that you move to the low cost provider.”
Listen at 32:03
Agents may give low-cost providers more volume but lower margins.
“That actually creates, in some sense, a better business where the low-cost guy wins a bunch more volume at much lower margins.”
Listen at 32:45
Investment professionals will eventually be displaced by competition from AI.
“in the long term, we'll probably get competed out.”
Listen at 33:12
Human investors and AI models will initially make better judgments together.
“The ability of me to use models to assist me and then to make judgments will make better judgments working together.”
Listen at 33:59
Companies should give everyone opportunities to participate in AI hackathons.
“everybody needs to have hackathons.”
Listen at 43:07
CEOs should enable agents to access company systems read-only.
“I want my GrokBot, Metamuse, agent, whatever, to be able to access every system in the company on a read-only basis.”
Listen at 43:34
Current conditions make this an exceptionally favorable time to start companies.
“it's the best time to start a company until tomorrow.”
Listen at 45:13
Easily achievable AI capabilities will quickly become widely available.
“if you can easily do something, then so can everybody else.”
Listen at 45:31
AI will enter investing first at short horizons, then longer horizons.
“I think it's going to start at shorter time scales and work its way up to longer time scales.”
Listen at 54:05
Carvana’s differentiated infrastructure and brand limit severe downside scenarios.
“Carvana's infrastructure and brand and everything else is just so differentiated that there are few scenarios where it works out really badly.”
Listen at 55:42
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.