Summary

The episode is a full teardown of Catrini’s viral “2028 Global Intelligence Crisis” piece: AI agents strip out the friction a services economy monetizes — the 2.5–3% card interchange, 5–7% realtor fees, the DoorDash home-screen moat — so $150–500k SaaS workers become $45k Uber drivers, prime borrowers with a 780 credit score default, and a credit crisis starts in private credit before metastasizing into insurance. Avi’s zinger: a single GPU cluster in North Dakota doing the work of 10,000 white-collar workers in Midtown Manhattan is “more economic pandemic than economic panacea.”
Both hosts call outright collapse “very unlikely,” but for different reasons. Avi’s is timing: the doomers “talk out of both sides of their mouth” — 84% of people have never used an AI application by the doomers’ own research, yet every consumer is supposedly agent-native by end-2028; “in 5 to 10 years absolutely,” but not in eighteen months. Avi says the response will be monetary — the Fed and Treasury “blasting money into the economy with a fire hose,” and “the helicopter money will come. That we can be sure of.”
Avi’s honest change of mind is the emotional spine: he used to dismiss AI doomers, but now “I’m struggling to debate the likely Luddites at this point” — because the likely Luddites were always wrong only while human intellect was the bottleneck that refilled displaced jobs, and that bottleneck is now itself being automated.
The optimist rebuttal fails, per Jonah, on one massive logical hole: it claims displaced workers won’t stay displaced because business-formation costs fell 70–80% in 18 months — but “nobody’s going to buy your freaking quilts because they suck.” AI-assisted intelligence is still a bottleneck; there will always be a ranking of best to worst, so a billion laid-off middle managers cannot all succeed simultaneously.
The one point of total agreement is the tradeable core: AI is deflationary, and “the moment that we see deflation in this economy rates are going to zero” with an instant refi wave. The menu: levered TLT, gold, Bitcoin as “the best passive hedge” for the fire-hose scenario, short human-output businesses (consulting-like IBM), long inflation-linked cash flows — down to Avi’s Miramax example, where Pulp Fiction rentals keep paying.
The regime call is “volatility is the word of the day”: hedge funds have compressed from 2–5-year theses to 3–6 months, which is why one article moved markets — and retail’s edge is speed, “it’s going to take them three to four days to get all of their positions on… it’s going to take you about 3 to 4 seconds.” Own anti-fragile, long-optionality assets: trading businesses, utilities, Poly Market stock, Hyperliquid tokens.
Concrete pair trade: long Apple, short IBM. Apple fell 1.5% on a sell-off whose source article says compute goes to the edge (token usage per person toward 10,000,000 a day, models running in your pocket) — “this is like Amazon stock selling off because there’s a tech crisis in 2000,” with ~50% upside into the article’s own mid-2027 melt-up. IBM, down 7.5% peak-to-trough after Anthropic said cloud code can automate Cobalt modernization, is the flipped adage: now you do get fired for choosing IBM.

Deep dive

1. Catrini’s doom loop: AI eats the friction the US economy is built on

The last 24 hours, per the hosts, were “entirely consumed” by Catrini’s article, which Avi compresses to one sentence — “AI is coming to eat us all” — but whose actual mechanism is that friction drives the services economy: agents kill the 2.5–3% card interchange, cancel unused subscriptions, and replace the 5–7% real-estate commission with an agent that “downloads the entire MLS database” for 1%. DoorDash’s only moat is being “an app that exists on your home screen”; once everyone vibe-codes a DoorDash equivalent, agents pick the cheapest and brand is abstracted.
The corporate math, as Jonah relays it: a company spending $95M on people and $5M on tech spends $70M on people and $20M on AI — $90M instead of $100M, a deflationary crisis. SaaS employees earning $150–500k become Uber drivers at $45k — except Uber also gets squeezed by an “agentic ride share finder” pinging a hundred cheaper marketplaces.
Downstream: wealth accrues to a few owners of AI production, a credit crisis starts in private credit and “metastasizes into the sources of permanent capital like insurance and reinsurance” — and unlike 2008’s day-one-default subprime, this time it’s prime borrowers: the 780 credit score, $500k-a-year San Francisco SaaS worker whose income stops recurring. Jonah’s deadpan coda: “then aliens invade the earth and we all die.”
Why it went viral, in Avi’s telling: it takes the likely Luddite argument and frames it in relatable stories with “incredible zingers” — the US built “a giant rent extraction layer on top of human limitations,” and the killer line: a single GPU cluster in North Dakota doing the output of 10,000 Midtown Manhattan white-collar workers is “more economic pandemic than economic panacea… how much money machines spend on discretionary goods. Hint, it’s zero.”

2. Avi’s pushback: the cycle always closes — and the timeline fails the sniff test

Avi’s history lesson, as told: the Industrial Revolution’s feared unemployment spike barely happened in aggregate, and the car completely decimated the horse industry — “basically everybody lost their jobs there” — but road construction boomed and roadside restaurants “popped up like wildfire.” “People always figure it out at the end of the day”; claiming this time is “so wildly different is kind of tough.”
His honest concession, kept as hedged: because “this is the first time that we’ve actually automated human thought and not just human physicality… maybe it does take a little bit longer” to figure this out — and against a near-term unemployment wave, “I don’t know if I can really argue super hard against that.” His only firm objection is speed.
The sniff test: the doomers “talk out of both sides of their mouth” — claiming we’re early while 84% of people have never used an AI application “according to their own research,” yet everyone transacts through agents within two years? “In 5 to 10 years, absolutely” — but not by end-2028, when so much discretionary spending sits with older consumers who won’t change behavior in a year. (The hosts’ opening gag makes the same point: xAI’s creator can’t even get a Twitter livestream working — “I don’t know how afraid we’re supposed to be of these AI engineers.”)

3. Avi can’t debate the likely Luddites anymore

Avi’s change of mind, worth keeping verbatim: he was dismissive of doomers at first, but “you do have to ask yourself, is this time different” — invoking Taleb’s turkey that “thinks everything is fine the day before Thanksgiving.” The likely Luddites were always wrong because human intellect was the bottleneck that refilled abstracted-away physical jobs with knowledge work; “now that knowledge work is being abstracted away by code… I’m struggling to debate the likely Luddites at this point.”
His answer to the behavior-change objection: consumers won’t do GitHub commits — they’ll download a shiny app that’s agentic on the back end, querying Uber, Lyft and “63 others” and guaranteeing a cheaper price. Avi’s clean counter: “Lyft is almost always cheaper than Uber, but people still use Uber.”
Where Catrini gets it wrong, per Avi: the policy response will be monetary, not political. “The US Treasury and the Federal Reserve will gang up and start blasting money into the economy with a fire hose… they’re not just going to sit idly by while Rome burns.” Bitcoin is “the best passive hedge for that scenario” — hedged exactly as spoken: BTC may go down first while people are getting fired and living in tent camps.

4. The rebuttal’s fatal hole: no room for a billion quilt entrepreneurs

The rebuttal concedes the displacement but attacks one assumption — “that displaced white collar workers would remain displaced” — arguing the same AI tools cut the cost of launching a business by 70–80% in 18 months and business formation exploded. Jonah “wanted it to be right” and couldn’t get there.
The part that holds water: when realtor fees drop from 5–7% to basis points, the savings don’t sit under a mattress — “every penny saved goes straight back into consumption,” furniture, finishings, a nicer house.
The hole: “I don’t care how easy it is for AI to help you pursue your passion of quilting after you got laid off from Meta. Nobody’s going to buy your freaking quilts because they suck.” AI-assisted intelligence is still a bottleneck — “there will always be a best product,” a ranking of best to worst — so mass simultaneous solo-entrepreneur success is “fooar.” His conclusion: “I think we’re in for some real turbulence.”

5. Avi’s real disagreement: there is still a cycle

Avi is explicit that he agrees with Catrini on radical transformation, massive tech layoffs and cost drawdowns; what he rejects is the claim of “no cycle” — that “the stabilizing cycle of capitalism that has happened over the last 350 years is a doom loop because we finally automated human intelligence.”
His mechanism: displaced workers flow to whatever AI can’t cheapen — physical delivery, food, furniture, art, government — increasing labor supply there and lowering those costs too, while AI lowers costs everywhere it touches. And if AI literally touches everything: “if you end up taking an 80% pay cut, but now costs of everything are down 70%, that isn’t as catastrophic as people might think.”
Jonah’s counter, undiluted: that’s “a battle I don’t think humanity can win” — if AI eats a third of the economy in two years, another third two years later, “people just can’t adapt and move fast enough… they’ll just be running away from Skynet, and that’s not a recipe for economic health.”

6. The point of full agreement: deflation sends rates to zero

Deflation screws borrowers: you don’t care that the dollar store is cheap again — “you care about your mortgage payment that you can’t afford because your salary went down.” So the government must print to counteract it.
Avi’s categorical call: “the moment that we see deflation in this economy rates are going to zero” — immediately, with everyone refinancing. Both hosts stress this holds even if the economy doesn’t collapse; Avi doesn’t expect serious deflation to stick because “the second there’s a sign of deflation the Fed is going to be hyper-reactive” — so the end state is more debasement and inflation.
The zero-rate menu: the “obvious” bonds trade — levered, since spot TLT “doesn’t give enough juice for the degenerate listeners” — plus crypto, gold, and equities framed as short human-output businesses, long inflation-linked cash flows. Avi’s example as told: his buddy runs Miramax, now a private-equity asset — “Pulp Fiction, people keep renting it” — plus, if you could, garnishing sports players’ wages.

7. Jonah’s energy stack and Avi’s volatility regime

Jonah’s slower-AI reasoning doubles as his bull case: physical constraints on data centers, chips and above all energy cap the pace. He’s “very very very bullish on energy” — a utilities basket (not one region: “you can get totally nuked” à la California), “your Exxons… your Chevrons,” uranium (spoken “URRA,” likely URA) on reactors restarting within 5 years, and Constellation Energy. At zero rates, utilities’ 3–4% dividends plus data-center demand get crowded into; energy trading businesses “are going to absolutely clean up.”
Avi’s frame is path-dependence: the Catrini thesis starts playing out, then the fire hose arrives — so “volatility is the word of the day.” Own anti-fragile, long-optionality assets at cheap levels: trading businesses, utilities, certain private-equity plays, “Poly Market stock or Hyperliquid tokens,” and Bitcoin as “an anti-fragile long-vol asset you can just be long by buying some tokens.” Start buying options “in a way where you can survive and not bleed out.” Also on the list: XLE, gold, and coming entries into European defense stocks, “a mega trend.”

8. Long Apple, short IBM

Apple fell 1.5% in a Catrini-led sell-off whose own document argues compute goes to the edge — token usage per person from zero to “10,000,000 tokens a day, whatever the article says,” models running “inside of your pocket pretty soon.” Jonah, “kind of all in”: the number-one luxury-device maker in history is the prime beneficiary, being puked by momentum-trading PMs with a few-weeks horizon — “this is like Amazon stock selling off because there’s a tech crisis in 2000.” Ride it into the article’s own mid-2027 melt-up, “probably up 50% with very little downside.” Avi’s additions: Apple is the only tech stock working (Google 10% off highs), and “Samsung too.”
The short leg arrived live on air: Anthropic announced cloud code can automate Cobalt modernization — the legacy language behind financial back-ends that IBM makes a consulting business updating — 24 minutes before recording. IBM: down 7.5% peak-to-trough. “Nobody ever got fired for choosing IBM… that’s over” — now you get fired for paying IBM millions for what you could do yourself. The trade: on an IBM bounce, “get really long Apple and really short IBM.”

9. The meta-read: a viral bear piece is a bullish tell

Avi’s psychology point: every stock Catrini named is down, Bloomberg is citing the article for the crash — yet “bull markets don’t end until nobody’s trying to call the top… this getting 20 million views in and of itself is bullish.” The article itself even calls for its sequencing: melt-up before melt-down. And it “could not have come two months ago when everything was at the highs” — it moved markets only because PMs “don’t have strong thesis right now.”
The structural claim: hedge-fund horizons have compressed from 2–5 years to 3–6 months, which is retail’s edge — “it’s going to take them three to four days to get all of their positions on. It’s going to take you, the retail guy, about 3 to 4 seconds.” Front-run or ride them; e.g., be early to the long-bond trade before every fund chases deflation.
Jonah’s closing modification of “humans always figure it out,” kept as said: “some people figure it out. Others blame the Jews and their society implodes” — America probably figures it out, but “India’s outsourced IT service economy may indeed tank,” and places that adopt AI badly may “revert to hardline communism.” His prescription: take risk now, because live P&L is “your best telemetry” into whether you’re winning or “about to have the rug pulled.” Avi’s warning label, via South Park: “phase one collect underpants, phase two question mark, phase three profit… this is not going to happen.”

Investing in Biotech with Verdad Capital Inside Dan Sundheim's Bets on Anthropic, OpenAI, and SpaceX