E126: Big Tech blow-out, Powell’s recession warning, lab-grown meat, RFK Jr shakes up race & more
2023-04-28 spoken.md · speaker-labeled ▶ watch ← E125 all episodes E128 →
Every number here is replayed from score_events — the same ledger the pool ranks on. Decay is what the 111 ideas nobody mentioned gave up this week; it applies only when an episode is processed.
Tier crossings
conviction thresholds crossed by this episode — 65 / 45 / 15 · ideas born here show where they landed. why 65 isn't always green
Kill dates that landed since E125
0 hit · 1 partial · 1 miss — windows that closed after 2023-04-21 and up to 2023-04-28, auto-scored against price data and never hand-set. verdict · R · α
| idea | verdict | R | α | closed |
|---|---|---|---|---|
| 📈 Coinbase's regulated-first franchise makes it a great business | MISS | -56.3% | -54.1 | 2023-04-23 |
| 🪙 Decentralized collateral-backed stablecoins have proven peg-resilient | PARTIAL | +0.1% | +2.3 | 2023-04-23 |
Who moved the board
each voice's force on conviction this episode — supports and opposes, weighted exactly as the replay applied them · share = % of this episode's movement
What got argued (17 ideas)
ordered by how hard each idea moved · quotes are verbatim from the transcript, timestamps deep-link into the episode
Friedberg's cost-curve verdict on the WSJ lab-grown-meat post-mortem, from inside the sector. Leg one: plant-protein substitutes (Beyond, Impossible) lose because they cost more, taste worse and use saturated fat for mouthfeel that the American Heart Association has flagged - the category never wins unless it is both cheaper AND better. Leg two: cultivated/cellular meat is many orders of magnitude away from cost parity, gated by strain engineering, ~60 tunable process variables and a hard capacity ceiling (of ~100m litres of biomanufacturing capacity worldwide only ~1m is rentable, against ~200 synbio startups, so capacity cost is up ~4x); a Merck executive's read is that fetal-bovine-serum demand sells a billion dollars and then zero because nobody can make money at $500 a burger. Near-term the money is in the high-value recombinant proteins already falling (insulin, pepsin, rennet), not in meat, so the listed alt-protein cohort de-rates further before the technology arrives.
So I want to just get back to the unit economics. The cost per kilogram or the cost per gram of the protein, we are still many orders of magnitude away on cellular meat. So the problems you're laying out are really down-the-road problems of optimization. Right now, we've got more fundamental problems on how do you actually get this stuff to be cost competitive.
Jason's call off hands-on GPT-4 / OpenAI-plugin deployment inside his own firms: roughly 30% of what knowledge workers do is already automatable today, and the cost of knowledge work ends up at about 10 cents on the dollar - a 90% collapse, not a 10% trim - on a five-to-ten-quarter timeline, with remote-work salary normalisation across countries stacked on top. Chamath's version is the Harvey.ai case: legal knowledge work gets bid down (the n+1 firm charges $600 instead of $800 and still pays the tool), so knowledge work is massively deflationary while physical labour gets more expensive. Sacks and Friedberg take the other side - Sacks says the same 30% shows up as productivity and therefore wage growth rather than displacement, and Friedberg calls the cost-savings frame the Luddite argument that misses the new product categories the leverage unlocks.
So I just don't see a world where, on the one hand, physical labor will continue to be more expensive, they'll demand more and more money to do the job that they're asked to do, and the knowledge work will become increasingly more deflationary because so much of it will be automated by AI that those folks will charge less and less.
It's going up for sale. And they believe, according to the Wall Street Journal, that bids will come in at $60 million, an 80% decline. And we talked about this commercial real estate would have this moment. A lot of the banks, the smaller regional banks, own this debt.
It looks like Russia is certainly not losing the way they used to be. They've captured 90% of Bakhmut, which has been the most violent, bloody battle of the war. And Biden at this point has no strategy to bring that to an end. In fact, he's rejected multiple attempts at a peace deal. And so now it looks like it's the Chinese who are in the driver's seat, potentially putting together some sort of diplomatic settlement.
By the way, we said that the first organizations to use this, like the Canary and the Coal Mine would be the consulting organizations. And today when Harvey got announced, one of the things that right on the heels of that, PricewaterhouseCoopers announced like a billion dollar investment into AI.
But when you show that chart, Jason, what it means is it's just really hard to find people. And so the only way you're going to get people off their butt to go into work, to sit in a chair, to do a job that you need them to do is to pay them more. And in finding that, wages will have to go up.
Hundreds of millions of dollars, and in several cases, billions of dollars have gone into these systems, and it's very likely that these companies may need several more years and several billion dollars. We are going to get there. The technology is progressing. The rate of progress is a little slower, and it's a little more challenged, I think, than the first round of investors had hoped.
And we've talked a lot about the statistic that I've shared a bunch publicly on how 70 percent of publicly traded companies that have gone public since 2020 are trading below their total cash invested since founding, which should translate to an estimate that call it somewhere in the order of 70 percent of private companies are probably worth less than their preference stack.
Episode digest
written during extraction and stored in data/extractions/ep126.json — the auditable source of truth, including everything market-adjacent that did not earn a capture
Big-tech Q1 blow-out week (MSFT +9%, META +12%, AMZN +10% after hours, GOOGL a mixed beat) split the besties across the two opposed big-tech ideas rather than resolving them: Friedberg, Chamath, Sacks and Jason all reinforced the lean-opex/cost-cut thesis with META as the proof, while Chamath simultaneously table-pounded his own bearish call that we are 'well past peak big tech' and these are now ex-growth cash cows valued on financial engineering. Two reversals, both Friedberg's: he now says Google is 'leap years ahead of Microsoft' on models and could be a quick turnaround story, against his own E106 chat-disrupts-search call, and fifteen days before its kill date he walked back the E032 synbio Netscape moment ('slower and more challenged than the first round of investors had hoped'). Sacks refused to revise his recession forecast against the earnings beat ('I'm sticking by it'), leaning on Powell's own sub-1%-or-recession framing, and put SF office at 30%+ vacancy with a five-to-ten-year absorption; Chamath reinforced the deglobalization/wage-inflation chain (ex-China world order, nearshoring, terminal inflation 'just roughly higher'). Two new ideas: Jason's 90%-knowledge-work-cost-collapse call (opposed by Sacks on productivity grounds and brushed off by Friedberg as 'the Luddite argument'), and Friedberg's bearish alt-protein cost-curve verdict - cellular meat many orders of magnitude from parity, plant protein dead on price and taste, the money is in recombinant high-value proteins. Diarization CLEAN: Jason top talker, all four addressed-by-name tests resolved to the right next speaker, no time gaps; the bare 'Friedberg' label (no first name) is the normal diarization variant and was canonicalized.