Summary

The LA fires are less an act of God than a documented failure to act, in Chamath’s telling: California’s timber harvest fell from ~6 billion board feet a year in the 1950s to ~1.5 billion, leaving 163 million dead trees that CEQA and vetoed bills (AB 2330, AB 1951, AB 2639, SB 103 on burying power lines) limited the ability to clear — after the same terrain burned in 2018. “This is the ultimate expression of negligence and incompetence.”
Friedberg flags insurance as the real economic crater: catastrophe-model loss probabilities can shift from ~1-in-1,000 toward 1-in-20 in fire zones, but state insurance commissioners won’t approve the rates, so carriers exit — State Farm canceled 1,600 Palisades policies months before the fire. The state’s fallback FAIR Plan holds ~$220M of capital plus ~$5B reinsurance against ~$6B of exposure in Pacific Palisades alone. One of three parties eats the loss — homeowner, insurer, or taxpayer — and the panel’s unanimous answer: “Taxpayer.”
The rebuild math doesn’t pencil: insurance reimbursement caps near $3M against a Palisades median of $7-8M, rebuild costs of ~$1,000/sq ft mean ~$7M cash (double it pre-tax), and the California Coastal Commission — voter-created in 1976 with authority “that exceeds legislative action” — can slow-roll beachfront permits for decades. Expected endgame: developers buy lots at 80% off and sit on them, and the state may seek a federal bailout.
The political call: this is a “crisis of competence”, not a DEI story — fire chief Kristin Crowley scored top-50 of 16,000 on her exam, and Chamath invokes Munger (“show me the incentive and I’ll show you the outcome”) to blame institutional mandates, not her. The fix demands wholesale replacement of the State Assembly, State Senate, and ballot-level rescission of three decades of bad law; Jason’s version: recall Newsom and Bass, “vote for executives” — Caruso, Bloomberg, Trump — regardless of ideology.
On Zuck killing fact-checkers for community notes, Chamath’s frame is that it’s the value-maximizing move, not a moral one: Zuck censored when Obama/Biden winds blew that way and flipped when Trump won — only Elon “was willing to torch $44 billion” for principle. Jason dissents hard: “There’s no human being who has censored more humans than him,” calling the oversight board a $150M punt and the Dana White board seat appeasement. Friedberg’s tiebreaker question — would Zuck have done this if Kamala won? “Probably not.”
Nvidia’s CES splits the panel: Chamath is “a little confused” by the $3,000 Project Digits box — if training is hitting limits and the market shifts to inference, why beat cloud accelerators? — while Friedberg sees the edge-compute bet: warehouse robots, machine-vision, and genome models can’t wait on a round-trip to the cloud, and $3K of compute “knocks a lot of Macs out of the field.” Both like the robotics test bench; Chamath repeats that Waymo and Tesla run away with autonomy, forcing auto OEM consolidation.
Cyan’s tradeable ideas: she’s bullish NVDA (“early adopters cannot use Claude without getting shut down because of scaling issues”), thinks Tesla should buy Uber (~10-15% of Tesla’s market cap — WhatsApp-scale), and says AI makes seed investing structurally harder — 30 teams per idea, forcing index-the-category bets — so she sits out hype cycles and owns the picks and shovels: “power and compute, lithography.”
Trump’s Greenland play is read as genuine strategy, not trolling: a melting Arctic opens the Northern Passage, and a Canada-Greenland arrangement gives the US “monopoly control over something that could become as important as the Panama Canal” while China and Russia militarize. Cyan goes further — “I believe either end” of the Panama Canal is operated and controlled by China, and “we are at war with China, whether we like to admit it or not, in my opinion.”

Deep dive

1. LA burned on schedule — the data was “in the realm of the knowable”

Friedberg’s both/and framing: you can have “incompetent planning and execution by leadership” and uncontrollable natural forces. Southern California’s rain season sits at essentially 0% of normal precipitation, and with 100-mph Santa Ana winds, “no matter how much underbrush you clear out… that is gonna create a fire hurricane.” Clearing should have happened — “that was bad policy” — but it wouldn’t have stopped everything.
Chamath rejects the wind excuse outright: the same terrain burned apocalyptically in 2018 (the Sepulveda Pass video), outlier weather events are modeled and accelerating, and insurers dumped SoCal fire coverage months before ignition. “All this data was in the realm of the knowable.” His evidence chain: timber harvest fell from ~6 billion board feet a year in the 1950s to ~1.5 billion, leaving 163 million dead trees, while CEQA and rejected bills — AB 2330, AB 1951, AB 2639 vetoed or blocked, SB 103 on burying power lines never reaching the governor’s desk — made clearing them nearly impossible. “There’s a level of incompetence bordering on criminal negligence.”
His budget counterfactual: California spent roughly $21 billion on homelessness and he suspects tens of billions on illegal immigrants — reappropriate that to protective infrastructure and “maybe there still would’ve been a fire… but it’s hard for me to believe it would’ve been as bad as it is right now.”
The disaster literally interrupted the taping: Friedberg left mid-show as the Kenneth Fire reached his childhood home and his 70-something parents evacuated. “I’m like, grab all the photos. Grab all the albums” — while his mom reached for jewelry. Jason’s aside: “We’re the last generation that will be thinking about this issue of grabbing the photos.”

2. The DEI pile-on misses — this is incentives, not identity

The mob went after fire chief Kristin Crowley, but Jason’s own research undercuts it: she scored top 50 out of 16,000 on the firefighter exam and climbed every rung from paramedic to deputy chief. His summary line: “We can confirm lesbians didn’t cause the Santa Ana winds.”
Chamath’s structural read — worth keeping: “It’s not her fault that she was selected. The real question is, what was she mandated to focus on once she got the job?” DEI becomes an institutional directive that “guides your compensation, guides your recognition, and so you do it” — Munger’s law: “Show me the incentive and I’ll show you the outcome.” If allowed to just fight fires, “she would probably do a good job.”
Friedberg draws the clean principle: “I don’t think that the mission of any public service organization should be to meet DEI metrics… The best ability to serve the public should be the objective, and that’s it.” He’s careful on the budget question — Crowley claims she fought the cut Bass imposed, though Chamath notes a counter-narrative that the department actually got an extra 50 (million). Breaking-news fog, flagged as such.
Cyan’s version: “Diversity is good unless that’s all you have” — the cost isn’t who gets hired, it’s bureaucratic attention. “You can be as diverse as you wanna be and not be able to put out a fire, and then it just really doesn’t matter.”

3. Insurance is where the economic devastation actually lives

Friedberg, who worked in the business, walks the mechanism: catastrophe models (RMS, Eqecat, now in-house and startup models) show total-loss probability in these zones shifting from ~1-in-1,000 years toward 1-in-20. At 1-in-1,000, a $1M home costs ~$10K/year to insure; at 1-in-20, the fair price is 5% of home value — “if you have a $10 million home, are you gonna pay $500,000 a year for property insurance? No.”
The regulatory bind: 50 state insurance commissioners must keep carriers solvent, keep rates down, and keep insurance accessible — three goals that can’t coexist when risk repriced. When the DOI refuses the rate increase, “the carrier’s like, ‘Okay, we got no choice,’ and they exit.” State Farm canceled 1,600 policies in the exact neighborhood that just burnt down, three-to-six months before the fire — same pattern as Tahoe and post-Santa Rosa wine country.
The state’s patch is itself insolvent: the FAIR Plan holds about $220M of capital plus ~$5B of reinsurance against ~$6B of exposure in Pacific Palisades alone. By suppressing insurance costs, the DOI props up real estate values and property-tax revenue — “at the end of the day, the bill is gonna come due.” Someone eats the repriced risk: homeowner, insurer, or taxpayer. The panel’s answer in unison: taxpayer, until voters “say enough is enough” and force legislative change — “I think we’re getting pretty close.”
Sacks’s free-market exhibit is his portfolio company Kin Insurance, writing direct-to-consumer policies in disaster-prone Florida and Texas by pricing off construction quality and weather models — “things that are not allowed in California, if you can believe it or not.” Harden your home, premium drops; California bans the feedback loop.

4. The rebuild math doesn’t pencil — and the Coastal Commission holds the pen

Chamath’s arithmetic: insurance reimbursement caps around $3M against a Palisades average of $4.5M and median closer to $7-8M — “folks will get less than half their home value back.” Rebuilding 7,000 sq ft at ~$1,000/sq ft means ~$7M of post-tax cash — “you might as well double it” pre-tax in California. Individual homeowners can’t do it; the conclusion on the show is the state ends up asking Washington for a bailout.
The permitting gauntlet may be worse than the money: the California Coastal Commission, created by voters in 1976 with authority that “exceeds legislative action,” takes “two decades, three decades sometimes” per approval. Jason’s dark scenario: political appointees slow-roll 50-70-year-old owners on PCH until “nature returned it to its natural state.”
Friedberg’s prediction of who actually wins: “I could see developers coming in and being like, ‘Dude, if I could buy all these lots for 80% off, I will’” — and then sitting on them while memory fades, as happened after 1962. Adam Carolla’s test, via Chamath: the real reckoning is whether displaced families — “the most single and only financially securing asset that they have,” times 120,000-200,000 families — metabolize a three-year permit wait into political change.

5. “A crisis of competence” — the recall-everything segment

The Karen Bass airport clip — refusing every question about her absence and the fire budget cut — sets Jason off: “You took the leadership job. I don’t give a if you’re in shock… I wouldn’t trust her literally to pick up my lunch.” His prescription is ideology-blind executive competence: “Whether it’s Bloomberg, whether it’s Trump, whether it’s Caruso, it doesn’t matter their ideology. It matters their effectiveness.”
Chamath’s is systemic: “wholesale replacement of the people that govern the state of California” — specifically the State Assembly, the State Senate, and ballot measures to rescind three decades of bad law. The world’s fifth-largest economy is run like “a weirdly disabled third-world country type operation with the wealthiest resources on planet Earth… We, as a populace in this state, need a reset. Otherwise, we deserve what we get.”
Cyan names the disease “woke imperialism, like a religion” and calls on Democrats to reclaim their party from it — “we’re not building some railway that’s never being built. We’re not solving homelessness with billions and billions of dollars.” Her credential: she was “signature number one” on the Chesa Boudin recall, weathered the harassment, and “I’ve never been happier to do something and get civically engaged.”
Her mea culpa for tech is worth keeping: for twenty years the industry outsourced politics — Ron Conway was the only person she remembers getting involved — because “they didn’t get involved until it affected them, kinda like the houses burning down.” Jason’s framing for residents: at 50% tax “you’re a shareholder of an organization known as California Inc…. Recall these people.”

6. Zuck’s flip: value-maximizing compliance, not conversion

The news: Meta fired third-party fact checkers for X-style community notes and is moving trust and safety from California to Texas. Zuck’s own words: complex moderation systems built “partly in response to societal and political pressures… This approach has gone too far.”
Chamath’s amoral read — the episode’s sharpest frame: under Obama and Biden, censorship “was the value-maximizing function for Facebook shareholders in that time”; now the winds reversed and so did Zuck. “Whether he morally agreed with it or not, almost didn’t matter.” The contrast: “Elon didn’t make a value-maximizing function. He made a moral decision… He was willing to torch $44 billion when he bought Twitter” — an outlier standard no other CEO should be conflated with.
The coercion receipts run both directions: Trump’s book describes Zuck “always plotting to install shameful lock boxes in a true plot against the president” and warns “he will spend the rest of his life in prison” if he cheats in 2024 — and asked at a press conference whether Zuck’s move responded to those threats, Trump answered “Yeah, probably.” Friedberg’s litmus test to the panel: if Kamala had won, would Zuck have done this or put Dana White on the board? “Probably not.”
Cyan’s more generous read: Zuck’s original Facebook profile listed him Libertarian, free speech was always “deep in his heart,” and the MMA fighting genuinely changed him — “I think he learned to stop acquiescing.” She predicts post-January 20 disclosures will reveal “just how much the government coerced him,” and he’s getting ahead of it. Her caveat: community notes can be gamed (she cites a report of Kamala-campaign manipulation on X), and “he’s gonna have to deal with advertisers next” — a problem X no longer has.

7. Jason’s dissent — and the Google China fact-check

Jason refuses the absolution: “His action was to be the greatest censor in the history of humanity. There’s no human being who has censored more humans than him.” The bill of particulars: a two-year ban on a sitting president, then a $150M oversight board created to punt the decision — “he’s so spineless” — despite holding god-level voting shares. “What is the point of being a billionaire… if you don’t get to say, I’m gonna do what I want?”
Friedberg pushes back hard: Zuck was dragged before Congress by officials who could imprison him, with directives amounting to “do this or you will be prosecuted” — “do you not think that there’s some degree of inherent complicit role that certain government officials had?” Chamath adds: “Not to mention a violation of our Constitution.” Jason doesn’t budge: “He could have just hired lawyers and fought it. He didn’t put up any fight.”
The exchange produces a real correction: Jason claims Sergey Brin killed Google’s censored China product on moral grounds; Friedberg corrects him — censored Google.cn ran live from 2006 to 2010, and the exit owed as much to failure against Baidu (which Google had separately invested in, fairly well) and a server hack as to philosophy. Disagreement preserved, Friedberg wins on the record.
Chamath’s closer defuses it: “However we got here, we’re here. And we should all be happy that we’re here” — the panel unanimously takes the win for speech, with Jason’s caveat standing: “if Kamala Harris had been voted in, he would double down on censorship.”

8. Nvidia goes consumer — Chamath confused, Friedberg sees the edge

Nvidia’s CES reveal: Project Digits, a $3,000 personal AI computer for researchers, plus a robotics test bench and driver-assist chips. Chamath’s skepticism, as hedged: “everything is telling us that we are reaching the limits of training” in LLMs, so the market shifts to inference — and it’s “not clear to me why that’s a better solution than all of the AI accelerators plus tensors that are now just prolifically being exposed” by Amazon, Google, and startups. His only guess: “maybe it creates some crazy DePIN market.”
Friedberg’s answer is that the bet isn’t LLMs at all: machine-vision models, graph neural nets for weather, “genome language models,” and real-time robotics — work he’s doing at Ohalo — need inference at the edge. “Do you really want that robot in the warehouse to be sending data to the cloud and waiting?” Tesla doesn’t drive your car from a server a thousand miles away; latency, dropped connections, and data-center dependence all argue local. And the price is disruptive: “the compute on this thing for $3,000 knocks a lot of Macs out of the field.”
On autonomy, Chamath holds his prediction-show line: “Waymo and Tesla are gonna run away with this market, and I think it’s gonna force a bunch of consolidation in the traditional auto OEMs.” The robotics platform “was cool” — the low-end PC remains the head-scratcher.

9. Cyan’s book: long NVDA picks-and-shovels, and Tesla should buy Uber

Her NVDA bull case is demand she can see: “we haven’t even barely touched what AI is going to do… early adopters cannot use Claude without getting shut down because of scaling issues, and I don’t think those are artificially created.” The consumer push is Nvidia growing beyond cloud GPUs “in case the music stops” — which she doesn’t expect. She also flags Jensen declaring Tesla one of the most valuable companies in the world long-run (while oddly backing Toyota), since no carmaker matches Tesla’s self-driving data.
The M&A idea: “I think they should buy Uber” — roughly 10% of Tesla’s market cap, 15% with a premium, WhatsApp-proportioned. Jason builds the bridge case: sell Model Ys to Uber drivers as an intermediate step, keep reinforcement learning running with safety drivers while each region legalizes robotaxis, and volume goes from 1.8M to 3-4M cars. Both find Amazon-buys-Lyft chatter puzzling — “a dying brand,” not global.
On AI and seed investing, her honest discomfort: a “Cambrian explosion of creativity” makes her job harder, not easier — “I think of an idea… and there’s 30 people working on it,” breaking the old taboo on backing competitors and forcing index-the-category thinking. Her response: “I tend to sit out a lot of hype cycles, so I invested in power and compute, lithography — all of the things that are going to be underneath all of this,” until consumer AI reaches a legible steady state.
The unglamorous alpha she does like: “there’s still a lot of spreadsheet companies out there” — e.g., AI ingesting an entire corpus of previous government RFP bids to compress a 30-day manual process, “and win more contracts. So that becomes like a national defense company at that point.”

10. Greenland is a shipping-lane play, not a Trump bit

Chamath’s decode of the “60 states” noise: as the Arctic ice shelf melts, the Northern Passage opens as a shipping lane, and a strategic arrangement with Canada and Greenland gives the US “monopoly control over something that could become as important as the Panama Canal” — before a militarizing China and Russia take it instead. The bombast is the distribution strategy: a sober tweet about “leverage in the Northern maritime shipping lane” and “nobody would’ve paid attention… It’s pretty smart.”
Cyan raises the stakes: “I believe either end” of the Panama Canal is operated and controlled by China. “We are at war with China, whether we like to admit it or not, in my opinion.” She’d extend statehood invitations outright — Taiwan, “it might be a bit controversial to even say India.” Friedberg’s more cautious rule for reading Trump: “it’s not the literal statement that matters as much as the vector and the magnitude of the vector.”
The episode exhales into Conspiracy Corner: Friedberg, fresh off Lex Fridman’s Graham Hancock interview, marvels that sea level was 400 feet lower just 12,000 years ago — Malta once connected to Italy, unknown ruins under Greenland’s ice. Asked for her pyramid theory, Cyan offers UFOs (“I don’t know if you ever tried to make a catapult, but it’s really hard”); Jason counters with X-Men mutants. Off-topic and kept in anyway — “No, no, I think it’s incredible.”

[Ride Home] Simon Willison: Things we learned about LLMs in 2024 Market Predictions, Rates & Inflation, DOGE, CES, AI Compute | BG2 w/ Bill Gurley & Brad Gerstner