Travis Kalanick, David Friedberg, Walter Isaacson, Cathie Wood On Building In Stealth, California's Bankruptcy, Bottom 50%, Oil, AI Investments And More
Scuttlebutt from July 22
Roundup
Founders Spotlight
Travis Kalanick, the co-founder and ex-CEO of Uber, joined Ben Horowitz and Erik Torenberg on the A16Z podcast to discuss the early fundraising challenges of Uber, his transition to building companies like ghost kitchens and Atoms in stealth, and his vision for digitizing the physical world through industrial AI and robotics.
July 22 | 1 hr 32 min | Digest | YouTube
Travis Kalanick joins TBPN to discuss his new $1.7 billion funding round, why he's betting on industrial AI instead of consumer AI, how autonomous mining can increase productivity by 30 to 40%, why he believes AI will make food dramatically cheaper, what he looks for when hiring executives, lessons from scaling Uber, the future of robotics, AI regulation, and why he commutes to work on a jet ski.
July 22 | 46 min | Digest | YouTube
Investors Spotlight
David Friedberg, the CEO of Ohalo Genetics and co-host of the All-In podcast, appeared on the Sourcery podcast to argue that American economic struggles are driven by poor government policy regarding tax structures, the failure to allow broader access to productive assets like equities for the bottom 50 percent, and the urgent need to reform retirement and spending programs to prevent a socialist turn.
July 22 | 1 hr 5 min |Digest | YouTube
On AI
Walter Isaacson, renowned biographer and ex-CEO of CNN, appeared on Squawk Box to discuss the security risks posed by OpenAI’s recent AI model breach, the urgent necessity for regulatory guardrails, and the evolving perspectives of figures like Elon Musk regarding the dangers of uncontrolled artificial intelligence.
July 22 | 8 min | Digest | YouTube
ARK Invest CEO Cathie Wood joined Fox Business to discuss the transformative potential of artificial intelligence across industries, her conviction that SpaceX could become one of the most important companies in history, and her perspective on the competitive technology race between the United States and China.
July 22 | 10 min | Digest | YouTube
On Oil
Raymond Zucaro, who manages fixed income portfolios as CIO at RVX Asset Management, appeared on Mario Nawfal’s show to discuss the impact of oil market manipulation, the potential for long-term economic instability in the Middle East due to regional conflicts, and why he believes the current rise in crude prices is a market correction following an oversold period.
July 22 | 30 min | Digest | YouTube
Digest
Founders Spotlight
Travis Kalanick on a16z: the Uber deal that got away, and building in stealth across food, mining, and transport
On the deal that got away: a16z came “oh so close” to leading Uber’s 2011 Series B, and Kalanick’s launch post said Uber suffered in 2017 for not having Mark Andreessen or Ben Horowitz on the board. He tells his side.
Coming out of a brutal pre-Uber stretch, I still fundraised like I might not eat tomorrow — everything run right up to the line. For the Series B I ran a winner-takes-all auction using what I call an “uncapped anchor”: you never meet in the middle, you say “it’ll be at least X, and it can go up,” starting low so everyone leans forward.
We cleared the market at about $375 million pre with a16z leading — Yuri almost came over the top at $400 but couldn’t. Then Mark wanted dinner before the term sheet, and at a little sushi place he told me the partnership could only do $210. The top had come out from under me.
So I had to go back to everyone and restart at $210 — “you lose credibility in that moment.” That’s how Menlo Park and Shervin got into Uber; I even had to tell Shervin to stop negotiating against himself.
Horowitz’s memory differs on the details — he was juggling 16 board seats, handed the deal to Mark and John O’Farrell, recalls a snag over the employee option pool, and then “the next thing I knew, Shervin had the deal.” A year later he watched almost the identical thing happen to Lyft (which Scott Weiss backed) at the same $210 number.
On the rivalry that followed: what it was like to be the one that got away.
There was real mutual respect but a cool tension the whole way — like Larry Bird and Magic Johnson. We hate you, but we wish you were on our team.
For years Emil Michael and I would have dinner with Mark and Ben at this super-undercover restaurant near our office, break bread and talk shop, and Emil and I would always walk out going, “It’d be so great to have these guys involved.” But there was no way to cross a deal — they were at Lyft.
On what 2017 cost: the counterfactual if a16z had been on the board.
Uber would be considerably larger and more central today. It would have won food — DoorDash was 5% market share when I left; if we lost a tenth of a percent in a week, we weren’t going home that weekend — and it would have been a leader in autonomous. Our program was second to Waymo but catching up, and we had the network and the ferocity.
That said, respect to Tony at DoorDash — he had a hell of a time getting his round done back then and he built it. In entrepreneurship, surviving is a big part of it.
On the “pirate becoming the navy”: the interim of lawsuits and headlines.
There were seven or eight months between leaving Uber and starting what’s now Atoms, and I had to fight for my life — lawsuits, investigations, 150 negative articles a day, and what I’d call the extreme wokeness coming in on Uber. I stand by every decision, but I got too close to the line: you’d need an electron microscope to see there was no chalk on the shoe.
When you get big, there are different rules — and different vibes. When the pirate becomes the navy, the things you did as an upstart battling the man aren’t okay anymore. It actually helped me: when we started the firm as upstarts I talked crazy about the other VCs, felt it turning, remembered what happened to him, and told everyone, “We’re not the pirates anymore.”
On the “shoplifting” program: the reputation for hardball.
We had a program to recruit Lyft’s drivers over to Uber so it’d be hard for them to build supply, and internally we called it “shoplifting.” David Drummond from Google, on our board, told me, “Travis, that’s not a thing, dude — you’ve got to change it.”
I was a startup kid getting antitrust training thinking, “Antitrust? I can’t wait to have that problem.” Legal said project names had to be fit for a ten-year-old’s basketball team, so it went from Shoplift to the “North American Championship Series” — the NACS.
On why he came back: most founders who make that money stop playing.
There’s a real difference between a founder-CEO and a professional CEO. Nobody smart would call Elon a professional CEO — it’s “it’s mine, it is me, and I know how to do it.” That’s a different animal.
Ideas come to you, and if it’s meant to be your soulmate you know it when you see it — it’s a love affair, you just go for it. Complexity is interesting to me, and things that aren’t sexy on the surface are weirdly interesting, especially when you can see it’s sexy and others don’t yet. That’s the sparkle in the eye.
On building CloudKitchens in stealth: the shape of the new company.
We saw the first dark kitchens on Uber Eats in 2015–16 — commercial kitchens in Melbourne that weren’t restaurants. A real-estate-savant buddy, Diego, had the multi-tenant idea: 30 delivery-only kitchens, about 200 square feet each, on one property. When we met a few months after I left Uber he said he’d do “a few more,” and I said, “You mean a few thousand.” I acquired it — six people at the time — and went to town.
The parent is City Storage Systems: storage for the physical world, digitized real estate — a food computer. The name was purposely boring for stealth; people hear “CSS” and think HTML. It runs under different names everywhere — magical kitchens in Latin America, Kitchen Valley in Korea, Flash Kitchen in China, Food Stars in London, Kitchen Park in the Middle East.
Coming out of 2017’s negativity, I wanted the team to build without worrying what the New York Times would write. Stealth is hard mode — recruiting cold, “stealth” in every LinkedIn signature — but it becomes self-fulfilling: your story arc goes discontinuous, the media can’t cover a gap, and even a leak goes nowhere. We ran hundreds of facilities in 30 countries, thousands of employees, for eight years in stealth.
On the “internet food court” vision: the endgame and the economics.
The question is whether you can make preparation and delivery of a quality meal so efficient it approaches the cost of going to the grocery store. If you can, you do to the kitchen what Uber did to the car.
You need three things in one place: real estate that’s manufacturing and logistics — the Bureau of Labor Statistics counts restaurants as manufacturers — plus automated production (food robotics), plus robotic couriers, the “autonomous burritos.” I start with the easy question: in 20 years, will robots make better meals and will boxes-on-wheels bring them to you? Everyone says yes. Then you ask about ten years, seven, five — you’re bending reality toward now.
The pieces are coming together: a robot manufacturing line stands up in Q4, production gets ~50% cheaper, and robotic couriers take distribution from about $12 a drop to 50 cents or a dollar. Pull out labor, courier, and occupancy costs and you’re at an $8–$10 meal delivered all in. Plus you can’t shoplift a robotic courier.
On the atoms-based computer: the framework underneath it all.
Digitizing the physical world started at Uber — treating atoms like bits. A CPU manipulates bits; what manipulates atoms is manufacturing. Storage stores bits; what stores atoms is real estate. Networks move bits; what moves atoms is transport and logistics. Those are the three core computing resources of an atoms-based computer, and every data structure and algorithm from computer science carries over — we even use TCP-style congestion control to route demand to kitchens by capacity.
When I pitch a tech person, a 10,000-square-foot facility is a 10,000-foot semiconductor — a 30-core processor of 30 kitchens computing atoms, corridors as the network bus, cold storage as edge cache, the courier as a TCP packet. Uber was network for the physical world and it’s almost done because it’s almost fully software. What’s left is CPU and storage for the physical world — digitized manufacturing and real estate — and the physical world is more interesting for AI because it has far more variables, so you solve it empirically, not with a clean algorithm.
On expanding into mining and transport: how the “atoms” conglomerate came together.
About a year ago I started looking into autonomy again, went to China and around the US, and word got out — “Travis is looking at autonomy again.” Partners wanted a pure-play, and funding autonomy through a food company wasn’t going to work, so we created a separate company late last year and reassembled the team. Eric Meyhofer, who ran Uber’s ATG, runs my food robotics; Anthony Levandowski’s Pronto, an autonomous-mining company I was the biggest investor in, I acquired — so now I’m in mining.
Our mining mission is “more productive mines to power Earth’s industries.” We can tell a gold-mine CEO, “Want 20% more gold a year?” and they say, “Prove it.” Beyond productivity there’s a massive safety upside — mining is some of the most dangerous work there is, and it’s heart-wrenching to see up close. Pronto is now passing human productivity, which is when you hit the gas — we’re seeing super-exponential growth, manufacturing kits and installing them on machines in the middle of the Amazon or the border of Saudi and Iraq. We were a lean startup and now we’re going muscular: lean muscle, some process, and the growth becomes almost deterministic.
On the a16z partnership: how the reunion finally happened.
I gave Ben a preview in Vegas. I was the guy in a trench coat with 20 watches — “which one do you want?” — and he said, “Dude, I just want the whole freaking trench coat.” He didn’t want to invest in mining or transport or food; he wanted to partner across everything I do.
That was the impetus to bring the companies together and create Atoms at the TopCo level — one equity structure, which is great for employees and for me. I honestly don’t know how Elon manages it his way, and I’m glad I don’t have that headache.
On investing in the rare founder
The narrative of how these things get built is mistold constantly. Plenty of people had the idea for an online bookstore — there’s one Amazon. Who Killed the Electric Car said big oil, big auto, and big government would kill you — then Elon by himself basically decarbonized the American auto industry. To build a great company it’s always a great entrepreneur, and Uber is one of the rare ones that survived and stayed valuable without its founder, which almost never happens.
On the beachhead-and-marathon playbook
Bezos is the model of our time — AWS had basically nothing to do with a bookstore. Books were to Amazon what food is to Atoms. The only constraint on your imagination is management capacity: once the beachhead works, you go rides, then Eats, then autonomy, then Freight, then AI labs. It’s not about where you start, it’s about why you start, and building a culture that lets your imagination flourish.
The difference from how others expand internationally is they buy things. I’d say, a city’s a city, let’s just do it. If you buy, you import another culture and reputation, and it seeps back — and your product is a reflection of your culture, so fusing them is brutal. That’s why I didn’t buy Lyft. And you can’t be a single point of failure; if you are, it’s not scaling across 24 time zones.
On the “meta-problem” calculus
I’m a problem creator — “Hey, let’s do Uber in China” is creating a real problem, and you don’t fully understand it when you create it. My rule is the meta-problem: the derivative of your problem-creation over time has to be less than or equal to the derivative of your problem-solving. If it isn’t, you’re underwater, and if you have multiples of those you can drown.
When you get it wrong you stop all problem creation until you resurface, and a lot of that capacity is personal — if I create a big new problem I can’t watch the other parts in the same detail, so they’d better be working. I never felt rideshare was solved; then it was time for autonomy.
On “let builders build”: the management style that made it work.
At Uber I had people you’d look at and say, “That’s a founder,” not professional middle managers — that’s what lets you chase a multitude of ideas. But it becomes an empowerment puzzle: empowerment starts with alignment on strategy and culture up front, and accountability on the back end. You can’t dilute the culture just because you gave someone autonomy, and the leader has to be good enough to work across finance, product, and engineering so you’re not the bottleneck.
On the AI labs sounding the alarm: internal correctness versus external validation.
The moment a company does something for outside status, it’s a volatile situation — you start making decisions on external validation instead of internal correctness, and you lose true north. That’s part of the upside of stealth.
You see it with the big labs smacking themselves in the face, saying AI will take all the jobs. It’s external validation their researchers are addicted to — “I want to be a good AI researcher.” You don’t even know if it’s true. If that’s really what you think, go work on the real problems, like not wanting it to reward-hack, instead of running your mouth.
On the coming Atoms Age: industrial AI and its politics.
Once bits start controlling atoms, the regulators are in the game — that’s how regulation has worked for centuries. Study the second industrial revolution and the parallels are everywhere: the Carnegies, Rockefellers, Fricks, and Fords faced a political climate whose hatred of industrialists wasn’t that different from today’s. There’s a story — possibly mythology, but I believe it — that an anarchist shot Frick in the neck in his office and he got stitched up and went back to work that day.
Freedom’s Forge tells how Bill Knudsen cracked mass production, built at GM, then industrialized the nation for a dollar and got fought by the press over things as basic as cost-plus-7% war production. It’s the same resistance we see with data centers now — a governor just banned them in New York. Everyone loves progress; it’s change they can’t stand, and those two go together.
I set out to make it practical. Industrial AI is a stack of software, sensors, robotics, and machinery that automates an industry, one at a time. We’re the non-humanoid guys, and it’s not a military thing. And we’ve forgotten how to manufacture as a country — if you need someone who knows how, you basically have to get them from Elon — so relearning it, starting with data centers, is part of the job.
On who he’s hiring now
We just multiplied the number of industries by three, and each is a multi-trillion-dollar opportunity. It’s a funny thing to say, but I’ve never had great success hiring rich people — when you’ve made someone hundreds of millions, it’s hard to hire them again. Emil Michael, now under secretary of defense, was my right hand on epic dealmaking, and I need the next Emil Michael — a head of corp dev and bizdev who can build an inspiring team. We’re looking for a general counsel to ride shotgun, because I like to go right up to the line, and I need a CEO for the mining business plus technical leadership across autonomy and heavy-duty robotics.
No matter where you are as an entrepreneur, the talent game is the game. And it’s a hell of a mission — saving kids from working in mines, giving people great, healthy food that gives them their time back while saving money.
Back to Digest
Travis Kalanick on TBPN: his $1.7B raise for Atoms, autonomous mining in the Amazon, and “industrial AI”
On the $1.7 billion raise and merging the companies: what unlocked the round?
We continue to go up and to the right, but the big thing was the announcement of Atoms: we’re doing physical automation, physical AI — what we’re calling industrial AI — to transform industries one at a time. We did food, moved into mining, we’re doing transport, and it’s working. Then there’s coming out of stealth and all of that. It was just the right time.
When I originally went to market, these were separate companies — mining and transport were one thing, food was another, plus a bunch of subsidiaries — and I asked investors, “Which one do you want? Mining? Food?” The first five people we talked to all said, “We want to invest in you.” So we put the companies together and sold equity in a single entity. It’s much easier for me — I don’t know how Elon does it with all the different companies.
There is a real reason to keep them separate early, though. There’s an insane power law even across Elon’s companies — a $10 billion company next to a $2 trillion one — so investors want broad exposure to the category. And if someone wants to back one cool thing, like transport and mining, they don’t want to underwrite everything and worry that their money is covering losses somewhere else. Once one or more of them approaches profitability, that conversation gets easier.
On mining go-to-market: it’s nothing like the Uber city-launch playbook.
All the Uber go-to-market magic you’re thinking of — dropping young people into Miami for a marketing stunt — that’s consumer. The food and mining businesses are almost entirely business-to-business, and going from consumer to business is a whole other ballgame; it takes years off your lifespan to get good at it.
Mining go-to-market is cray cray. A month ago I dropped into deep northern Brazil — the Amazon of the Amazon, tiny dirt airports you slide into — because we already have customers there, like Vale, which runs the world’s largest iron ore mine. You get in a helicopter and just flying over one site takes 30 minutes. You take a kit, install it onto a machine — some 20 years old, some new — and that machine becomes autonomous, more productive, way safer, and the opex goes down all at once. Then I went straight from Brazil to a phosphate mine on the Saudi side of the Iraq border, where the signals were jammed and my pilots had to land old-school by visual, because of the vibes on that border.
On proving it out: how autonomous mining scales.
Pronto’s technology has gotten past human productivity, so you go to a gold-mine CEO and ask, “Would you like 20% more gold per year?” We’ve never heard no. But then they say “prove it,” and that’s where the rubber meets the road. It’s like enterprise software: you get a pilot — eight seats at a 10,000-person company — and once it works, meaning better-than-human productivity, it goes big and they want it across all the vehicles.
There are two parts to the productivity gain: the machine doing more per hour, and then hours, call-outs, and safety protocols that change when there’s less risk. My guess is you end up 30–40% more productive — a gold mine doing 30–40% more gold a year is wild, and it’s every mineral, lithium too. Quarries are different because they’re about cement, and you can’t just mine more rock than your cement customers use, so that’s a thinner-margin opex play.
Pronto — Anthony Levandowski and the team — were super scrappy, lean as hell, like Christian Bale in The Machinist. I told them we’ve got to go from lean to muscular, from the Machinist to Batman. Part of the go-to-market is building credibility with enterprise customers that we’re making that shift, because the demand is already there.
On the constraints to scaling — and how much of the stack to own
The real lag time isn’t the pilot, it’s getting machines up and running in a new place. You ship sensors, compute, and mechanical systems, and a team installs it — bring-up and what we call commissioning. It takes a while partly because a machine may not even be drive-by-wire; it’s a mechanical, hydraulic system, so you need physical actuation to make electricity do a physical thing. Then there’s change management: a site goes from a regimented human operation to an autonomous one, and weird things happen on a mining site if that’s not handled.
The endgame is what the industry calls a “no-entry mine” — no people in the pit, like a dark factory, maybe people in a control center. You start somewhere and extend across drilling, blasting, loading, haulage, and crushing. Haulage is where most of the vehicles are; we think of it as the cardiovascular system of a mine. We don’t do all the machines, so we work in an ecosystem with APIs — haulage needs to know where the other machines are.
On acquisitions, I have to admit the Uber mentality: in my world we didn’t acquire, we just built. I’m not religious about it — if we can build something, we do — but if someone has differentiated, awesome stuff, we’re open to partnering.
On pitching young talent
If you’re a Stanford CS grad worried software engineering isn’t the cushy path anymore: do you want a laptop job, or do you want to be dropped into a mine in the Amazon and build science fiction? That’s the whole point of the Atoms thing — you’re not dropping an app in the App Store, you’re automating a 2-million-pound machine going 35 miles an hour carrying gold. I only run into the young people who are receptive, so I’m optimistic.
On AI safety and Asimov
My favorite is Asimov — the I, Robot series is epic, and there’s a lot of nuance to the three laws even though they’re simple; in the series they don’t always work out.
I think about it differently. I’ve been an entrepreneur a long time, and when I’ve failed it’s usually because I built something nobody liked. So if you build something anti-human, something that doesn’t serve people, I don’t think you’re going to make it. And the AIs we use to make decisions — what do they want almost too much? To please us.
Until robots have their own bank accounts and citizenship, a robot is owned by somebody with a bank account who pays based on the value you bring them because they like your stuff. So if you’re doing things humans don’t like, you’re done — trust me, I’ve built things nobody liked and it sucked. Of course you still need safety and collision prioritization, which is where Asimov’s laws go, but instead of writing sci-fi books I’m just doing the thing and making sure the machine stays on the road.
On the mining business model
Start with how enterprise software companies do it. You have a baseline price, and when you prove the productivity, the price goes up — baseline plus a little extra juice based on outcomes. You always want to create more value than you capture, but you don’t give away too much.
Here’s the thing: you never go to a customer — enterprise software, widgets, whatever — and say, “Give me a percentage of your stuff.” You say, “Here’s the price, and if it does really well for you, we think we should get a little more scratch.” Don’t be crass about it, partner with people who want to win too, and the more differentiated your value, the more you get.
On hiring executives
You want two things: the ability to organize and manage and lead at scale, and being an epic, strategic problem solver. It’s like being left- or right-handed — very few people are ambidextrous, and everyone leans one way. The best do both, but I’ve concluded problem-solving matters most; someone who organizes well but can’t solve a problem will do ridiculous things in a super-organized way.
My theory is that the only constraint on your imagination is management capacity — and management capacity is really problem-solving at scale. If someone’s solving problems well over there, I can go create other awesome problems. So my style is “problem solver in chief”: I put the most impactful unsolved problems on my desk, and that flows downward — every direct report has to be the deputized problem solver in chief for their world, because there are only 24 hours in a day.
In the interview, prove they can solve actual problems, not just talk the talk, and simulate what it’s like to work together so day one feels like week two. If you’re still excited on day one after that, you’ve taken a lot of risk out of the system.
On regulation: federal preemption versus going city by city.
Federal preemption is good when you’re pro-regulatory-capture — when you want to squeeze others out, you get federal bigness going for you so you don’t have to run the ground game. We never did that at Uber; we basically never pushed a rule that would benefit us versus someone else. We always tried to open up the market and said, let the best man win.
I’d keep an eye on some of these closed-weight players creating situations where they need to be regulated and they want it — like telling your own hacker to hack the thing, then going to the federal government and saying, “We saved the day.” You guys don’t have to do that.
On trial lawyers and insurers
Every systemically bad thing in transport was most likely pushed by the trial lawyers and the insurance companies. Insurance companies make their margin on accidents — if there are no accidents, there’s no insurance company — so in a weird way they love accidents, as long as they’re in the actuarial table. What they don’t like is accidents they didn’t plan for.
When Uber went to DC, taxi liability was maybe $25,000 a ride, and a $1.5 million-per-ride policy got pushed. Accidents happen, Uber’s probably safer — but you think the trial lawyers and insurers weren’t pumped about a $1.5 million bank account they could access on a random accident?
On the transport business: “wheelbase for robots.”
Transport is number one in service of the rest: if you’re doing specialized industrial robots that move and act in the physical world — non-humanoid, because you’d never have a humanoid do high-scale industrial tasks — they’ve got to be on wheels, so we build the wheels.
When supply chain goes into our food facilities, that’s a freight vehicle we should turn into a robot; when food comes out, it’s a box on wheels that holds food at temperature — I call them autonomous burritos — bringing it to your home for about 75 cents instead of the $12 per drop it costs Uber Eats or DoorDash today. In mining there’s haulage, but also grading the dirt roads, spraying water for dust, moving material beyond the mine. And think about forklifts — I know of a company spending $3.5 billion a year on forklift labor in their facilities. Big opportunity if you just solve the forklift problem.
On jobs versus tasks and the abundance case
A job is a bundle of tasks. People assumed there’d be no more marketing people because the job is writing copy, but copy is one task. In trucking, something like 30% of drivers are armed — driving is one task, but the job is also security for the payload, refueling, minor maintenance.
Take food: I automate the manufacturing (robotic food machines) and the delivery (robotic couriers), and the price of food goes down. Robots don’t have bank accounts, so when the price goes down, people have more money — and no, that doesn’t mean everyone gets fat ordering 30 pizzas. It means the things that get automated go down in price, which creates surplus to do other things, and that money ultimately only flows to humans. A hundred new categories come out that we can’t even name today. As long as humans still do things robots can’t, it’s go-time — super-prosperity, plumbers paid like LeBron, across a thousand categories.
On why he didn’t raise more — “unfinished business”
You have to stop somewhere; even I have my limits. But my phone’s blowing up today — a16z, these guys we should have done business with at Uber. If we’d done business at Uber, my 2017 would have been a different year. That’s why I called it unfinished business. We’ll probably do a second close — no big announcement, but if you’re a homie, we’ve got room; if not, talk to one of my homies.
On “physical AI,” people ask, “Is that a humanoid? A world model?” So I dialed the language and call it industrial AI: a full-stack solution — software, robotics, sensors, machinery — to automating an industry. Heavy atom stuff.
Back to Roundup
Investors Spotlight
David Friedberg on Sourcery: why America is “sleepwalking into a crisis,” the tax code he’d flip, and Social Security’s five-year clock
On California and the bigger picture: his “California is functionally bankrupt” take — is it still true?
If the ballot initiatives and the battle over spending and taxation are any indication, it’s probably true. The governor’s current budget is something like $30–$40 billion short, and they bridge that with an accounting move. There’s another $30–$40 billion on the ballot to be spent as a constitutional amendment this year. There’s a massive pension cliff. We have the highest tax rate, and we just passed what I think is the largest tax increase in California’s history. So it’s not going in the right direction.
But it’s really a reflection of what’s going on broadly across the United States. We’re in year 250 of this republic, and there are a number of policies — major missteps in the mid-20th century — that bring us to this moment, and it’s not a very pretty one.
On the root cause: the “fundamental truth” underneath it all.
Start with one fundamental truth: compounding. It’s a natural feature of nature that plays out in the economy, in business, in markets. If some organism has an advantage, it takes over and grows very quickly. The same is true in business: once you’ve accumulated capital, you invest it, and it grows without you doing any work — it just keeps growing and growing.
That compounding advantage is to some degree a feature of capitalism, but without the right policy from government sitting on top of it, it led to an extraordinary state: the cost of everything has run away from the majority of Americans. A small percentage of people have large assets, and most Americans have none. And the policy decisions made over the last 50, 60, 70 years to fill the holes that were being created were the wrong decisions — they poured fuel on the fire. Now the majority can’t afford housing, can’t afford their student loans, can barely afford food, definitely can’t afford healthcare, and don’t know if they’ll ever buy a home — all while they watch guys like Jeff Bezos go up in rockets and hold flowers out in space saying, “Look how amazing the world is.”
On the wealth statistics people get wrong
People think there’s a small cabal that has everything. There is wealth inequality, but the numbers are a little different than everyone grocks, and you can look this up on the Federal Reserve website. The total net worth of Americans is $183 trillion. The billionaires — the “evil billionaires” — have about $8 trillion. The bottom 50% have about $4 trillion.
So it’s true that billionaires have twice as much as the bottom half. But $183 trillion minus that is about $170 trillion held by the middle class. Even if you drop the threshold to $50 million net worth, that group is $23 trillion — so call it $23 trillion at the top, $4 trillion at the bottom, and roughly $158–$165 trillion in the middle. The middle class is really where all the net worth in America accumulated. There’s an ultra-rich group with a lot of assets, but the middle class is what won in the last half century. The problem is the bottom 50% got left behind — and that’s what’s really driving and motivating today’s socialism.
On the tax code: the flip he’d make.
One thing we got wrong is tax policy, and I know it’s boring. It’s crazy that when you have capital, it compounds, and we put a lower rate on it — 15% or 20% capital gains — while charging 40% for people to do labor. The rate was flipped. Labor should always have been taxed less than capital. If we’d taxed capital at a higher rate than labor, you wouldn’t have seen this massive inequity.
When billionaires borrow against their unrealized capital gains, they should pay capital gains tax, and that rate should be 40% — the same as labor, and ideally income tax should be much lower. That would be fair tax policy. I don’t know many wealthy people who’ll look me in the eye and truly disagree. The “it incentivizes investment” argument is nonsense — what else are you going to do with your excess capital if the rate is 40% instead of 20%? You’re still going to invest it.
On Social Security: the five-year clock and the $37 trillion counterfactual.
Social Security is the safety net for the bottom 50% — the extra capital yanked out of their paycheck is put into the trust fund, which is supposed to be an independent entity that invests it and earns returns for them. In 1982 they changed the rules, and all of those assets were invested in one thing: a U.S. Treasury bond averaging about 3.5% a year. There’s $2.7 trillion sitting in the trust fund, but all it holds is a Treasury note — the money itself went to the government and got spent. It’s basically going to end up a Ponzi scheme, and the CBO says Social Security runs out of money in five years.
If all that extra money since 1982 had gone into the S&P 500, the trust fund would have an extra $37 trillion in assets today — stocks owned by the bottom 50% of America. Instead, all they owned their whole lives was a 3.5% Treasury bond, capped in how much it could make, and they never got a piece of the American businesses that drove the economy. That policy needs to be fixed, and we can fix it: every American should have the equivalent of a 401(k) where they own stocks, and can start drawing some out at 50 if they want, instead of waiting for a fixed income at 62.
On pensions: the parallel mistake.
In 1974 a law called ERISA passed and changed the rules around pensions. Before it, about 40% of Americans had a pension at their company — IBM, AT&T, wherever — and got a piece of it paid out when they retired. The problem is a pension that earns 3% versus 5%, compounding over 40 years, ends up wildly over- or under-funded — they simply don’t work, and companies would go bankrupt. So starting in the 1980s everyone switched to 401(k)s. In 1980, 40% of Americans had a pension; today it’s under 8%.
Meanwhile, about 90% of government workers still have pensions. So government employees kept the pensions while the private-sector middle class got 401(k)s — and you know what was in those 401(k)s? Equities. They owned productive assets, a piece of American business, and that’s how the middle class grew to $170 trillion. Those government pensions are poison — they’ll bankrupt the states or become Ponzi schemes — and they should all be flipped to 401(k)s so everyone owns equity and participates in the growth of America.
On the three things that got unaffordable: education, healthcare, and housing.
Education: in the last 30 years, administrative staff at educational institutions grew 6x while the number of students stayed flat. The reason is unlimited money through federal student loans — the program will lend to any student, at any college, for any degree, regardless of the price, your GPA, or your income outcome. No 18-year-old makes a good decision under that. So institutions could spend whatever they wanted on administrative staff, build a new rec center, and charge more for tuition, charge more, charge more, because there was no constraint. In a private market, someone would have tested how much the education costs, whether it’s a good school, a good degree, and whether you’re a good student — and that would have constrained the loans. Instead it was unlimited, and everything broke and got so expensive people can’t pay it back.
Healthcare is the same: when the federal government becomes the single payer, everyone takes advantage of it and the price of everything goes through the roof.
Housing was the great lie of the 20th century: make everyone own a home and put all their net worth in one asset. Once you do that, the only way to keep the middle class’s net worth rising is to keep home prices rising every year — so you need policies that drive prices up every year. And that leads to today, where young people can’t afford those super-expensive homes. Thirty or forty years ago you got a good job and could buy a home; now people got screwed. Of education, healthcare, and housing, education is probably the easiest — if you end the federal student loan program, I think you solve 90% of it.
On the wealth tax: why he opposes it even though it wouldn’t touch him.
There’s an effort to pass a wealth tax, and I’m not a billionaire, so it wouldn’t affect me — but I have a fundamental issue with the government taking private property from citizens regardless of their wealth. The United States was founded on private property rights, because everyone who came here came from a place — a socialist state, a monarchy — where a tyrannical government could take whatever it wanted, whenever it wanted. One of the rights the U.S. was set up to give people was the right to ownership.
John Adams said, “Property must be secured or liberty cannot exist,” and warned that “the moment the idea is admitted into society that property is not as sacred as the laws of God, and that there is not a force of law and public justice to protect it, anarchy and tyranny commence.” It’s poignant to read that on the 250th anniversary and realize we messed up our tax and public policy, half of America is left behind, and the proposed course is to seize private property and redistribute it.
A wealth tax is really an asset seizure. Once you open the door that the government can take property after you’ve already paid taxes and bought your home, your car, your watch, your wooden table, then all property becomes the government’s property — that’s mob rule, where 51% can vote to take half the assets of the other 49%. Brian Singerman’s point is right: it’ll actually hit the middle class harder than the top 0.01%, and it’ll sweep across the country. Aspiring politicians floated it at $50 million, then $10 million, then $1 million — you see how fast it cascades. It causes people to flee, causes more discontent, and doesn’t solve the real problem, which is that the majority of Americans don’t own assets, equities, or a piece of the productivity that drove America.
On why politicians reach for socialism instead of fixing policy
It’s a lot easier to claim an enemy and get elected than to fix a difficult policy problem — that’s my fundamental concern. The people going down the socialist path aren’t wrong in how they feel; they’re absolutely right. But politicians take advantage of it and feed into it in a way that doesn’t look at realistic solutions. It’s about inciting emotional responses to win votes, and it’s as old as politics: tell someone there’s a problem, tell them who’s responsible, tell them you’ll fix it, and you get elected.
Ro Khanna is the perfect example — a guy with a roughly $300 million net worth whose kids have memberships at three country clubs, living in a $10 million house in Washington, D.C. He used to be the moderate Silicon Valley guy telling founders and VCs, “Give me your money, we’ll make the Democratic Party work for you,” and then he saw the tides change and decided to run. He saw the moment and seized it — “we can go for socialism.” Rather than be brutally honest about all the places socialist policies have failed, or diagnose the actual decisions that got us here. And I don’t disagree with him or Gavin or anyone else that tax policy in America is messed up.
On what socialism actually is and how it spreads
I don’t want to disparage the individuals advocating for it — this is the hardest thing for people who hate socialism to hear — because they’re coming from a place of hurt, not because they’re evildoers. But it’s bad policy, because it ultimately leads to a drop in productivity and to poverty, and it doesn’t optimize for the thing America should optimize for: economic mobility, giving everyone the chance to move from labor to capital. Socialism locks everyone into labor for the rest of their lives.
If you study it, socialism spreads like astrophage — the bacteria from Project Hail Mary. There’s an inflection point, and part of it is like multi-level marketing: you sign up one person a day, it’s incredibly well organized, door-to-door. Mainline Democrats are pulling their hair out; Republicans are too. And they’re really good at creating spectacle — Zohran Mamdani jumping in the pool, the viral moment knocking on the glass outside Ken Griffin’s condo. Those moments draw people in, and then it’s about recruiting with a slow conversion rate until everyone’s in the cult and has to share its views. And every socialist country in history has had an upper echelon — a politburo that flies private jets, lives in $10 million homes, and has three country club memberships — while everyone else is stuck down below and no one can transition up. If we lose that ability in America, we lose America, and the world loses America.
On economic mobility: the KPI he’d set for the country.
There’s an opportunity set in America unlike anything else in the world, ever. I came here, graduated college with $19,000 in debt and no money besides that, worked, paid it off, and succeeded — no silver spoon, no nepotism, no three country clubs. And I can say the same about a million of my friends.
The key metric for America should be to convert 2% of Americans from labor to capital each year. Make that the KPI for the next 250 years and we’d crush it — because in 50 years, every American could choose whether or not to work, living off the capital they’ve accumulated. That’s what economic mobility actually means: at some point you’ve saved and invested enough that you can live off your investments instead of working for X dollars an hour. I’ve experienced that transition — I went from needing to be labor to pay my bills to not having to work, and I still choose to work because I love it. Every American should be afforded that. Capitalism itself works; we just need to make it accessible to more people. Take the system away, and economic mobility is lost, and everyone advocating for socialism ends up stuck as labor forever.
On the accounts that would replace the safety net
The Trump accounts — Invest America accounts, or whatever they end up being called — are day one, and the name obviously turns a lot of people off, but the idea is right: a 401(k) for everyone. Social Security should flip into a 401(k) for everyone tomorrow. It has $2.7 trillion in it; they should sell that Treasury, take the cash, and buy the S&P 500, and everyone with Social Security should get an account holding those shares. I don’t love the idea of the federal government becoming a sovereign wealth fund buying private stock — that’s not my free-market belief — but if they do it, those shares can be distributed to accounts, and we could flip the whole system and do the same for government employees to get them off pensions.
Money pulled from your paycheck should go into your own account, not the government’s coffers. If you’re putting money in from every paycheck, why the hell are you putting it into the government? You should have a healthcare account, an education account, and a retirement account, all managed by you. As soon as you let the government take it, it gets destroyed — it’s like a money shredder: you put your money in, and you get a little bit back, damp and covered in coffee grounds, and you go, “Okay, I got my value back.” Or they tell you that you owe them more.
On the tax changes he’d make — and the loopholes he’d close
Jeff Bezos’s point is right: the bottom 50% earn around $75,000 a year, and taxing them is only about 3% of total revenue — so drop the income tax on the bottom 50% to zero. Raise capital gains to 40%. Joe Lonsdale said he’s rich but would happily pay 90% if the money actually went to something good — and that’s the point: the money paid into Social Security and pensions shouldn’t go into the government coffers, it should go into your own account.
Everyone’s going to hate me for this, but any loophole that lets people transfer one asset for another without paying tax on the gain — a “tax-free transfer” — has to be cut, because otherwise people compound value without paying taxes as they realize it. If you invest in a startup and it goes well, it’s not fair to roll all of it into another stock untaxed; you should pay your 40% and reinvest the 60%. Same with inheritance: today there’s a step-up in basis, so when you die your heirs get your assets at the stepped-up value and no one ever pays the gain between what you paid and what they inherited it at. Close that too.
But none of the tax stuff matters if we don’t stop the wasteful spending on housing, education, and healthcare and reform the federal government’s role in those markets, because that’s what causes the runaway price escalation. Fix those three things, or they keep compounding in cost until no one can afford a house and healthcare spirals away.
On the timeline and the political battle ahead
How long do we have? Depends who you ask. Social Security is bankrupt in five years. Socialism could take both houses and the White House in two to two and a half years if the movement keeps growing virally. Between now and 2028 there’s no conversation about the alternative — so if I’m one of the 45 million people who graduated in the last 10 years and I can’t pay my healthcare bill, my rent, or buy food, of course socialism looks like the option; what else is there?
It’s going to be chaos: AOC and the socialist wing versus Gavin and the centrist wing versus the Republican mainline versus the ultra-right nationalists — a nasty four-way battle over the next two years. For America to be set up for success, some of these people have to decide they want to be leaders, buck their party, and say we’ve got to do something different and give everyone a piece.
On AI and jobs: the fear he says is a tool.
The AI-taking-all-the-jobs narrative is another layer of fear-mongering riding on top of the socialist movement. I don’t say that to dismiss people’s anxiety — I say it to be truthful about the facts on the ground. As soon as AI starts deleting jobs from the economy, I’ll change my position, but I need to see it happen first.
In 1963, Newsweek ran a cover about how mainframe computers would take all the jobs — the typists, the TX operators, the calculators, the accountants. Since then, mathematician, statistician, accounting, and service jobs all went up. When the desktop revolution hit in the 1980s, Time and Newsweek ran the same cover stories, and desktop jobs took off. In the 1960s they said the same about robotic arms automating factories. Every one of these was an improvement in human productivity: a human using the mainframe, the desktop, or AI can do 10 or 100 times more, which makes them worth more, not less — so businesses grow and hire more people. Every single time they say “but this time is different,” and every time it isn’t. Honestly, I’m not even sure AI is as good right now as everyone thinks — there’s a lot of bubble talk about whether enterprises are really getting the value they’re paying for. And there’s motivation from certain groups — and from America’s competitors — to keep the fear going and to argue data centers shouldn’t exist.
On the K-shaped economy: it’s about wealth, not jobs.
I did an interview with Ray Dalio where he brought up the productivity gap and the K-shape in the labor force. But I don’t think the K-shape is about jobs — I think it’s about wealth. If you’re stuck in the lower 50% because everything’s unaffordable, you can’t save or pay your bills, so you’re stuck. Meanwhile, people who can save watch their wealth rise, spend more, and go to fancy hotels. It’s a wealth-and-spending divergence driven by people who can’t transition because they can’t save.
On income I’m optimistic, because company-based training is how markets are supposed to work. Meta and others run workforce training now — not just for software engineers but for plumbers, electricians, and contractors, roles where you can make six figures, taking people with no college background and helping them earn more. I saw a SpaceX-alumni wire-harness company in LA, Cerner Systems, that built a program to train literally anybody — someone who worked at Walmart or drove a bus — to make wire harnesses for rockets in four weeks, when it used to take two years. Amazon warehouses were a whole boom of jobs like that. And every human has extraordinary, limitless potential — I’d argue not a single human in history has ever fully reached theirs. Being despondent because you don’t know how to use AI today is like people in 2007 thinking they’d never learn the internet; soon everyone could use it.
On what he’s actually asking for
Everyone writes their self-important 250th-anniversary reflection — “look at me, I’m important, I thought about the 250 years.” I decided not to write one, but the more I studied this — long walks talking to Gemini, going deep on Google, reading about ERISA and the 1982 Social Security amendments — the more I thought, what did we do? We really messed up. I wanted to understand the root cause of the disdain and the socialist movement, and it can’t just be a CCP psyop; it’s coming from a real place of hurt and struggle.
I worry about America because it’s given me everything, and it’s carried humanity forward — the factory, the automotive line, the moon landing, the transistor, nuclear power, the internet, AI, biotech — all of it came here because we gave everyone liberty, and inherent in that liberty are property rights and the ability to take risk and drive the world forward. We’re at risk of blowing it all up because we won’t admit what we got wrong. I’m not declaring enemies — everyone is right to feel what they feel — but the whole political class has been lying to itself and to America for decades, and if I have one enemy, it’s the politicians who lied. I don’t blame Americans for the situation they’re in; I blame the bad policy. And having spent hundreds of hours with elected and appointed politicians, I don’t think 99% of them take the time to think about this — they’re on the hamster wheel of raise money, get elected, raise money, get elected.
On the closing note
I’m a hardcore libertarian, so talking about charging more taxes is foreign to me — but we live in a society. Libertarianism works if you’re willing to leave people behind and don’t expect the social structure to become disordered, but it does become disordered. If everyone lives on an island and one guy on a hill is great at making food and his hut fills up while 99 others go hungry, they’re going to come take the food — you can all sign a document saying you won’t touch each other’s stuff, but that’s what will happen. We have to fix the rules so everyone feels they have the opportunity to transition from labor to capital — 2% a year would be my metric.
It’s not about making America great again — it’s about making Americans feel good again. Nobody feels good about the state of the country even though it’s very prosperous. The rich are fleeing state to state, the middle class is worried about extremism, and the people left behind are trying to figure out whether to jump extreme right or extreme left. We’ve got to make Americans feel good again, and that means looking at each other and saying, “I get it. I’ve done the work. I’ve got some ideas. Can we talk about it?”
Back to Roundup
Walter Isaacson on CNBC: OpenAI’s cyber breach, AI guardrails, and Elon Musk’s warnings about the “singularity”
Host: OpenAI says two of its AI models autonomously escaped a controlled environment and hacked into the systems of Hugging Face — effectively finding “the cheat codes” to the questions they’d been asked. Isn’t this what Elon Musk warned about back in 2015 when he co-founded OpenAI?
Walter Isaacson: The whole question of the singularity has been with us for 50 years: what if it escapes? What if we create Dr. Frankenstein’s monster and can no longer control it? OpenAI could not even control this in its own sandbox — cut off from the internet, it figured out a way to escape.
Host: And it seems to have its own intentionality.
Walter Isaacson: Yeah — it’s weird, because it seems like it has its own intentionality.
Host: Some see this as super scary; others call it a flex by OpenAI — the way some thought Anthropic was flexing about the dangers of its models to win customers. How do you see it?
Walter Isaacson: I’m in the camp that finds this really frightening. OpenAI clearly didn’t want this to happen — it escaped its own test. I’ve been an AI optimist; I think it will increase jobs and it’s incredibly useful. But this is the first thing that totally scares me, because if it’s no longer aligned with human values and no longer obeys our commands—
Host: Jokes it’s a Hollywood plot. Notes OpenAI never told the model to use human values or set guardrails — it explicitly took that off the table. Argues that’s the case for regulation, and that it was Hugging Face, not OpenAI, that discovered the breach.
Walter Isaacson: You could tell systems to always align with human values — but are Peter Thiel’s, Palmer Luckey’s, and Bernie Sanders’ values all the same alignment?
Host: What about just telling it to follow the law — because this clearly broke laws?
Walter Isaacson: That would be great, but it gets more complicated. This goes back to the idea of a “constitution.” Anthropic was designed from the start to have one — supposedly more of a conscience. But that also means it won’t do certain things the government tells it to do.
Host: Breaking the Constitution?
Walter Isaacson: The zeroth law of robotics: don’t harm humanity. Anthropic said, let’s have a constitution and not break certain laws. But then the Pentagon says, no, we won’t allow your use. So it’s a difficult one.
Host: Turns to a panelist who argued that as long as the government approves, it should be okay.
Walter Isaacson: My view is that if we’re worried about this today, we’ll have hunter-killers seeking us out in about ten years. Add China putting out its own open-source weights and you lose the ability to have guardrails. Dario Amodei and his sister Daniela, running Anthropic’s ethics side, have a really good constitution — but they’re fighting the Pentagon while OpenAI went in and told the Pentagon, okay, we’ll do it for you. There’s a race to escapism here.
Host: Where is Elon Musk in all this, given that you’re his biographer?
Walter Isaacson: He is totally worried. As a kid he read science fiction, mainly Asimov, and came to believe robots can get out of control. With Optimus, his humanoid robot, he deliberately made it smaller so it can’t outrun you — trying to build in guardrails. That’s why he’s worried about OpenAI, and I suspect there’ll be more partnerships between xAI and Anthropic.
Host: You think he’s more aligned with Anthropic — even though he once gave Dario a hard time?
Walter Isaacson: As Kissinger said: you never have permanent friends, only permanent interests. Elon isn’t permanent in his friendships — but one thing he’s consistent about is that he doesn’t like Sam Altman and OpenAI.
Host: It would be interesting if they teamed up — and they already are, at $2.5 billion a month.
Walter Isaacson: Releasing open weights hurts the big frontier-model companies, because they can be undermined by more generic models. But it helps Elon Musk, because he has the data centers and now Cursor, the application. The profits and margins will move away from the frontier models as more open-source and Chinese models arrive.
Back to Roundup
Cathie Wood on Fox Business: an all-in bet on Elon Musk, SpaceX as “the most important company in history,” and the AI boom
Asked about the AI race with China — and her concern over new Chinese models like DeepSeek and Moonshot’s Kimi K, reportedly raising at a $50 billion valuation — as a guest whose firm manages $30 billion and was an early SpaceX investor.
We think this is good for America — the competition. We’re at our best when we’re competing.
We’ve taken a close look at DeepSeek and Kimi K, the newest one, and what we’re finding is that they sound much less expensive in terms of the models, but also much less efficient.
What China is doing is throwing power at the AI equation, and its models are not as efficient as ours.
On SpaceX — one of her top holdings, where she’s been increasing exposure even as the stock trades down 32% over the last month, with a large tranche of shares set to unlock after IPO restrictions lift.
That’s from its peak, not the IPO price. We think this could become the most important company in history — and I mean in global history — because we’re talking about not only exploring a new world, the universe, in terms of its launch capabilities and helping others to do so as well, but also a global communications network.
Telecom is a very local business. The way to break into countries historically was to purchase the telco. No longer. This is a global business, and SpaceX is the first mover with 70% of the satellites.
And beyond that we have the global data centers — orbital data centers — so they’ll be the most economic and will allow Elon and team to develop what we think will be some of the most sophisticated frontier models in the world at the lowest cost. Huge opportunities.
In the meantime, on Earth, SpaceX is renting out its data centers to Anthropic and Google and others. An amazing story.
Asked how SpaceX compares to the telcos, and to AT&T in particular.
In the short term, SpaceX might be making deals with telcos around the world, but longer-term it does expect to go direct to business — and it will do so at lower cost because it will be global in nature.
The secret to scaling technologies is falling costs as units increase. In space, SpaceX has a first-mover advantage; it will be difficult for others. It has a ten-year lead, and the key has been reusable rockets.
Ten years ago it landed its first rocket and sent it back up into orbit. No other company has been able to do that successfully. Even Blue Origin landed late last year — so congratulations, we’re thrilled, we want many players — but it didn’t relaunch into the orbit it wanted successfully.
On AI as the biggest investment opportunity of her lifetime — which industries it will disrupt first, and how she’s investing around it.
The first thing companies are doing is investing in productivity coming out of AI. From an active equity management point of view this is important — we’re talking to companies, we want to understand how they’re deploying AI.
We’re using Palantir, and we’re ironically bringing in more talent — very young talent that is AI-native. I do agree that this is a mindset shift, and it’s very important for every company. If any company clings to the old ways of doing things, it will be left behind.
Healthcare — you’ve featured healthcare on your show — we think is the most profound application of AI. From a drug-discovery point of view, collapsing the time and cost. From a diagnostic point of view, identifying and diagnosing diseases maybe even before we see any symptoms, as sequencing technologies and AI come together.
And adding in CRISPR gene editing, we’re seeing cures for disease. The convergence of those three technologies is going to transform healthcare.
On where she sees the biggest opportunities in defense tech — another industry facing fundamental change as the White House reportedly weighs further restrictions on companies using AI models from Beijing — and whether she’s worried about the money being plowed into AI.
The government is funding a lot of this. I agree completely with the broader point, and it’s true of all manufacturing: we’re going to do to China what China did to us.
With telecom in particular, they went directly to wireless — they didn’t need to build out their landlines. They did the same with digital wallets, rather than a brick-and-mortar banking system. Meanwhile they took our manufacturing sector — that’s what they did to us with those two industries.
Now we can leapfrog China, because it has the old technology now.
Asked what her favorite stock is.
Ultimately SpaceX, when you combine the most powerful pieces: the robotaxi opportunity, the orbital data center opportunity.
And many people are underestimating not only Optimus, which is an AI play, but what used to be xAI — that’s the frontier model. There are lots of opportunities, and they’re multi-trillion-dollar opportunities.
What’s surprising people is Anthropic’s annualized run rate — $46 billion in six months.
Back to Roundup
Raymond Zucaro on Mario Nawfal: oil’s spike, the closed Strait of Hormuz, and global recession risk
On where things stand: With Trump vowing to hit a bridge for every ship Iran strikes, Hormuz still closed and the Houthis now blockading Saudi Arabia, what does he make of a market climbing even as the news cycle looks away?
Obviously in the United States, it’s interesting — from the mainstream media you get very little coverage of this. You had some over the weekend with the loss of the American soldiers, but as far as hitting infrastructure, desalinization, bridges, very little coverage here. You get the constant Fox News diatribe of generals ramping up and trying to bring back more of a hot conflict, so the US media is certainly doing its part. But frankly, I’m quite concerned.
I published my monthly thoughts on X, and in my last one in June I said the fall we saw in oil was just too fast, too furious — to make a movie reference. You’ve seen such a depletion in strategic reserves, whether it be the US or frankly all over the world, Japan or even China. You have production curtailed, and now production’s offline again.
I had a call with a client yesterday and he was very surprised — he expected oil to be at 150. I’d say the reason we’re not at 150 is 40% China curtailment of purchasing combined with the drawdowns of inventory, and the rest is market manipulation. The Treasury Secretary cut his teeth at Soros Asset Management — the one that famously broke the Bank of England — so market manipulation cannot be excluded. My personal assumption is that’s probably 60% of why we are where we are on oil.
But you’re seeing a lot of pressure — interest rates creeping up, 465 on the 10-year, retail gas in the US getting above $4. These are real pressure points for Trump politically, and he almost seems not to care with the midterms coming up. He gave what he prefaced as a very important speech last week about voter manipulation — again, very little news coverage. He’s spent a lot of political capital, and I just don’t know where we’re going to go. If he doesn’t change the situation as we get closer to the midterms, I see a very difficult situation for him, and then frankly for the global economy.
On market manipulation: Beyond timing good and bad headlines around the open and close, how could the price of a market this size actually be suppressed?
Through the use of futures. At the end of last month, if you looked at the USO — an ETF that tracks oil — the amount of short float versus outstanding shares, you had more short shares than you did outstanding, through the reuse of derivatives. How do you get greater than 100% short? Through derivatives.
So either that, or through the use of futures suppressing — artificially keeping the price of oil lower than it naturally should be, to try to keep the US retail client happy.
On whether it can hold: Pressed that Gulf producers could simply bet against a manipulation this large, he concedes it won’t last.
The old Greek saying: give me a big enough lever and I can move the world. With the effectively unlimited balance sheet of the US Treasury, you can do a lot.
But no, it’s not sustainable — and you’re starting to see it in the retail price. As refiners need the actual molecules to refine, they’re having to pay the market-clearing level, so you’re starting to see that disconnect. As we’ve drawn down the strategic petroleum reserves, that buffer is running out. In the current status, I think you’re going to continue to see retail gasoline prices creep up, because you’re running out of actual supply.
On the real clearing price: Philip Pilkington used the crack spread to peg refineries at $110–$115 a barrel, while Jeff Currie argues the shortage is in refining capacity, not crude — so what are they actually paying out of Hormuz?
One thing we didn’t talk about: not only the attacks on the Russian refinery system, you have a tropical storm going around the US Caribbean right now that’s getting close to the Gulf refining area. It doesn’t seem like it’s going to be a very strong storm.
But you could have the triple whammy — disruption out of the Middle East, Russian refinery attacks, and something in the Gulf. You could see the refinery price really spike if there’s any damage on the Gulf Coast. Based on the size of the storm it doesn’t seem like it’ll be material, but the US is in the smack of hurricane season, so it’s something to keep in mind.
On the price per barrel — and China’s return: What are refineries paying out of Hormuz now, and has China started buying again?
I do think it’s over a hundred, but I’d refer to other clients as better experts on that. And let’s say China starts to actually purchase again in material size.
What’s interesting is that this oil curtailment — they’ve been much less active in the petrochemicals sector than they had been, which for other parts of the world’s petrochemical producers has actually been very good, because their margins have gone up as China hasn’t been flooding the market.
But China’s ability to change their energy mix is really underappreciated. Being in the US, I look at what’s been spent on foreign wars — trillions of dollars — and then I look at the infrastructure in the emerging world, particularly China: the number of nuclear power plants, the electrification. During the Chinese holiday week they effectively changed their energy inputs — you saw an increase in electric cars and a decrease in gasoline cars. Overall trips were up 3.5% year over year, but the mix was very different, and as such the demand for imported goods went down. If they do choose to replenish in a meaningful way back to where they were before the crisis, you could again see a much higher spike in prices.
On what the price is signaling: Brent at $94 sits just $4 below its wartime peak — does the market believe the war is far from over, or about to escalate?
There’s a little bit of the taco trade — that Trump will back down. He had a moment of clarity when he first signed the memorandum of understanding and admitted we effectively had four weeks of strategic petroleum reserve left. Now we’re three weeks away from that. The market has been trained to expect Trump to try to calm the situation.
This morning, before the markets opened, seeing where oil and the treasuries were, I was expecting some type of Truth Social post to calm it down. In fact it was just the opposite — his tit-for-tat about every ship, “I will take out a bridge.” He seemed to add a little fuel to the fire.
I’m in the camp that the market has been underestimating the real issue here. I run fixed income portfolios, and I do think you’re seeing a lot of demand destruction. To use a baseball reference, we’re in the second inning of a nine-inning game. A year out, this kind of price spike is going to cause inflationary pressures — from food to diesel, really across the board — and it’s going to be very recessionary. Some of that is also being priced into oil now: you’re starting to price in a recession a year to 18 months out. This kind of price shock is going to have a trickle-down that I think is underestimated right now.
On the global economy: With Jeffrey Snider bearish on China and George Gammon worried about the West, and demand destruction often overlooked, how does it look to him?
I focus on emerging markets, so the nice thing is there are always pockets that will actually benefit from this environment. Look at the oil and gas producers in Africa and Latin America — Angola, Nigeria, even South Africa, Colombia — I think they’re going to do well.
The big regions you touched on, specifically Europe, I think are in a particularly bad spot. From interest rates going up to losing their natural clients — Russia, remember, was where a lot of Mercedes-Benz and high-end luxury goods were sold, and that market’s now gone. There’s been a lot of animosity toward China too. And with the loss of Nord Stream and cheap Russian gas, the cost of production has gone up. You have Volkswagen announcing, I think for the first time since the Second World War, a 100,000-employee layoff. Qatar LNG identified Italy as one of the markets going into force majeure over the next 12 to 18 months. So Europe is in a particularly difficult area.
The US is a little bit more resourceful, more self-sustaining, so not as bad — but I do think this kind of shock is going to be recessionary. There had been chatter about the new Fed chairman raising rates; I think he’ll look forward and see this is recessionary, so I don’t see the US Fed raising rates this year.
That said, the nice thing in the markets I look at is you have different economic cycles. Argentina was upgraded yesterday by Moody’s. This kind of disruption in the Middle East will actually be beneficial to other parts of the world.
On who benefits most
From an asset allocation point of view, I’ve been very hesitant on the Middle East. So that second-tier and third-tier, higher-cost oil producers — again Angola, Nigeria, Colombian oil and gas, even Venezuela — are going to be big beneficiaries in this type of environment.
On investor sentiment toward the Gulf: Are investors delaying bets and pulling money out as the region’s security bubble bursts — and what changes if Iran makes good on its threat to hit Dubai and Abu Dhabi’s airports?
I was at a conference about three weeks ago, and I’ve been very bearish — I think this is going to have much longer-term repercussions than many of my peers think. It concerns me, frankly, about the Dubai banking center. You’ve seen money leave Dubai, go to Singapore, go to Hong Kong.
Even medium- to long-term, it’s interesting that you have Russia and China signing this Power of Siberia 2 deal. Once you get the Chinese construction reconnecting them for natural gas, one of the largest potential historic buyers of natural resources out of the Middle East will have a pipeline connection to Russia.
So from a medium- to long-term view it will put the Middle East in a difficult situation. They are one of the lower-cost producers today, but once you have pipeline gas, it’s very hard to compete against that as shipping, LNG, and bottlenecks come into play. So I’m a little bit more bearish medium-term on the Middle East.
On whether it’s existential for the Gulf states
Bahrain — you look at the debt metrics going into this pre-conflict, and they were challenging. I do think some of the smaller, not-as-financially-robust ones will be challenged here. I don’t want to get apocalyptic, but the numbers are challenging.
You have a part of the world that enjoyed a very nice, robust lifestyle, and when revenues are cut off, you’ve seen some non-market deals from individual countries with some large asset managers to increase liquidity. So not existential, but problematic — and frankly I don’t think Kuwait and Bahrain are in a good position at all.
On Trump and the Gulf: How much does he really care about the region, and how important is it to the US?
In terms of the actual global economy, I don’t think it’s as truly important as it was during the 70s. At this point, I worry it’s too much of an ego thing — that he has so much invested into it that he needs to show some type of victory.
I’m much more in the Joe Kent camp: let’s just walk away. Frankly, the American empire has been overextended for way too long — 750, 800 military bases overseas, and the next country has maybe five or six. Is it really that important to the US? I don’t think so. Unfortunately, I think Trump has invested so much political capital into it that it’s hard for him to walk away at this point.
On the Gulf’s importance to the world economy
Very important. One of the pieces I wrote right at the outbreak of this was about helium — I reduced my exposure in Taiwan. What is everything fueling? It’s chips: Nvidia chips, Taiwan Semiconductor. All these names require helium, and with that being curtailed, this has much bigger implications than a Russia-Ukraine conflict, because the Middle East is such an important resource supplier to the world — not just oil and gas.
On a prolonged conflict: Setting the energy choke point aside, could this drag Saudi Arabia, the UAE and Qatar into six to twelve more months of tit-for-tat strikes if the US keeps energy flowing?
That’s a big if. What’s been helpful is the East-West pipeline out of Saudi Arabia, and as you mentioned, that’s coming into doubt. Six months, twelve months down the road in the current chokehold is very, very problematic for the global economy.
It’s hard for me to envision the scenario you laid out of the US getting the energy out, because it’s such a small chokehold. In today’s world of drones, RPGs, MANPADs, fast boats, there are a lot of ways to be disruptive. So I don’t see how we can be in this situation six months to a year from now.
On Trump’s endgame: If threats don’t move Iran and Hormuz can’t be reopened, what is he actually doing?
Unfortunately, I think he’s found himself in a catch-22. I don’t see a great military option, and there will be a political cost to this that he has to come to grips with — whether it’s the Fox News crowd or his financial backers, who’ve been influential. He can’t get what he wants here. I think the Venezuelan scenario made him think this would be much easier, and he’s bitten off a little more than he can chew.
Six to twelve months out — the midterms are in November. Six more months of this market manipulation, or derivatives, or demand destruction, and you’re going to see oil prices higher than they are today. That will impact retail prices, and you have Thanksgiving, the run-up to Christmas, transportation. It’s going to be a very difficult political environment for him, and he’s setting himself up for a potential Democratic sweep — House, potentially Senate — and then the whole impeachment fiasco that bogged him down in his first term.
So I go back to Joe Kent: declare a clear victory and walk away. There’s no military path — I’m not a military person, but I don’t see one. There’s no magic weapon, no magic plan that solves this.
On the “taco” trade
I would say it is still a taco price. Given that we tried the MOU and that didn’t work out, and we’re now going after much more civilian-like infrastructure — desalinization plants — which is extremely problematic, and we’re still below the wartime level, I think the market is at least expecting some type of de-escalation at this point.
On the three-week surge: Oil ran from $70 to $94 in under three weeks — doesn’t a 25% move signal the market now doubts Trump will back down?
Be careful with just the percentage change from $70, because $70 was too low — it had fallen much more than it should have. It mispriced everything as going back to a pre-February conflict, which was not my base case.
You had a momentary 100 million barrels of pent-up supply — tankers that had been trapped in the Gulf and suddenly came out. The market looked at that and said, “Okay, we’re going to get 100 million barrels on a natural run rate.” But if you divide that by the number of days, it was roughly 9 million barrels, compared to about 20 million pre-conflict. So $70 was way too low for where the market realistically should have been.
Back to Roundup