SpaceX Financials, Does AI Increase Unemployment or Leisure, Chimp Civil War | Diet TBPN
TL;DR
SpaceX’s core business looks wildly profitable, but xAI drags the combined picture deep into the red — The Information reports SpaceX generated $18.5 billion in revenue and nearly $8 billion in EBITDA from launch + Starlink, yet posted an almost $5 billion net loss after folding in xAI and roughly $13 billion of AI-related capex.
The hosts frame the coming SpaceX/xAI IPO as classic Musk financial engineering — their read is that investors buying into a dominant space-and-telecom business will also be financing Elon Musk’s still-unproven AI ambitions, much like Tesla cash flow funding the next big bet.
Alex Tabarrok’s sharpest AI labor point is that ‘40% unemployment’ and a ‘three-day work week’ can describe the same amount of work — he argues the real question is distribution, not just productivity, citing U.S. annual work hours falling from about 3,000 in 1870 to 1,800 today without society collapsing.
The show treats AI’s labor future as a policy problem as much as a technology problem — Tabarrok’s suggested ‘AI dividend’ is basically more shared leisure via holidays or a shorter week, while critics warn that even welfare-improving change still creates painful losers, from rust-belt decline to ‘deaths of despair.’
A chimpanzee ‘civil war’ in Uganda became the episode’s weirdly gripping metaphor for human institutions breaking apart — researchers tracked a once-cohesive group of roughly 200 chimps that split after key males died and a new alpha rose, leading to more than 24 deaths in coordinated attacks by 2021.
Ferrari closes the episode as a case study in manufactured scarcity done right—until maybe not — recapping a Wall Street Journal piece from Acquired’s Ben Gilbert and David Rosenthal, the hosts note Ferrari sold only 330,000 cars in its lifetime yet built 400 million fans, though one host says recent models like the F80 make it feel like the company is ‘making too many cars’ and losing its way.
The Breakdown
Artemis splashdown opens the show with real space-drama energy
They kick off on NASA’s Artemis 2 return, calling re-entry and splashdown the “13 minutes of things that have to go right.” The hosts lean into the immediacy — 24,000 miles per hour, 5,000 degrees Fahrenheit, a Pacific splashdown off San Diego at 5:07 p.m. Pacific — and one of them says it feels like the astronauts “just left.”
SpaceX’s numbers are strong — until xAI enters the spreadsheet
From there, they jump to The Information’s report that SpaceX brought in $18.5 billion in revenue but posted nearly a $5 billion loss after consolidating xAI, which SpaceX acquired in February. Their tone is basically: of course — SpaceX is a very profitable company that bought a very unprofitable AI company, and now IPO investors will be underwriting Musk’s next moonshot.
The expensive part of the future might be chips, not rockets
The most striking number for them is capex: nearly $13 billion for chips and data centers, about 50% more than the rocket and satellite divisions combined. They joke that you’d assume rocket factories would be the expensive thing, then compare AI infrastructure to buying ASML lithography machines that can run around $400 million each.
Launch demand is still outrunning supply
Even with the xAI drag, they emphasize that SpaceX’s core engine is humming: nearly $8 billion in EBITDA from launch services and Starlink in 2025. They tie that to broader market demand — saying companies can’t get enough launch capacity, referencing a founder they spoke to building a new low-Earth-orbit GPS system and teeing up Firefly Aerospace’s Jason Kim as a future guest.
xAI’s censorship fight and model-building scramble
They briefly hit two xAI side plots: David Sacks praising the company for challenging a Colorado AI law he says could require censorship of truthful answers, and Wired’s report that xAI approached Black Forest Labs to license image tech but got turned down. Their read is that Black Forest, a 70-person startup, may have decided it didn’t want to be both partner and future prey while xAI trains its next image model at Colossus 2.
Is AI mass unemployment — or just more weekends?
The labor section centers on economist Alex Tabarrok’s argument that 40% unemployment and a three-day work week are, in raw hours, nearly the same outcome. The hosts clearly enjoy the mind-bender, walk through the math, and highlight his historical point that U.S. work hours fell from roughly 3,000 a year in 1870 to 1,800 today, with more childhood, retirement, and leisure along the way.
Then the episode takes a left turn into chimp geopolitics
In a very “slow news week” pivot, they unpack a Wall Street Journal piece on a deadly chimp split in Uganda’s Kibale National Park. What makes it stick is the human framing: a 20-year social order unravels after bridge-building males die, a new alpha rises, the group fractures, and former allies start carrying out coordinated lethal raids — which the hosts immediately read as a metaphor for tech industry factions and HOA politics.
Ferrari as the luxury brand everyone knows but almost nobody owns
They end with a long riff on Ben Gilbert and David Rosenthal’s Wall Street Journal essay about Ferrari. The hosts love the core insight — Ferrari isn’t just Hermes-on-wheels, it’s scarcity fused with sports fandom — and marvel that the company has sold only 330,000 cars in its lifetime versus Ford’s 2.2 million in the U.S. last year alone, before closing on a personal note that Ferrari may once again be overproducing and drifting from what made the brand magical.