All-In PodcastWhy Iran's oil shock is really about China's energy needs
Brent crude swings tracked the escalation. Goldman flagged inflation; Chamath argues China's oil exposure to Iran makes Beijing the true off-ramp lever.
CHAPTERS
- 0:00 – 2:50
Brad Gerstner returns: State of the Union shout-out and the “Trump accounts” rollout
Jason welcomes Brad Gerstner back and asks what it was like being recognized by the President during the State of the Union. Brad describes the behind-the-scenes experience and updates progress on the kids’ investment accounts initiative ahead of its July 4th launch.
- •Brad had no advance notice the President would mention him; it wasn’t in the prepared speech
- •Reflections on civic institutions and the State of the Union as a tradition
- •Update: 100,000+ kids/day signing up; millions claimed; ~30M eligible for at least $250
- •Goal framing: broaden ownership and participation in capitalism via kids’ equity accounts
- 2:50 – 3:48
A philanthropic ‘equity pledge’ idea to fund kids’ accounts
Jason proposes a voluntary giving pledge where founders commit a small portion of their equity to children’s accounts over time. Brad confirms the idea has come up and hints at upcoming announcements.
- •Jason connects the program to inequality and UBI-adjacent narratives
- •Proposal: tech founders pledge a small percentage of shares over years/decades
- •Potential impact: fractional ownership in top companies for millions of kids
- •Brad: “Stay tuned” for major announcements ahead of July 4th
- 3:48 – 6:14
Iran war market shock: oil volatility, historical context, and odds of escalation
The conversation pivots to the Iran war’s economic impact, focusing on Brent crude swings and the risk premium tied to the Strait of Hormuz. Jason cites historical oil shocks and prediction-market odds for U.S. involvement.
- •Brent crude whipsaws: mid-$80s to ~$119 and back near ~$100
- •Ships hit in the Strait; closure used as leverage by Iran’s leadership
- •Historical analogs: 1970s oil shock, Gulf War, 2008 peak, Russia/Ukraine spike
- •Polymarket probabilities cited for U.S. boots on the ground
- 6:14 – 10:22
Macro fallout vs. ‘Trump doctrine’: inflation, GDP drag, and market PTSD
Brad lays out Goldman’s revised inflation and growth expectations and argues markets may be overpricing a long conflict. He contrasts a pragmatic, limited-objectives approach with prior neocon ‘nation-building’ doctrine.
- •Higher expected PCE and core PCE; lower GDP forecast; higher unemployment risk
- •Equity multiple compression framed as sentiment/uncertainty driven
- •Argument: limited goals (degrade threats) vs. democracy promotion
- •Market ‘flashbacks’ to Iraq/Afghanistan creating quagmire fears
- 10:22 – 15:18
Off-ramp debate: declare victory, avoid escalation, and prevent Gulf catastrophe
Sacks and others argue for de-escalation, warning that escalation could trigger attacks on regional oil infrastructure and even desalination plants. They outline cascading scenarios: energy supply collapse, water insecurity, and broader humanitarian disaster.
- •Sacks: Iran’s military degraded—time to declare victory and exit
- •Risks: retaliation against Gulf oil/gas infrastructure even if the Strait reopens
- •Desalination plants as soft targets; water supply threat for ~100M people
- •Secondary risks: Israel sustaining heavy damage; possibility of nuclear escalation
- 15:18 – 19:18
Domestic politics and war duration: midterms risk and MAGA coalition strain
Jason argues that a prolonged conflict could be politically fatal for Republicans, accelerating Democratic momentum ahead of midterms and 2028. The group disputes framing, but agrees long wars tend to be unpopular and an off-ramp is essential.
- •Jason: prolonged Iran war undermines core ‘no new wars’ expectations
- •Claims of coalition backlash from prominent MAGA-aligned commentators
- •Midterm electoral risk: long war + inflation/unemployment shocks
- •Sacks: political instincts favor short, decisive action; resist neocon objective-creep
- 19:18 – 26:49
‘All roads lead to China’: energy dependence, summit leverage, and global pressure on Iran
Chamath reframes the war as fundamentally about China’s oil dependence and strategic bargaining. Brad adds that China’s lack of military response and continued summit plans are key investment signals, implying pressure for resolution.
- •Chamath: upcoming Trump–Xi meeting incentivizes a ‘grand bargain’
- •China’s oil exposure via Iran/Venezuela framed as severe vulnerability
- •Game theory: China needs stability more than the U.S. (U.S. produces/consumes ~20M bpd)
- •Brad: if the U.S. steps back, neighbors/Asia will pressure Iran to stop disrupting shipping
- 26:49 – 32:18
AI revenue ‘nuclear moment’: Anthropic and OpenAI scale at historic speed
The show returns to tech with eye-popping revenue run-rate figures for Anthropic and OpenAI. Brad argues recent model/agent improvements have shifted AI spend from IT budgets toward labor augmentation, driving explosive demand.
- •Reported run rates: Anthropic ~$14B; OpenAI ~$20B; unprecedented growth curves
- •Brad: crossing a threshold with newer models/agents (e.g., coding/agent platforms)
- •Claim: can’t reach multi-billion monthly revenue by displacing only IT budgets
- •Brad advocates IPOs: capital access, institutional demand, retail participation
- 32:18 – 43:49
Is enterprise AI revenue ‘experimental’ or production-grade? A heated reality check
Chamath challenges the durability and ROI of enterprise AI spend, arguing much is checkbox-driven experimentation rather than embedded in mission-critical workflows. Brad and Sacks respond that certain domains (especially coding) are already delivering strong value, even if broader transformation takes time.
- •Chamath: regulated industries can’t yet rely on AI without stringent controls/SLA
- •Example: Amazon tightening human review after AI-assisted code incidents
- •Brad: distinction between experimental run rate revenue vs. true ARR; notes some production use cases (gov/military/large enterprises)
- •Sacks: enterprise breakout is coding assistance; massive latent demand; change management slows broader rollout
- 43:49 – 47:49
The AI capex J-curve: gigawatts, data center economics, and time-to-profitability
The group debates when frontier model companies become profitable given massive infrastructure costs. Chamath provides a data-center build perspective, tying intelligence to energy and estimating payback periods per gigawatt.
- •Jason frames the profitability question via a ‘J-curve’ and long payback analogs (Amazon/Tesla/Uber)
- •Chamath: 1GW data center costs ballooning (multi-step increases) to ~$50B for full powered shell + infrastructure
- •Revenue heuristic cited: ~10B annual revenue per GW; ~5–6 year payback to break even
- •Potential curve-shrinkers: better silicon, efficiency gains, open source adoption
- 47:49 – 1:00:31
AI’s PR nightmare: doomer messaging, trust collapse, and regulatory backlash
Chamath argues AI leaders’ inconsistent public messaging—ranging from apocalypse to ‘tokens as a utility’—is fueling fear and political pushback. Brad and Sacks agree the sector is failing at communications, especially in the U.S., where sentiment is markedly worse than in Asia.
- •Messaging spectrum critique: ‘sentient doom’ vs. disruption warnings vs. token-utility framing
- •Poll/sentiment: U.S. far less optimistic than China/Asia about AI benefits
- •Sacks: fear narratives amplified by media + possible regulatory capture strategies
- •Consequences: proposed bans (e.g., AI legal/medical advice) disproportionately harm low-income users
- 1:00:31 – 1:04:59
Doomer think tanks and data center NIMBYism: protests, cancellations, and lost capacity
Sacks highlights well-funded ‘doomer’ organizations shaping discourse and local opposition to infrastructure. Chamath quantifies cancellations and warns that poor messaging is now materially reducing U.S. data center buildout, with Texas positioned as a beneficiary.
- •Claims: organized FUD around power prices, water usage, and data centers
- •Data point: significant share of protested data centers get canceled; acceleration since 2024
- •Chamath estimates multi-gigawatt cancellations translating into large foregone annual revenue
- •Geography: cancellations concentrated in Virginia/Indiana; Texas presented as more permissive for builds
- 1:04:59 – 1:07:39
Open source’s role and the ‘bigger-than-expected TAM’ thesis
Chamath presses Brad on open source and local model adoption potentially pressuring frontier labs’ economics. Brad argues open source is complementary—enterprises mix frontier planning with open-source execution—and the continued revenue surge implies an enormous total market.
- •Chamath: startups shifting a large share of tokens to open models; frontier models used for hardest jobs
- •Brad: enthusiastic about open source; enterprises pursuing ensemble strategies
- •Observation: frontier labs still add billions in revenue despite open-source progress
- •Takeaway: TAM expansion can support multiple winners (frontier + open ecosystems)
- 1:07:39 – 1:20:22
Washington’s ‘Millionaire Tax’ and the migration of wealth: Schultz, state math, and class-war politics
The episode closes on Washington State’s new tax and Howard Schultz’s move to Florida, expanding into why state-level wealth taxes often backfire. Chamath and Sacks argue these policies can drive out high earners, shrink the tax base, and inflame class warfare; Jason pivots to anti-fraud and pro-growth reforms as an alternative.
- •Policy: additional 9.9% tax for $1M+ earners starting 2029; estimated $4B raised; ~30k households affected
- •Schultz relocation framed as emblematic of high mobility among wealthy taxpayers
- •Chamath cites California simulations: negative NPV outcomes, overestimated gains, underestimated avoidance/migration
- •Sacks warns of normalization of federal wealth-tax proposals; Jason argues reforms in housing/education/healthcare could undercut socialist momentum
