Ideas
Power is AI's binding constraint
Chamath came out of the DC AI summit convinced that electricity, not chips or capital, is the binding constraint on AI: in the absence of power, AI is not going to be the thing we think it can be. He argues that constraint will create an enormous appetite by the federal government to do deals in energy and power generation and get players on the field, and that the combination of Trump's AI speech, the executive orders and the clarity of the budget bill gives operators in these markets a long runway to execute, which is why he came away risk-on.
Fix regulation to accelerate US nuclear
Friedberg says energy is the biggest unsolved issue in America besides the federal deficit and the debt, and that the bottleneck for nuclear is regulation rather than technology: he personally thinks the US needs to fix the regulatory roadblocks in nuclear, and notes Energy Secretary Chris Wright is spending most of his time on accelerating nuclear deployment. He wants a deep dive on where the US actually is in the cycle of nuclear build-out, making it the energy theme he is monitoring most closely.
Nat gas scales US power fastest
Friedberg argues natural gas is the only realistic way for the US to scale power generation from one terawatt to two faster than currently projected, which is what the AI build-out requires. His reasons are cost and speed: methane plants run about half the cost of solar and solar is roughly twice the cost of nat gas, a gigawatt can be stood up in under two years, the footprint is about 20 acres versus 4,000 acres for solar, methane burns cleaner than coal and oil, and the US already has abundant domestic supply, which is what makes nat gas a reliable source right now.
AI content licensing deals have peaked
Chamath reads the New York Times' roughly $20 million a year Amazon content licensing deal as the peak of AI content deals rather than the start, arguing the dollar value of such deals only goes down from here. Taken to the limit, he thinks enforceable copyrights and patents are a very fragile assumption in five years once models can independently derive patented material or two competing AIs invent the same thing from scratch. He says he is underwriting the value of these rights to zero for his own businesses, which is why Beast deliberately kept its OpenAI training licence short, and is focused on real defensible moats rather than a piece of paper to sue with, making content owners whose upside depends on licensing revenue unattractive.
AI licensing deals multiply, helping NYT
Jason takes the other side of Chamath's bet and guarantees the Amazon deal is the beginning of AI content licensing, not the peak. He argues the New York Times signed it to establish a paying customer and therefore demonstrable damages against OpenAI, which could support an injunction and force expensive removal from the training corpus, and that $20 million a year is about 1% of Times revenue dropping straight to the bottom line. He expects the Times to sign roughly ten such deals, funding more journalists and fact-checkers, with gated real-time content becoming a distinct advantage that model makers have to pay for.
Post-tariff growth strong; Fed should cut
Chamath says Powell is deliberately smoothing the first half to justify a political decision, when Q1 and Q2 must be segregated because Q1 was pre-tariff and Q2 post-tariff. The Q2 run rate, a 3% print with a large surplus and moderating inflation, is what Q3 and beyond should look like, so on the data alone the Fed should be cutting; refusing to cut is the only remaining lever to slow the administration into the midterms. He argues that injecting a 100 basis point cut would send the economy gangbusters into 2026, and says he is extremely risk-on right now.
EU deal is huge US stimulus
Sacks calls the 3% Q2 GDP print way ahead of expectations, with everything humming on all cylinders, and says the bigger story is the new EU trade deal: Europe opens its market to US products at zero tariff while European goods pay 15% coming into the US, plus $600 billion of European investment, $750 billion of US energy purchases and hundreds of billions of American defense products on top of the NATO commitment to 5% of GDP. He adds it up to roughly $2 trillion, effectively stimulus into the US economy over about three years without money printing, and argues the tariff-threat strategy is demonstrably working as counterparties capitulate and markets sit at all-time highs.
Trade cash may fund nuclear buildout
Jason says he asked Commerce Secretary Lutnick where the $600 billion from the EU and $550 billion from Japan actually go, and was told it is at the president's discretion and that Lutnick is advising him to spend it on nuclear. That would put roughly a trillion dollars of foreign money behind new nuclear power plants and small modular reactors, which Jason calls absolutely brilliant, while cautioning that these are still handshake deals that have to be inked, approved and executed before any of it lands.
This All-In Podcast video, published August 01, 2025,
features Chamath Palihapitiya, David Friedberg, Jason Calacanis, David Sacks
discussing Power infrastructure, URA, NYT, SPY, US energy exports, Small modular reactors.
8 trade ideas extracted by AI with direction and confidence scoring.
Speakers:
Chamath Palihapitiya,
David Friedberg,
Jason Calacanis,
David Sacks
· Tickers:
Power infrastructure,
URA,
NYT,
SPY,
US energy exports,
Small modular reactors
