All-In Podcast2025 Predictions: Tech, Business, Media, Politics!
Jason CalacanishostChamath PalihapitiyahostDavid FriedberghostGavin Bakerguest
CHAPTERS
- 0:00 – 5:38
Introductions, Ski Trip Banter, and Polymarket Partnership
The hosts welcome guest investor Gavin Baker, joke about his firm size and recent ski trip with the ‘besties,’ and introduce their partnership with prediction market Polymarket. They explain that select ‘super predictions’ from the episode will be turned into tradable markets, mainly for non-U.S. users.
- •Gavin Baker of Atreides Management describes his firm as a $4B crossover investor in public and private tech/consumer companies.
- •The group recaps their ski week, noting Chamath learning to ski in his 40s and friendly ribbing around skiing abilities.
- •They unveil an All-In section on Polymarket to list markets tied to their 2025 predictions, allowing performance tracking and community participation.
- 5:38 – 9:26
Biggest Political Winners of 2025: Fiscal Conservatives and a Younger Class
The panel revisits last year’s political calls and lays out who they think will win politically in 2025. Themes include fiscal discipline, centrist MAGA realignment, and a generational shift as younger leaders enter power.
- •Chamath names ‘fiscal conservatives’ as 2025’s biggest political winners, predicting austerity will reveal immense waste, fraud, and abuse in federal spending.
- •Friedberg says ‘young candidates’ (average age of Trump’s cabinet ~40–45 vs. ~60 under Biden) signal a durable shift toward younger officeholders.
- •Gavin frames the winners as Trump plus Gen X/elder Millennials (Elon, Sachs, JD Vance, Vivek, Rubio, Tulsi), arguing they’ll focus less on Boomer retirement issues and more on future generations.
- •They debate whether independent centrists actually swung the 2020 and 2024 elections toward Biden then Trump.
- 9:26 – 17:34
Biggest Political Losers: Putin, Neocons, Progressivism, and Extremes
Attention shifts to who will lose in politics: Putin, pro-war neoconservatives, and progressive identity politics. They also criticize small extreme factions on both left and right and discuss the changing Republican donor ecosystem.
- •Gavin picks Putin as biggest loser, arguing Xi will see Russia as a liability as Europe rearms, prompting China to decouple and Trump to cut a tough peace deal that leaves Russia humiliated.
- •Friedberg expects pro-war ‘neocons’ to lose to the JD Vance / Elon non-interventionist wing, though Gavin notes Trump may still use aggressive rhetoric tactically in negotiations.
- •Chamath names ‘progressivism’ as the primary political loser, citing likely rightward shifts in Canada, Germany, France, and the UK as backlash to identity politics and scandals like UK grooming gangs.
- •Jason targets the ‘racist vocal minority’ on both ends (DEI extremism vs explicit racism), predicting those fringes will be increasingly rejected.
- •They review last year’s call that Koch-network Republicans were big losers as MAGA replaced the old GOP donor machine; new power donors cited include Elon Musk and Miriam Adelson.
- 17:34 – 27:50
Biggest Business Winners: Robots, AI Full-Stack, and Stablecoins
The crew names their biggest anticipated business winners, converging on robotics and AI while diverging on specific vehicles like stablecoins and full-stack compute. They highlight how capital and infrastructure advantages will shape competition.
- •Friedberg declares 2025 ‘the year of the robot,’ spotlighting Unitree’s sub-$2K quadruped G02 and $16K G1 humanoid as proof that capable, API-addressable robots are now mass-market accessible.
- •Use cases for robots include scientific field work, autonomous farm monitoring, perimeter security, and even militarized applications (e.g., mounted weapons).
- •Gavin broadens the ‘robots’ theme to include Tesla FSD and AI agents; he argues big companies with capital to afford extended test-time compute will gain disproportionate advantage.
- •He predicts another GPU crunch and is bullish on NVIDIA and inference compute providers.
- •Chamath picks dollar stablecoins as biggest business winner, noting they processed ~$8.5T in Q2 2024—over double Visa’s volume—and have decoupled from crypto speculation.
- •He expects Trump to attack high credit card fees, accelerating stablecoin adoption and potentially quadrupling the market, while debate continues over stablecoins’ regulatory and geopolitical implications.
- 27:50 – 32:43
AI Platforms, Tesla/Google, and Stablecoins vs. U.S. Dollar Hegemony
They delve into how AI and payment infrastructure reshape business winners, with Jason championing Google’s Gemini Deep Research and xAI’s Colossus buildout, while Gavin warns of risks to U.S. monetary advantage if stablecoins become a reserve currency.
- •Jason touts Tesla, xAI, and Google as 2025’s big winners, praising Google's Gemini Deep Research as a step-function above rivals for complex analytic work (multi-step web research with citations).
- •He claims Gemini’s deep research outputs rival McKinsey/BCG-style reports and notes OpenAI’s $200/month o1 Pro as evidence of the high compute costs involved.
- •Gavin stresses that if a constellation of stablecoins displaced the dollar as global reserve currency, the U.S. would lose crucial advantages like borrowing in its own currency and inflating away real debt.
- •The group discusses stablecoins in enterprise payments, citing Jeremy Allaire’s pitch to rebuild Chamath’s and Jason’s payment rails to avoid 3%+ processor and platform fees.
- •They agree stablecoins are economically compelling but debate the national-interest tradeoffs.
- 32:43 – 46:54
Biggest Business Losers: Government Vendors, Legacy Defense, Mag 7 Concentration, OpenAI Risk
Predictions turn to business losers: government-heavy service providers, old-line defense and consulting, index-heavy mega-cap tech, and OpenAI’s valuation. The segment surfaces both business-model fragility and regulatory/political headwinds.
- •Gavin calls ‘government service providers’ with >35% of revenue from government the biggest losers, expecting fiscal conservatives to attack time-and-materials and cost-plus inefficiencies.
- •Friedberg extends this to ‘Defense 1.0’ (Boeing, Lockheed, Raytheon) and large consultancies/IT outsourcers (Accenture, Cognizant, TCS, etc.), predicting tech-native defense (Anduril, Palantir) and AI tools will erode their margins.
- •Chamath warns that despite their fundamental strength, the Mag 7/8’s extreme index concentration (~40% of market cap) makes them vulnerable to large absolute dollar drawdowns as capital rotates.
- •Jason names OpenAI as his biggest business loser, arguing its $150B+ valuation is unsustainable amid Google/xAI competition, open-source migration, court fights over its nonprofit-to-for-profit transfer, and Microsoft’s ownership of its code and compute stack.
- •They contrast OpenAI’s real-but-costly ~$12B projected revenue with meme-y bubble names like Truth Social and MicroStrategy, which Jason also sees as mispriced but easier targets.
- 46:54 – 57:04
Biggest Deals of 2025: Auto Megamergers, Robotics Buildout, Intel and AI Lab M&A, Waymo
The conversation moves to M&A and transformative deals, from auto consolidation and hardware buildout to AI lab acquisitions and potential ride-hail–autonomy marriages. They see pent-up deal flow returning after regulatory gridlock.
- •Chamath predicts a ‘collapse’ of traditional auto OEMs into mega-mergers, citing the Honda–Nissan agreement as the start of a wave driven by Tesla’s lead and Chinese EV competition; he describes many OEMs as ‘melting icebergs.’
- •Gavin mostly agrees, expecting Chinese OEMs to seize their domestic market and global share unless heavy protectionism intervenes; he notes governments may prop up ‘national champions.’
- •Friedberg foresees ‘blockbuster’ financing deals for U.S.-based robotics and advanced manufacturing, likely combining private capital and federal support to counter China’s dominance in drones and automation.
- •Gavin anticipates a tidal wave of general M&A after four quiet years, plus a major event around Intel and consolidation of independent frontier AI labs, since labs that rent cloud compute can’t be lowest-cost providers.
- •Jason speculates on mega-deals in autonomy/transport: Tesla buying Uber, Amazon buying DoorDash, or Waymo merging/partnering with ride-hail, particularly as Lina Khan’s antitrust era wanes.
- •Friedberg underscores Waymo’s traction: in ~15 months in SF, it reached ~22% market share (vs. Lyft ~22%, Uber ~55%), and is expanding to LA, Austin, and beyond on a cheaper hardware stack, suggesting another big financing, IPO, or strategic deal is likely.
- 57:04 – 1:12:59
Contrarian Beliefs: Bank Crisis, 5%+ GDP, Rising Socialism, and OpenAI Collapse
This segment is devoted to ‘most contrarian beliefs,’ surfacing tail risks and non-consensus macro views: a potential U.S. bank crisis, a productivity-driven growth boom, a resurgence of socialism, and Jason’s repeated prediction that OpenAI loses its lead.
- •Chamath posits a small but real chance of a major U.S. bank crisis in 2025, driven by the sheer dollar size of total debt (~$70T) at 5–6% rates, which replicates the pain of 10% rates decades ago and could trigger reserve shortfalls.
- •He notes total interest expense relative to GDP is what matters, not just the nominal rate, and suggests CDS as asymmetric insurance despite expecting to lose money in most scenarios.
- •Gavin’s macro-contrarian belief: the U.S. will see at least one year of >5% real GDP growth within four years as AI and deregulation supercharge productivity, potentially creating a late-1990s-style boom.
- •He also predicts leading AI labs will eventually stop releasing frontier models to the public to prevent knowledge distillation and IP theft (e.g., DeepSeek copying GPT-4-level performance).
- •Friedberg’s contrarian call: socialist movements will grow stronger in 2025 despite MAGA’s win, as AI/deregulation create sharp winners and losers, cut government jobs, and push some toward demands for universal healthcare, childcare, and broader safety nets.
- •Jason aligns, noting AI will not only displace blue-collar roles but also high-earning white-collar workers (developers, PMs, marketers), intensifying inequality. He reiterates his ultra-contrarian ‘total collapse of OpenAI’ as a 2025 possibility.
- •They also joke about Jason’s past wish that Jeff Bezos would run for president and float a new prediction that Bezos may sell The Washington Post in 2025.
- 1:12:59 – 1:22:56
Best and Worst Performing Assets: HBM, CDS, Chinese Tech, Mag 7, Enterprise SaaS
The group makes concrete asset-class calls. They highlight memory and China tech on the long side versus enterprise software, cost-plus contractors, real estate, and legacy autos on the short side, while debating index concentration risks.
- •Best-performing assets: Gavin chooses high-bandwidth memory (HBM) producers SK Hynix and Micron, highlighting that HBM is already a bigger cost component than TSMC in many AI accelerators and supply is chronically tight.
- •Chamath picks being long CDS (credit default swaps) as his ‘lottery ticket’—a low-probability but massive upside hedge against systemic credit or bank events, comparing it to Ackman’s COVID-era CDS trade.
- •Friedberg goes long Chinese tech ETFs/stocks (e.g., Alibaba), betting on a potential U.S.–China ‘great deal,’ China’s low power costs and industrial capacity, and a political throttle-up of entrepreneurship after years of crackdowns.
- •Jason backs the Mag 7 as best performers, arguing their internal use of AI will drive extraordinary earnings expansion as they automate and offshore more work before customers even see many AI benefits.
- •Worst-performing assets: Gavin and Chamath both hammer enterprise application software and the broader ‘software industrial complex’—big legacy SaaS/ERP platforms that wrap heuristics around CRUD databases and rely on golf-trip sales rather than product value.
- •Friedberg again singles out ‘vertical SaaS’ whose pricing power and moats will erode as companies build AI-native internal tools.
- •Jason adds legacy auto OEMs and real estate (especially overbuilt markets like Texas) as awful 2025 assets, given affordability crises and looming lease/loan repricings.
- 1:22:56 – 1:30:19
Most Anticipated Trends: AI Acceleration, Nuclear Buildout, and Exit/DPI Boom
They outline big structural trends they’re watching: rapid AI capability scaling, a nuclear renaissance to feed compute, and a resurgence in M&A/IPOs as regulatory pressure eases. Jason demos Google’s Gemini Deep Research as an example of AI’s new capabilities.
- •Chamath’s ‘most anticipated trend’ is actually the *absence* of regulatory can-kicking: he cites the supplemental leverage ratio and other arcane rules as potential levers to bail out banks or debt markets; if these are *not* relaxed, it’s a sign elected leaders (not the ‘deep state’) are truly in charge.
- •Friedberg expects 2025 to be the year nuclear power buildout becomes explicit U.S. policy, driven by the need to match China’s massive generation expansion and support AI’s power demands; he notes many smart people are launching nuclear startups in anticipation.
- •Gavin anticipates AI will make more progress per quarter in 2025 than per *year* in 2023–24 due to three multiplicative scaling axes: pretraining, test-time compute, and now reasoning (via synthetic ‘show your work’ traces and reinforcement learning).
- •He cites Arc-AGI benchmarks and Ilya Sutskever’s warning that reasoning models are inherently unpredictable, like AlphaZero’s non-human chess moves, implying we’re on a ‘straight shot to ASI.’
- •Jason live-demos Gemini 1.5 Pro with Deep Research, showing it creating a bank-risk report from 160+ web sources and concluding Citigroup and Wells Fargo are higher-risk, illustrating how AI agents can already synthesize extensive analysis.
- •Jason’s macro trend call: a surge in exits and DPI (distributions to paid-in-capital) as Lina Khan’s antitrust regime wanes and IPO and M&A markets reopen after several frozen years.
- 1:30:19 – 1:38:59
Media, Declassification, and AI-Native Entertainment
The hosts pivot to media they’re excited about in 2025, from prestige TV and AI-native games to the potential ‘content bomb’ of government declassification under Trump. They also revisit last year’s media predictions.
- •Jason’s ‘media’ pick is the internal turmoil and repositioning at legacy outlets like Washington Post, LA Times, and CNN as billionaire owners push them toward centrist or heterodox coverage, provoking newsroom revolts.
- •Chamath is most excited by the ‘enormity of the files’ he expects Trump to declassify—JFK, Epstein, Diddy, UFOs, and other long-suppressed topics—arguing this will produce an unprecedented media event.
- •Friedberg looks to AI-native video games, where generative content and dynamic storylines let studios make cheaper, more personalized, and innovative experiences, potentially drawing in non-gamers.
- •Gavin’s pick is non-AI: 1923 Season 2 from Taylor Sheridan’s Yellowstone universe, saying 1883/1923 hit him like Game of Thrones and praising Sheridan’s run that includes Sicario and Day of the Jackal.
- •They briefly review 2024’s accurate media calls, like MrBeast’s global success on Amazon and Demis Hassabis’s Nobel for AI—both aligning with prior predictions.
- 1:38:59 – 1:43:03
Polymarket Super Predictions: Immigration, National Debt, Mag 8, and Cloud Wars
The group designs specific Polymarket questions to track their boldest predictions, focusing on immigration, national debt, index concentration, and cloud provider growth. They debate appropriate thresholds that would attract two-sided betting.
- •Jason’s immigration market: Will Trump deport at least 750,000 people (5% of his stated 15M target) in his first year, using official government deportation stats as reference.
- •Chamath’s index-structure market: Will Mag 8 representation in the S&P 500 fall below 30%, effectively testing whether concentration unwinds in 2025.
- •Friedberg proposes a cloud-revenue race: which of Microsoft Azure, AWS, or Google Cloud will see the largest absolute revenue increase in 2025, noting Google’s higher growth off a smaller base.
- •They settle on a U.S. federal debt line: will the Treasury’s December 2025 federal debt report come in above or below $38T, a test of whether Trump can moderate the current ~$2T/year debt expansion.
- •They briefly debate how quickly immigration and debt policy can realistically change within a single year.
- 1:43:03 – 1:55:33
Bonus: UFOs, Drones, Pyramids, and Simulation Talk
In a long, speculative coda, Gavin raises UFOs and recent drone sightings, prompting a wide-ranging discussion about UAPs, historical accounts of extraterrestrial materials, Arthur C. Clarke’s vision of advanced civilizations, and whether the government is hiding proof of alien life.
- •Gavin references credible reporting in major outlets (NYT, WaPo, New Yorker, Atlantic) and declassified pilot testimonies on UAPs, plus historical lab accounts claiming to have seen extraterrestrial materials in the 1950s skunk works era.
- •He notes concentrated waves of UFO reports around key tech inflection points: nuclear weapons era (1945–1960) and now AI, speculating there may be a pattern linking dramatic shifts in human capability to elevated ‘observations.’
- •Chamath gives a ~20% probability the U.S. government is sitting on evidence of extraterrestrial life; Gavin takes the over at ~25%.
- •Friedberg is skeptical of ‘bodies in craft,’ arguing from an Arthur C. Clarke–style perspective that sufficiently advanced civilizations wouldn’t need to move biological bodies; they’d manipulate matter and gather information at a distance, making classic UFO narratives anthropocentric.
- •They humor the idea that Trump could release UFO files; Jason jokes that if Trump proves extraterrestrial life, he’ll back a constitutional change for a third term.
- •They also briefly discuss theories about how the pyramids were constructed, with Jason favoring hydraulic canal/float methods over alien engineering.
- •The episode ends with the usual All-In sign-off and banter about 2025 being a year to ‘kick ass.’
