Skip to content
All-In PodcastAll-In Podcast

OpenAI's GPT-5 Flop, AI's Unlimited Market, China's Big Advantage, Rise in Socialism, Housing Crisis

(0:00) Bestie intros! Gavin Baker, Ben Shapiro, and Phil Deutch join the show (7:32) GPT-5 underwhelms, benchmark saturation, OpenAI's product excellence (18:14) AI's TAM, ROIC, and energy implications (27:27) China's major advantage and how the US can counter it (43:46) Political parties picking winners, energy subsidy reversal, how the bureaucratic branch was created and how to fix it (52:38) Socialism, OBBB impact on energy, does socialism in the US rely on Mamdani in NYC? (1:10:00) Tariffs update: $125B+ revenue so far, India and Switzerland hit hard, Trump's ad hoc approach (1:27:51) Nvidia chips being smuggled to China (1:31:13) Apple's $700B in buybacks since 2015 and failing AI strategy (1:40:09) Summer reading recommendations Follow the besties: https://x.com/chamath https://x.com/Jason https://x.com/DavidSacks https://x.com/friedberg Follow on X: https://x.com/theallinpod Follow on Instagram: https://www.instagram.com/theallinpod Follow on TikTok: https://www.tiktok.com/@theallinpod Follow on LinkedIn: https://www.linkedin.com/company/allinpod Intro Music Credit: https://rb.gy/tppkzl https://x.com/yung_spielburg Intro Video Credit: https://x.com/TheZachEffect Referenced in the show: https://www.bloomberg.com/news/articles/2025-08-08/openai-s-gpt-5-met-with-mixed-reviews-confusion-in-first-day https://www.anthropic.com/news/build-ai-in-america https://www.youtube.com/watch?v=whE75zEFBpU https://www.census.gov/library/visualizations/time-series/demo/historic.html https://www.justice.gov/opa/pr/two-chinese-nationals-arrested-complaint-alleging-they-illegally-shipped-china-sensitive https://www.ft.com/content/6f806f6e-61c1-4b8d-9694-90d7328a7b54 #allin #tech #news

Jason CalacanishostBen ShapiroguestChamath PalihapitiyahostDavid FriedberghostGavin BakerguestPhil Deutchguest
Aug 9, 20251h 50mWatch on YouTube ↗

CHAPTERS

  1. 0:00 – 7:40

    Cold Open, Host Banter, and Guest Introductions

    The episode opens with light banter about nicotine, side hustles, and eBay before Jason Calacanis introduces the panel: David Friedberg, investor Gavin Baker, Daily Wire co‑founder Ben Shapiro, and energy investor Phil Deutch. They briefly plug the upcoming All‑In Summit, highlighting a high-profile speaker lineup and the event’s mix of tech, investing, and policy.

    • Calacanis jokes about nicotine patches and product ‘infomercials’ on the show.
    • Friedberg vents about co-hosts bailing last-minute on an earlier recording time.
    • Gavin Baker is introduced as CIO of Atrides, a public–private crossover tech fund.
    • Ben Shapiro plugs his upcoming book “Lions and Scavengers,” framed as builders vs destroyers in Western civilization.
    • Phil Deutch is introduced as an energy expert with deep ties to public service.
    • Quick promo for the All‑In Summit speaker lineup and scholarship tickets.
  2. 7:40 – 19:30

    Ben Shapiro’s Book, Journaling, and a Gen‑X Reality Check

    Ben explains that his new book grew out of personal journaling amid global turmoil, including post–October 7 debates at Oxford about Israel. The hosts riff on midlife angst, meditation, and the weirdness of Gen‑Xers watching societal norms flip, setting a tone of cultural unease that recurs throughout the episode.

    • Ben describes the book as a contrast between people who build and those who tear down Western civilization.
    • He admits journaling about debates where people justified October 7, leading to a more personal, diary-like book.
    • Jason and Ben joke about Trapper Keepers and adolescent nostalgia.
    • The Gen‑X hosts reflect on using journaling and meditation to process a ‘topsy‑turvy’ world.
  3. 19:30 – 32:50

    GPT‑5 Launch: Underwhelming Upgrade and UX Innovation

    The group dissects OpenAI’s GPT‑5 release, which followed quickly on the heels of xAI’s Grok‑4. While GPT‑5 tops some benchmarks, Gavin emphasizes it doesn’t decisively beat Grok‑4 on the hardest tests, a first for OpenAI. Friedberg argues the truly meaningful advance is OpenAI’s automatic model-routing UX, which hides complexity from end users.

    • Jason calls the GPT‑5 keynote messy: flat/wrong charts, typos, lack of Jobs‑style polish.
    • Gavin explains benchmark saturation and introduces Humanities Last Exam and RKGI‑2—on which Grok‑4 slightly outperforms GPT‑5.
    • He notes his fund is an xAI shareholder and admits bias while still stressing data-based observations.
    • Friedberg highlights GPT‑5’s router that auto‑selects models based on query complexity as a major UX leap.
    • Gavin mentions the router was broken for the first 12–16 hours, suggesting early performance was suboptimal.
    • They agree OpenAI’s product and distribution might keep it highly valuable even if it loses pure model supremacy.
  4. 32:50 – 40:00

    AI Economics: Returns on Capital, Productivity, and the TAM Debate

    Ben Shapiro asks whether AI’s productivity gains will arrive fast enough to justify enormous capital spending. Gavin answers by pointing to rising returns on invested capital at public hyperscalers, AI‑driven revenue acceleration (Meta, Microsoft), and the difference from the dot‑com era’s ‘dark fiber’ overbuild. The panel frames AI as silicon replacing human OPEX, already showing real economic impact.

    • Ben notes that in the dot‑com era, productivity lagged investment by years and wonders if AI will be similar.
    • Jason references estimates of $50–$100 of AI spend per business user, yielding a potential multi‑trillion TAM.
    • Gavin says ROIC has actually risen for big AI investors, with Meta’s AI ad tools and Microsoft’s Copilot driving real revenue impact.
    • He contrasts AI GPU utilization (clusters melting from use) with underutilized dark fiber from 1999–2001.
    • They emphasize AI’s early returns are visible sooner than the internet’s were at a comparable stage.
  5. 40:00 – 48:00

    AI’s Power Hunger: 50 GW Demand, SMRs, and China’s Buildout

    Phil Deutch and David Friedberg dive into AI’s colossal electricity appetite and its consequences for energy policy. Anthropic wants 50 GW of US power in three years—about an entire year of national capacity addition—while China adds ~1 TW every 18 months and leads in solar and nuclear. The panel debates US subsidies for renewables vs nuclear, regulatory overhang, and how hyperscalers’ carbon‑free demand is reshaping the grid.

    • Phil cites Anthropic’s request for 50 GW as roughly 5% of US power, equivalent to a full year’s capacity additions.
    • Hyperscalers want continuous, location-specific, largely carbon‑free power and are relatively price‑insensitive.
    • China’s incremental electricity capacity is dominated by solar (on an additions basis) and a massive nuclear ramp.
    • Phil argues rule of law and contract reliability are prerequisites for sustained energy innovation; China’s lacking here.
    • Friedberg warns that US solar/wind subsidies distorted capital allocation away from nuclear; Phil counters that nuclear’s main problem is regulation, not subsidy competition.
    • Both agree SMRs plus AI‑driven demand and reduced bureaucracy are finally pulling real money into advanced nuclear.
  6. 48:00 – 1:00:00

    US–China AI Race: Corporatism, TikTok, BYD, and Existential Stakes

    Ben frames the AI race with China as existential, citing TikTok as a live psyop and potential future control over AI tools used by US businesses. The panel explores China’s ‘state-sponsored corporatism’—subsidizing specific sectors like EVs, social apps, and coffee chains—as an alternative model that can outcompete the US in narrow domains. Friedberg argues AI is more like the internet: a broad platform with many competitive vectors, not a single ‘winner takes all’ race.

    • Ben stresses AI’s dual-use nature: economic and military; a Chinese lead would affect freedom of the seas and US allies.
    • He backs Trump-style tariffs as justified pushback against China’s targeted industrial policies.
    • Jason cites Luckin Coffee and BYD as examples of China using capitalism to beat Western firms at their own game.
    • Friedberg says no one ‘wins AI’ globally, but advantages in drones, supply chains, and manufacturing can be decisive.
    • Gavin describes DeepSeek engineers allegedly being barred from meeting Americans inside China, underscoring CCP sensitivity.
    • All agree China is treating AI as an existential competition; the US must treat it at least as seriously.
  7. 1:00:00 – 1:10:00

    Entrepreneurship vs Central Planning: China’s 2018 Peak and Its Limits

    Friedberg walks through data on China’s company formation peaking in 2018 before plummeting amid crackdowns on tech entrepreneurs and private education firms. Gavin adds examples from the late Soviet Union to show capitalism’s superiority under resource constraints. Phil and Ben emphasize the need for rule of law and stable policy for long-term innovation, highlighting how CCP overreach undermines its own economic miracle.

    • Chart: Chinese new company creation peaked ~2018, then fell sharply as the CCP cracked down (e.g., Jack Ma, tutoring bans).
    • Brookings research suggests China’s poverty escape was driven by peasant entrepreneurship more than central planning.
    • Gavin notes Xi recently met entrepreneurs, with pride of place given to ‘real economy’ and AI, but social media de‑emphasized.
    • Phil cites his own wind blade factory experience in China: daily contract renegotiations and control issues over the corporate chop.
    • Consensus: China can mobilize quickly but stifles bottom‑up innovation by attacking successful private actors.
  8. 1:10:00 – 1:25:00

    Imperial Presidency, Administrative State, and Federal vs State Power

    The conversation pivots to US governance: why power is centralizing in the presidency and administrative state, and what that means for AI regulation and industrial policy. Ben Shapiro traces the trend back to the early 20th‑century administrative state, where Congress offloaded hard decisions to regulators. He argues the real future of reform lies in powerful state governments (e.g., Florida vs California), not federal heroics.

    • Gavin’s parents argue the ‘imperial president’ arose because Congress no longer legislates effectively; Ben agrees and goes deeper.
    • Ben blames Congress for creating agencies, then dodging accountability by delegating broad rule‑making powers.
    • He suggests the Founders anticipated branches jealously guarding power, not deliberately offloading it.
    • Federal preemption of state AI laws could be good if it blocks extreme state-level authoritarianism (e.g., California).
    • Ben advocates building an ‘America Party’ from the bottom up—winning governorships and state legislatures—rather than chasing the White House alone.
  9. 1:25:00 – 1:39:40

    Socialist Drift, Cost of Living, and the Housing Crisis

    Friedberg delivers a stark warning: socialism is likely to spread in the US because nearly half the workforce is government-paid, costs have exploded, and political answers are always ‘more government’. Ben underscores that failed policies rarely cause retraction; politicians simply push more aggressive versions. Jason and Ben argue that housing deregulation, new city-building, and encouraging mobility would do more to defuse socialist sentiment than any ideological fight.

    • Friedberg cites inflation in groceries, rent, and home goods as near‑universal pain points, even for the affluent.
    • He claims ~50% of Americans are directly or indirectly employed by government, fueling dependency and inefficiency.
    • Policy failures (e.g., rent control, food stamps) typically lead to bigger programs, not rollbacks.
    • Ben notes cities like LA and San Francisco keep electing left governance despite decades of poor results; socialism ratchets.
    • Jason argues solving housing—via deregulation and massive building—would lower anxiety and weaken redistributionist politics.
    • They highlight collapsing geographic mobility as a cultural shift from the traditional American willingness to move for opportunity.
  10. 1:39:40 – 2:03:40

    Tariffs: Revenue Boon, Trade War Risk, and Uncertain Endgame

    The hosts analyze Trump’s aggressive tariff regime, which has pushed US average tariffs above Smoot‑Hawley levels for the first time in a century. They discuss polymarket odds on tariff revenue, impacts on inflation and demand, and how tariffs clash with goals like re‑shoring. Ben and Gavin stress Trump’s flexibility—he backs off when markets react badly—but warn that policy volatility is corrosive for long-term investment.

    • Tariff revenue is annualizing above $300B; prediction markets see $200–$500B in 2025.
    • Ben warns that lack of immediate inflation may reflect falling demand, not benign effects.
    • Tariffs to punish India for buying Russian oil illustrate the mix of geopolitical and economic motives.
    • Phil sees globalization as a net positive historically but is unsure if the security‑driven deglobalization trend will pay off.
    • Gavin emphasizes that stability is crucial: Trump likes headlines and change, but investment thrives on predictability.
    • Several panelists put the odds of tariffs ‘working’ vs ‘backfiring’ around 50/50, underscoring genuine uncertainty.
  11. 2:03:40 – 2:10:00

    GPU Smuggling to China: Export Controls Meet Black Markets

    They briefly cover DOJ arrests of two Chinese nationals accused of smuggling high-end NVIDIA GPUs to China via Singapore and Malaysia. Gavin argues that given the value of these chips, smuggling is inevitable, but the reported ~$1B volume is too small relative to US hyperscaler clusters to be strategically decisive. He likens NVIDIA’s despec’d H20/B50 chips sold into China to ‘legalized marijuana’ versus hard drugs.

    • DOJ alleges ALX Solutions re-routed state-of-the-art GPUs from the US to China via third countries.
    • FT estimates at least $1B of GPUs smuggled in three months; chips fetch a ~50% black-market premium.
    • Gavin says we can’t keep illegal drugs out; we won’t fully stop high‑value chips either.
    • A billion dollars of GPUs is meaningful but tiny compared to multi‑billion US AI clusters.
    • US strategy of selling down‑specced chips to China is compared to legalizing low‑potency cannabis to blunt the black market.
  12. 2:10:00 – 2:22:00

    Apple’s $700B Buybacks vs AI, Data Centers, and Innovation

    The episode wraps with a critique of Apple’s decade-long $700B+ share buyback spree, contrasted with its relatively thin product innovation (AirPods, M‑chips) and absent AI strategy. Gavin is baffled by Apple’s poor Siri and AI experience and argues they should have diverted a slice of buybacks into massive data center and AI efforts. Friedberg cautions against betting against Apple’s system, while Jason points to an opening in AR, home energy OS, and AI‑native devices.

    • Chart shows Apple buying back more stock than the entire market cap of all but 12 S&P 500 firms.
    • Jason lists what Apple could’ve bought instead: Disney, Netflix, Tesla, Uber, Robinhood, BMW.
    • Gavin says investors love buybacks but sees a missed opportunity to fund data centers and real AI competition.
    • He calls Apple’s AI products ‘terrible’ and “almost like they’re trying to lose.”
    • Friedberg reminds everyone that people have been predicting Apple’s demise for 30 years; the iPad was mocked at launch.
    • Phil flags Apple’s potential to dominate AR glasses and home energy/device orchestration given their ecosystem strengths.
  13. 2:22:00

    Recommendations and Sign‑Off: Books, Shows, and All‑In Summit

    In a lighter close, the hosts and guests share summer reading and viewing picks—from sci‑fi like Project Hail Mary and Moonfall to business biographies and TV crime series. Jason plugs All‑In Summit scholarships and teases big party plans. They end on a mix of optimism about American resilience and concern about policy experiments that could shape the next decade.

    • Ben recommends Netflix’s ‘Department Q’ and laments the lack of great new Broadway musicals.
    • Gavin mentions the post‑apocalyptic ‘Moonfall’ series and classic fantasy like ‘Empire of the East’.
    • Friedberg reads Andy Weir’s ‘Project Hail Mary’ but nitpicks its scientific assumptions.
    • Phil cites the Musk biography, a book on the King James Bible, HBO’s ‘The Penguin,’ and ‘Dope Thief.’
    • Jason plugs Monocle magazine and several biographies: Mel Brooks, Barry Diller, Larry Charles, Keith McNally.
    • They invite listeners to apply for discounted Summit tickets if genuinely unable to afford full price.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.