2025 Predictions: Tech, Business, Media, Politics!
2025-01-04 spoken.md · speaker-labeled ▶ watch ← E208 all episodes E210 →
Every number here is replayed from score_events — the same ledger the pool ranks on. Decay is what the 138 ideas nobody mentioned gave up this week; it applies only when an episode is processed.
Tier crossings
conviction thresholds crossed by this episode — 65 / 45 / 15 · ideas born here show where they landed. why 65 isn't always green
Kill dates that landed since E208
3 hit · 0 partial · 1 miss — windows that closed after 2024-12-20 and up to 2025-01-04, auto-scored against price data and never hand-set. verdict · R · α
| idea | verdict | R | α | closed |
|---|---|---|---|---|
| 🪙 Solana's post-FTX recovery runs - the besties are still holding | HIT | +78.5% | +51.7 | 2024-12-29 |
| 🏛️ Third-party candidates take a Perot-plus share of the 2024 vote, at Biden's expense | HIT | +106.2% | +79.4 | 2024-12-29 |
| 🌍 Red Sea blockade spikes freight rates and re-ignites goods inflation | HIT | +154.2% | +127.6 | 2024-12-23 |
| ⚡ The 2024 black swan is a wider Middle East war and an oil shock | MISS | +4.1% | -22.6 | 2024-12-23 |
Who moved the board
each voice's force on conviction this episode — supports and opposes, weighted exactly as the replay applied them · share = % of this episode's movement
What got argued (14 ideas)
ordered by how hard each idea moved · quotes are verbatim from the transcript, timestamps deep-link into the episode
After 2024's compute buildout, 2025 is the year of autonomous hardware: humanoid/quadruped robots (Unitree at $1,600-16,000), FSD crossing into mainstream adoption, robotics 'slowly then all at once'.
2025 will be the year of the robot... [Unitree's] GO2 robot is $1,600, has an API... we just placed an order today... It takes a long time for things to work, and then all of a sudden they happen faster than you could have ever imagined... this could be the year of the big rollout.
Profoundly agree — the year of the robots inclusive of FSD. I think FSD works today and it's going to cross into mainstream adoption... it's going to continue compounding at an accelerating rate.
Agents from the labs/clouds replace per-seat enterprise application software; AI-native competitors are built an order of magnitude cheaper and price at $10 what incumbents sell for $100 — fissures show in the 'software industrial complex' (and vertical SaaS pricing compresses further).
Enterprise application software... 2025 is going to be the year of agents... if the labs and the big cloud providers dominate agents, which seems likely, enterprise application software is going to be in a lot of pain... they don't have their own models, they don't own their own compute.
The software industrial complex — large, bloated enterprise software companies... wrapping heuristics around a CRUD database... you're going to start to see fissures in those businesses in 2025. [Disclosed skin: founded 8090 to attack exactly this — 'my 30% engineering team does the work of 300 people... you show up and say $10' where incumbents charge $100.]
Anyone with >35% of revenue from the US government — service providers, cost-plus contractors, and the consulting complex (Accenture, Cognizant, Wipro, Tata) — gets scrutinized and squeezed as DOGE audits the bill.
Government service providers. You do not want to have the United States government at any level as over 35% of your revenue.
Would you add the traditional consulting companies to the mix? The Wipros, the Tatas, the HCLs, the Accentures, Cognizants... all in real trouble... Cost-plus is a fancier way of saying human labor arbitrage. All of that stuff will get extremely scrutinized in 2025.
Honda-Nissan is the signal: legacy OEMs are melting icebergs, uninvestable next to Tesla and Chinese makers — 2025 brings the realization and a wave of auto mega-mergers, with European OEMs (Volkswagen, Stellantis) in real trouble.
This is the year we will see the collapse of the traditional auto OEMs... a massive wave of consolidation... the European OEMs are in real trouble. What does Volkswagen do? What does Stellantis do? These are melting icebergs... it will trigger the realization that these auto OEMs are uninvestable.
I agree 100% with Chamath. They're going to lose their Chinese business because they don't make competitive products anymore... they'll be caught between Tesla and the Chinese OEMs... absent really significant government support, they're all in deep trouble.
O3-style test-time compute plus reasoning models mean GPUs/accelerators sell out in 2025 the way they did in 2023 — inference compute is the derivative winner; 'long NVIDIA and chip makers.'
Inference compute is going to be the derivative winner... we're going to run out of GPUs, accelerators, compute in 2025 the same way we did in 2023. [Jason: 'So you're also long NVIDIA and chip makers?'] Absolutely.
HBM is a bigger share of Nvidia GPU COGS than TSMC, only SK Hynix and Micron can make it, it's been sold out for two years, and test-time compute makes it more critical than ever.
The companies that make high bandwidth memory... HBM is a bigger part of NVIDIA's COGS on GPUs than Taiwan Semi is. Today there's two companies that can make it, Hynix and Micron... it's been sold out for the last two years. So high bandwidth memory would be my pick.
Everyone dumped Chinese tech; a Trump-Xi grand deal, massive cheap electricity buildout, and the CCP throttling entrepreneurship back up make beaten-down, single-digit-multiple Chinese tech (Alibaba etc.) poised for a big 2025 run.
I went with Chinese tech stocks, Chinese tech ETFs... they're trying to line up the great deal with China... cost per kilowatt hour is already lower... these stocks are pretty cheap. I was just looking at Alibaba... poised for a pretty good run in 25 if the macro works out.
Really cheap... I tend to agree with a lot of what David said. I think Trump and Xi both want a deal... there are really, really high quality companies, some of them at mid-single digit multiples. [But he keeps a personal 'no-China guideline' since the Longtop fraud.]
Stablecoins decoupled from crypto volatility ($8.5T H1-24 volume, 2x Visa), crossed the point of no return, and quadruple or quintuple in 2025 while finally attacking the Visa/Mastercard duopoly.
The biggest business winner of 2025 are going to be dollar-denominated stablecoins... $8.5 trillion of transaction volume — more than double Visa's over the same period... we're going to finally attack the duopoly of Visa and MasterCard... stablecoins could quadruple or quintuple by the end of 25.
Gemini deep research is 'a step function higher' than anything in market; Google's data + distribution (YouTube, Gmail, Drive) produce things nobody anticipates — Google among 2025's biggest winners.
I am just absolutely amazed at the comeback that Google has made with AI... Gemini deep research... blows anything else in the market out of the water, period, full stop. It is a step function higher... Google is going to come out with things people didn't anticipate.
Record consumer debt, unaffordable mortgages, and two straight years of falling Texas prices/rents where building is allowed — real estate is among 2025's worst assets.
Defense budgets shift to tech-oriented, ROI-driven spend (Palantir/Anduril world); the clunky cost-plus primes — Boeing, Lockheed Martin, Raytheon — are deeply challenged in 2025.
I went with the old defense and aerospace providers — Boeing, Lockheed Martin, Raytheon... defense budgets are going to shift towards a more tech-oriented and ROI-driven rationalization... this government contracting business is going to be deeply challenged this year.
Total Pax-Americana indebtedness at ~5% rates equals the pain of 10% rates a generation ago; a mark-to-market or credit-default shock triggers a reserve issue at a major bank — low probability, huge asymmetry, expressed as long CDS ('92 times out of 100 this goes to zero').
I think you're going to see a banking crisis in one of the major mainline banks... 5% rates today on $70 trillion is equivalent to 10% rates 25 or 30 years ago... I would be long CDS... 92 times out of 100 this goes to absolute zero... I hope I'm completely wrong. [Explicitly framed as low-probability insurance.]
Deregulation plus AI-driven power demand make US nuclear buildout an inevitability announced in 2025; smart people are quitting good jobs to start nuclear companies — the leading indicator.
My most anticipated trend is the announcement of buildout of nuclear power in the United States in 2025... nuclear is an inevitability... I know a lot of very smart people who have left very good jobs to start nuclear power companies in anticipation of this happening in 25. So I'm very bullish.
Top-7/8 concentration approaching 40% of the indices historically foreshadows a big drawdown; independent of fundamentals, even a 10% Mag-8 retrace is trillions of dollars. Super prediction: Mag-8 share of S&P falls below 30%.
The absolute dollar drawdown of the Mag-7 will be in the trillions of dollars... concentration of the top seven, eight companies in the indices is approaching 40 percent... these historic concentrations have generally foreshadowed a big drawdown. [Super prediction: Mag-8 representation on the S&P 500 shrinks below 30%.]
I think the Mag 7 is going to be the best performing asset. I'm taking the other side... the internal application of AI will allow these companies to have earnings growth that people will not be able to comprehend over the next couple of years.
Episode digest
written during extraction and stored in data/extractions/ep209.json — the auditable source of truth, including everything market-adjacent that did not earn a capture
The 2025 predictions episode — Sacks absent (newly in the White House), Gavin Baker guests as fourth chair. Consensus theme: AI eats incumbents — Gavin's enterprise-application-software pain call gets doubled by Chamath ('software industrial complex', with his 8090 skin disclosed) and tripled by Friedberg (vertical SaaS again); Gavin adds HBM/Micron and a 2023-style compute shortage ('long NVIDIA — absolutely'); Chamath calls stablecoins the year's winner with 2x-Visa volume receipts and re-attacks the V/MA duopoly. Genuinely contested: Chamath's Mag-7-drawdown-in-the-trillions (super prediction: Mag-8 below 30% of S&P) vs Jason explicitly taking the other side with Mag-7-best-asset. Chamath and Gavin agree legacy autos are melting icebergs headed for mega-mergers; Friedberg fades the defense primes for the Palantir/Anduril world, goes long beaten-down Chinese tech (Gavin: 'really cheap' but keeps his no-China rule), and calls the US nuclear buildout. Chamath's contrarian is a low-probability mainline-bank crisis expressed as long CDS ('92 of 100 times this goes to zero' — captured at strength 1 accordingly). Skipped as untradeable: Jason's OpenAI-total-collapse contrarian (private; no clean public leg), Gavin's frontier-labs-stop-releasing-models and 5%-GDP-year-within-four-years, Friedberg's socialism-rises-in-2025 (a notable thread — it reappears in his E286 debt-spiral thesis), Waymo mega-deal speculation, Unitree order (private), and the Polymarket supers on deportations and federal debt. Label quality: clean — all four speakers individually labeled, no merges detected; Gavin weighted as guest ×0.5 by the system.