E280
OpenAI vs Anthropic IPOs, Anthropic $3T, Zuck's Price War, China Ends Open Source?, Trump Accounts
2026-07-11 spoken.md · speaker-labeled ▶ watch ← E279 all episodes E281 →
0
ideas born
4
ideas moved
6
captures · 3 voices
2
dissenting
+41.7
conviction added
-34.4
decay · 54 silent
Every number here is replayed from score_events — the same ledger the pool ranks on. Decay is what the 54 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
▼ -10.8
🤖 AI compute capex supercycle has 12-24 months of runway
green threshold
→
watch
74.4 → 63.6
▲ +30.3
🤖 Frontier AI labs keep durable premium pricing (not commoditized)
dormant
→
ember
0.0 → 30.3
Kill dates that landed since E279
2 hit · 1 partial · 1 miss — windows that closed after 2026-07-03 and up to 2026-07-11, auto-scored against price data and never hand-set. verdict · R · α
| idea | verdict | R | α | closed |
|---|---|---|---|---|
| 🤖 Agents replace the browse-and-click layer and disintermediate consumer apps | PARTIAL | +21.3% | -1.1 | 2026-07-11 |
| 📈 Adobe takes the AI valuation retrade before Figma does | HIT | +41.7% | +20.4 | 2026-07-04 |
| 🏛️ State AI-regulation patchwork advantages incumbents over startups | MISS | -20.9% | -42.2 | 2026-07-04 |
| ⚡ The US power bottleneck is turbines, transmission and storage, not generation | HIT | +110.5% | +89.2 | 2026-07-04 |
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 (4 ideas)
ordered by how hard each idea moved · quotes are verbatim from the transcript, timestamps deep-link into the episode
▲ MSFT
🤖 Frontier AI labs keep durable premium pricing (not commoditized)
ember 27
CONTESTED
▲ +30.3
0.0 → 30.3
Open source went from 19% last year to 11% this year [of enterprise AI spend]... anyone who's saying that these closed models are going to lose or are somehow losing, you're just not seeing it in the data... this is a case of the spirit is willing, but the flesh is weak... it is just phenomenally convenient to go with the frontier labs... there's only two companies making meaningful revenue... I could see AI easily becoming another tech market that becomes a duopoly.
the share of wallet, is actually increasing to the frontier labs, while the share of tokens, ... The difference between spending 3 bucks on a cheap model or 15 bucks on an expensive model to replace a $200 an hour consultant, ... The non-consensus argument might be that intelligence is not converging at all, ... as it becomes recursive, you actually extend the lead, ... the distance between the frontier and everybody else doesn't converge, it actually extends,
the SpaceX IPO where we were also investors and we also bought in the IPO, ... trading at $2 trillion on roughly $35 billion of forward revenue. ... I still think there's a lot of meat on the bone on SpaceX, on Anthropic, on OpenAI. ... these things can be compounders. They're going to compound at the rate they compound revenue. ... well over 30% for the next many years.
▲ GOOGL
🤖 Anthropic IPO (~Oct 2026) is attractively priced at $1.5-2T
watch 54
CONTESTED
▲ +11.0
24.7 → 35.8
I suspect that if you can get out now, you should get out now before all of that starts to seep into the water table. ... Enterprise is probably a little bit more brittle because there are fewer buyers and they're more demanding. Consumer, on the other hand, ... when I use Fable 5, the problem is that it's nerfed on a bunch of things that I would normally research. ... it shouldn't really matter what model I'm using
rumored to be trending over 100 billion in revenue ... I think it would be a blockbuster IPO. ... Altimeter would be a buyer at scale and at size in both of those IPOs. ... If these guys end the year over $100 billion, ... they're on a revenue trajectory that they could 3 to 5x again next year. ... 200 billion of incremental revenue is incomprehensible in the history of Silicon Valley, ... intelligence is the largest TAM we've ever seen in the history of the world.
I sat down with my CTO today... right now, our token costs are doubling every 45 days... [and the downstream productivity?] maybe 5% max... you need to use a lot more tokens to get to this next iteration of improvement because we've effectively already asymptoted... everybody in the next three or four years will for sure go through [this reckoning]... [and per Fable's own numbers] the actual ROI [of AI on S&P-493 EPS] was somewhere between 1 and 2%.
Episode digest
written during extraction and stored in data/extractions/ep280.json — the auditable source of truth, including everything market-adjacent that did not earn a capture
Brad guests for Friedberg (vacation); Chamath dials in from the 8090 factory and exits early to 'sell enterprise software'. TRILLION-DOLLAR IPO QUEUE: SpaceX round-tripped to ~$150 (IPO price; ran to $200 first) = 'priced to perfection' at $2T on $35B forward revenue (Brad, who bought the IPO: 'textbook' — staged lockup, early index inclusion after exchanges modified rules, $75B raise); Anthropic confidentially filed June 1, Polymarket 65% IPO-this-year; OpenAI privately filed, GPT-6 rumored within 30 days, ~$70B exit-rate revenue, mojo back per Brad — his blockbuster-buyer-at-scale call captured (str-3, conflict: Altimeter invested in both); Chamath's counter captured (get-out-now-before-the-reckoning + enterprise-brittle/consumer-safe-harbor = OpenAI relative value again + Fable nerfing anecdote). THE ROI RECKONING (Chamath's captured capex oppose, 8090 receipts): token costs doubling every 45 days vs ~5% productivity lift — 'we've effectively already asymptoted'; his Fable-assisted S&P analysis strips NVIDIA-circularity out of 'AI EPS lift' → S&P-493 EPS +9% mostly pricing power + buybacks, true AI ROI 1-2%; 'they should report the EPS gains attributable to AI'; predicts the what-happens-on-an-earnings-miss test — companies cut token spend before people. Brad's rebuttal: experimental bucket, largest TAM ever, millions of independent buyers, Jensen designing chips with AI can't unplug, Snowflake-style optimization never dented revenue growth. CTO receipts on X: Uber's Praveen (99% engineers on AI, 70% of PRs agent-attributed, 2,500 agentic skills, agentic pods = FDEs embedded per department), DoorDash's Andy Fang (benchmarked routing: Fable for hardest work, Kimi 2.6 for lower-level, no quality degradation), Databricks' Ali (SAME model, different harness = 2x cost savings; GLM 5.2 strong), Dekagon founder (90% of usage now on post-trained open models — mature-use-case pattern; Sacks' maturity frame: immature use cases need the most powerful general model), Nikesh (enterprises want 'model fungibility' but nobody has abstracted memory/context/history away from the model yet). FRONTIER-PREMIUM DEBATE both directions captured: Brad's share-of-wallet-rising + premium-workload economics + non-consensus recursive-divergence thesis (intelligence gap may WIDEN as revenue→compute→better models compounds); Sacks' open-source-share-of-spend falling 19%→11% + technically-incapable-enterprise reality + duopoly-by-revenue (Anthropic ~$60B ARR vs OpenAI ~$40B, 'top two and then everybody else'); Jason's dark-tokens caveat (open-source usage doesn't show up as revenue — it shows up at NVIDIA/neoclouds) + his own 95%-cheaper Bittensor/TAO subnet setup (GLM 5.2 via OpenRouter; hourly agent crons, trend-spotting agent on all All-In transcripts; conflict: he holds TAO) + Lovable ($600M revenue, 30 months) and ElevenLabs both building their own models to leave the labs (Brad's counter: will Mati really use an inferior in-house model?). Zuck's price war: Meta MuSpark 1.1 'strong agentic coding model at a very low price' — same-quality-at-1/100th-cost vector (Chamath: 'they flubbed' the open-source scorch play, now attacking on cost) — era rule: meta-open-source-ai-leader born later, no retro-attach. CHINA MAY END OPEN SOURCE (Reuters, anonymously sourced): regulators met Alibaba/ByteDance/z.ai about restricting overseas access to top models; AI-research leaks becoming national-security offenses; Manus employees pulled back from Singapore; motive per report = fear Washington deploys Mythos against Chinese interests; Sacks: overstated — the pattern is open-until-you-catch-the-frontier then close (Qwen, GLM going closed; ByteDance always closed; same as OpenAI and Android before them); 'the best thing that could happen for America is if China sprouted their own Doomer community... we need a Chinese Yud... get their P-Doom up'; his real worry = mid-level bureaucracy + Congress responding to doomer pressure, not the president. Brad's DC readout: beating China is THE unifying force; GLM 5.2 'has watermarks from Mythos all over it' → US anti-distillation steps coming; frontier labs told him they'd ship open-source models if demand existed (Reflection + NVIDIA Nemotron carrying the US open flag). Sovereign AI: no country wants to 'subjugate themselves to technical risk' — UAE Falcon, Saudi Humane, Japan's $6B Neotera physical-AI consortium; Chamath's UN AI commission seat (Benioff co-chair, w/ Jensen, Brad Smith, Tom Brown for Anthropic — 'the future is open source for all these countries'). Chamath's energy-gap stat captured on electricity-prices-double (3 Californias short by 2050); Taiwan's 2-3 week LNG buffer as blockade risk (digest). TRUMP ACCOUNTS launch (Brad's baby, 4 years in the making — Invest America Act): $1K/child at birth into a zero-fee S&P 500 account; app went live July 4, #1 in App Store, 1.5M accounts + $1B deposits in 24hrs; Dell $6.25B (250/child for 25M lower-income kids), Gwen Shotwell $350M in SpaceX shares, Micron $250M employer match, Brad $100M (all Indiana kids under 5); $5K/yr contribution limit + $2.5K employer tax-free; 18yo rollover to IRA/Roth (CPA meta: convert in the 0% bracket college years → $200-300K tax-free compounding base, millionaire by ~28 if maxed); president ordered auto-creation of all 50-70M minor accounts within 90 days; ~25 states adding money; $100B philanthropic target year one; supplemental-IRA-for-adults expansion + TikTok-fee/Intel-shares as future account contributions floated; Sacks: 'as big as Social Security'; $2-4T flowing to S&P accounts over 15 years = structural equity-demand tailwind (noted; no near-term capture — policy/philanthropy). Labels: clean full names throughout.