Teahose.
SIGN IN
NEW HERE — WHAT TEAHOSE DOES
We read the entire AI & tech firehose — so you don't have to.
PODPodcastsAll-In, No Priors, Acquired…
NEWNewslettersStratechery, Newcomer…
PAPPapersPhysical AI research
PHProduct Huntdaily launches
VCInvestor ScoutSequoia, a16z, Benchmark…
CLAUDE DISTILLS →
7 reads, 30 sec each — free, 6 AM ET.
+ a live graph of the companies, people & themes underneath.
HOME/SOURCERY NEWSLETTER/ICYMI: 10 Wild Hot Takes From th…
NEWS
// NEWSLETTER ISSUE
SOURCERY NEWSLETTER

ICYMI: 10 Wild Hot Takes From the All-In Summit 2026

DATE September 27, 2026SOURCE SOURCERY NEWSLETTERPARTICIPANTS MOLLY O'SHEA
// SUMMARY

1. Key Themes

AI Safety Debate Is Becoming Institutionalized, Not Just Ideological

The push for independent oversight of AI is gaining traction among mainstream investors, not just doomers or skeptics.

"We had @elonmusk come here & say, 'I tend to agree with Dario. The risk is augmenting & we need to have independent peer review.' I thought that was a great idea." — Brad Gerstner

The Energy-AI Nexus Is the Real Infrastructure Story

Several speakers converged on the idea that AI's energy demands — not AI cognition itself — are the actual bottleneck and opportunity, with natural gas and compute providers positioned to benefit.

"We're seeing a lot of the techniques that were used to help shut down energy development in this country are now being used to shut down..." — Toby Rice "Demands still far exceed supply. That's great for us as a provider because if we can get compute online, there will be a revenue stream." — Kent Draper

Hardware Efficiency, Not Just Model Scale, Is the Next AI Frontier

There's a counter-narrative to "bigger models" — that radical energy efficiency gains in AI hardware could reshape the value stack away from software.

"nanojoules per image, building AI hardware thousands of times more energy efficient and why he believes software could become a de minimis part of the value stack." (Naveen Rao, description)

The Body-as-Platform / Biohacking Wave Is Entering Mainstream Tech Discourse

Longevity, wearables, and human optimization are increasingly framed as an extension of the tech stack, alongside AI.

"the human body is the next major technology platform, biohacking, wearables, peptides, GLP-1s, ketones and treating the body like an API." (Michael Brandt, description)

2. Contrarian Perspectives

AI doomerism is financially motivated, not evidence-based. Naveen Rao argues that fear around AI risk is largely a narrative tactic rather than grounded in data, and challenges doomers to put their money where their mouth is.

"I don't think we have any credible evidence other than vibes." "If everybody is so concerned about this, stop working on it. Don't go IPO. Don't make a lot of money for…" — Naveen Rao

The lettuce/food-safety scare was overblown ("a psyop"), while processed food is the real danger. Jonathan Neman flips the usual AI-doom narrative, arguing that ultra-processed food — not AI — is the bigger existential threat to health, and that a single contamination event unfairly tainted an entire food category.

"There was 1 farm from 1 supplier serving iceberg lettuce to Taco Bell that was unsafe, and unfortunately, they scared the whole country from eating not just lettuce, but…" — Jonathan Neman "AI's not gonna kill us, but processed food will."

AI development should accelerate, not slow down — safety comes from controls, not deceleration. Wen Sang pushes back on the "pause AI" camp, arguing speed and safety aren't mutually exclusive, backed by Genspark's own hypergrowth as proof of demand.

"It's about putting more controls in place, but it's not about slow down." — Wen Sang (Genspark hit "$250M ARR in 12 months, 7,000+ business clients in 8 months")

3. Companies Identified

  • Unconventional AI — AI hardware startup — Mentioned for its energy-efficiency thesis and massive funding round — "Unconventional AI raised $475M Seed at $4.5B [valuation]"
  • Sweetgreen — Fast-casual salad restaurant chain — CEO used the summit to defend fresh food against processed food narratives — "you can eat lettuce now PS: that was a psyop"
  • EQT — Energy company (~$30B, NYSE: EQT) — Cited for connecting AI/data center growth to natural gas and LNG demand — "fear around AI & data centers is obscuring the bigger energy story"
  • Genspark — AI agent platform — Cited as a fast-growth case study for "speed, not slow down" philosophy — "hit $250M ARR in 12 months, 7,000+ business clients in 8 months"
  • Anti Fund — Venture fund (Geoff Woo, Jake Paul) — Represented views on consciousness, free will, and constitutional reform tied to AI-era technology — "Spend all your time maximizing, spending & burning as many tokens to get ahead while money & status are still..."
  • IREN — AI infrastructure / neocloud compute provider — Highlighted for selling out capacity amid compute scarcity — "IREN has already 'sold out' their 2026 capacity"
  • Ketone-IQ — Biohacking/ketone supplement company — Used as a lens on the "body as technology platform" theme — "Some people want to breathe underwater. Other people want to neuromaxx & be really smart.. dunk a basketball."
  • Nervana (acquired by Intel, ~$400M+) and MosaicML (acquired by Databricks, ~$1.3B) — Prior companies founded by Naveen Rao — Mentioned to establish his credibility/track record — "Rao Founded: - Nervana, acq. by Intel for $400M+ - MosaicML, acq. by Databricks for $1.3B"

4. People Identified

  • Brad Gerstner — Founder & CEO, Altimeter Capital — Framed the AI safety debate and Elon Musk/Dario Amodei peer-review consensus — "I tend to agree with Dario. The risk is augmenting & we need to have independent peer review."
  • Naveen Rao — Founder & CEO, Unconventional AI — Serial AI hardware entrepreneur pushing efficiency-first, anti-doomer stance — "I don't think we have any credible evidence other than vibes."
  • Jonathan Neman — Co-Founder & CEO, Sweetgreen — Contrarian on food safety fears vs. AI fears — "AI's not gonna kill us, but processed food will."
  • Toby Rice — President & CEO, EQT — Connected AI infrastructure growth directly to energy policy and gas development — "techniques that were used to help shut down energy development... are now being used to shut down..."
  • Wen Sang — Co-Founder & COO, Genspark — Advocate for AI acceleration with guardrails — "Everybody could work like a Jensen Huang, right? He says, 'I love working.'"
  • Geoff Woo — Co-Founder & Managing Partner, Anti Fund — Philosophical take on free will, consciousness, and status in an AI-abundant world — "Spend all your time maximizing, spending & burning as many tokens to get ahead while money & status are still..."
  • Nick Shirley — Independent journalist — Investigated government fraud/waste in California's high-speed rail project — "Since 2008 California has spent $15 billion and has not connected a single city or laid a single track."
  • Jake Paul — Co-Founder & Managing Partner, Anti Fund — Argued for updating the Constitution for an age of advanced weapons/autonomous systems — "As weapons systems get more advanced, a pistol or a…"
  • Kent Draper — Chief Commercial Officer, IREN — Signaled compute scarcity as a durable revenue driver — "Demands still far exceed supply."
  • Michael Brandt — Co-Founder & CEO, Ketone-IQ — Framed the human body as the next tech platform via biohacking — "treating the body like an API" (description)

5. Operating Insights

  • Efficiency-per-unit metrics matter as much as raw capability. Naveen Rao's focus on "nanojoules per image" signals a shift toward energy-cost-per-inference as a key competitive metric for AI hardware builders — an operating lens founders in AI infra should adopt now.
  • Sell into scarcity, not hype. Kent Draper's comment that IREN is already "sold out" for 2026 capacity underscores a tactical lesson: in infrastructure-constrained markets, locking in forward demand/contracts ahead of supply is a durable moat.
  • Growth-at-speed can coexist with "controls." Wen Sang's framing — controls instead of deceleration — is a reusable playbook for AI product teams navigating regulatory/safety pressure without slowing shipping velocity.

6. Overlooked Insights

  • The "SaaSpocalypse" framing (mentioned only in the timestamp/section header, tied to Naveen Rao's segment) suggests a thesis that AI hardware efficiency gains could make software a "de minimis part of the value stack" — a potentially significant re-rating risk for SaaS-focused investors that wasn't elaborated on but was clearly flagged.
  • Government fraud as an emerging investor talking point. Nick Shirley's California high-speed rail exposé ($15B spent, zero track laid, $200B+ to complete) was framed as nonpartisan and drew direct commentary from summit host David Friedberg — signaling growing crossover interest among tech/investor audiences in public infrastructure accountability as its own theme.