Gutenberg Lab — Morning Paper

Neuroscientist's 35-Year "Low-Volume, High-Intensity" Health Protocol — Morning Light, 3x/Week Lifting, and Ruthless Peak Protection

Featuring

Daisy (a16z) × Andrew Huberman, PhD (Neuroscientist, Stanford Professor, Huberman Lab Podcast Host)

Bottom line

Optimizing cognitive performance isn't about maximizing every health variable, but about identifying and protecting your personal peak cognitive windows while building daily rhythms sustainable for decades.

Three-Line Summary

Neuroscientist Andrew Huberman reveals the health protocol he's practiced for 35 years, tailored for tech founders: three pillars of morning light exposure, 3x/week low-volume resistance training, and sleep optimization, plus three founder-specific habits — protecting peak cognitive windows, daily disconnection practice, and organizing your day like an athlete. Challenges Silicon Valley's traditional "blitz and crash" culture with evidence-based approaches to sustained high performance.

Three Key Points

1. Three workouts per week, one hour each, two all-out sets suffice: Huberman's 35-year resistance training protocol uses only two work sets to failure per exercise (after warm-up and medium set), with 3-4 minute rest between sets. Total volume is "very low," yet a D1 athlete testified it delivers "the same results with a third of the volume I was doing before." Progressive overload (increasing weight or reps every workout) is key.

2. Productivity in peak cognitive windows is exponentially greater: Huberman's personal peaks are 9:30 AM - 12:30 PM and 4:30-7 PM. "If you're in one of those windows and you allow your attention to be captured by something that's not worthwhile towards your company build, you're not just losing that three or four hours, because the productivity that comes in those three or four hours is exponentially greater than the productivity that comes from doing that work at some point later." Only family/coworker emergencies should interrupt.

3. Phone accessibility drains cognitive resources even when not in use: Research shows that having your phone in the room (even off, in a bag) requires additional cognitive resources to maintain the same focus level as when the phone is completely inaccessible. Huberman uses a "supermax prison lockbox" for a separate device containing X (Twitter) and Instagram, allowing one-hour sessions before physically locking it away.

Editorial Perspective

Huberman's principle of "organizing your day like an athlete" strikes at the essential question facing operators during platform transitions. As practitioners who've navigated the shift from i-mode's official site economy to the App Store economy and now the AI transition, we've witnessed the contrast between "people who take a ton of stimulants and just blitz, blitz, blitz, blitz, blitz — these are the same people that crash and disappear" and those who "figure out the maximum amount of real work that you can do every day consistently" plus weekly reset. The advice from Huberman's father (physicist, Xerox PARC) — "You have to be a long distance runner. This is an endurance event" — is validated by Huberman's own 25-year adherence to a consistent schedule. The phrase "exponentially greater" productivity in peak cognitive windows isn't merely about efficiency optimization, but about the strategic question of where to bet your limited cognitive resources.

Source: a16z "Andrew Huberman's Three Rules for Better Health and Focus" (November 3, 2026)

https://a16z.simplecast.com/episodes/andrew-hubermans-three-rules-for-better-health-and-focus-3LRT9Uzz

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

AI Agents Deliver 50:1 ROI While World Model Developer Puts 90% Odds on Living in a Simulation

Guests

Jason Calacanis (host, angel investor, Launch Accelerator founder) × Oliver Cameron (Co-founder & CEO of Odyssey, former self-driving car developer for 8 years)

Bottom line

AI agents have moved from experimental to mission-critical, delivering 50:1 return on investment while automating 7+ full-time positions, and the CEO building world models to train robots believes there's a 90% chance we're living in an advanced simulation.

3-Line Summary

• Grok Bot users burned through weekly limits in days; Elon Musk called serving costs "unsustainably high relative to revenue" with 800,000 paid users doubling in a month
• Jason's company spends under $40K/year on AI tools, gets $2M+ in value, automating 7+ full-time positions and becoming 100% faster at average tasks
• Odyssey CEO Oliver Cameron is "90% certain this is a simulation," citing the double slit experiment (light behaves differently when observed) and universe's observational limits mirroring compute constraints in simulations

3 Key Points

1. The Last 90 Days as Inflection Point

Jason argues "the last ninety days are the equivalent of the first two years" of AI development. The shift came from agents gaining context and access through virtual computers inside Grok Bot and similar tools. Previously brittle systems now reliably complete multi-step tasks. Jason demonstrated designing jackets for Founder University Japan in 15 minutes, replacing 100-200 hours of designer/PM work.

2. World Models Learn Physics from Video

Odyssey's world models train on vast visual observations to learn physics, dynamics, and cause-and-effect. Oliver explains: "really attentive learner watches thousands of years of video, takes copious notes on every aspect of physics and motion." The company sells its base model (Odyssey 3) to robot labs, humanoid companies, self-driving car makers, and drone companies. Customers do post-training and deploy to physical devices. Already generating revenue.

3. 66% Oppose Data Centers Anywhere in US

ABC/Washington Post/Ipsos poll (October 2026) shows 66% of Americans oppose data centers anywhere in US, with 80% opposition among people 30 and under. OpenAI plans post-midterm PR campaign including renaming data centers as "AI compute campuses." Jason argues space data centers are inevitable: "Low Earth orbit has ~1,000x the surface area of Earth available, and you don't have anybody to complain."

Editorial Perspective

The unsustainable economics of Grok Bot — Elon Musk openly stating serving costs exceed revenue — is not a sign of immaturity but of demand overwhelming supply capacity. This mirrors the early internet's growth pains when bandwidth couldn't keep up with usage. Musk's transparency echoes Uber and Lyft's initial discount strategy: lock in users, then raise prices. The difference? 50:1 ROI means customers can absorb 10x price increases and still profit. Meanwhile, those losing jobs to automation capture none of these gains, making political backlash inevitable. Oliver Cameron's 90% confidence in simulation theory, coming from someone who builds world models daily, carries weight — if unobserved regions are rendered at lower fidelity to save compute (as the double slit experiment suggests), we're already living the optimization strategy Odyssey uses in its simulations.

Source

This Week in Startups "Odyssey CEO Oliver Cameron is 90% sure we live in a simulation | E2358" (November 3, 2026)

https://thisweekinstartups.com

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

• --

General-Purpose Robots Are Already Working in American Factories — Tutor Solves the Full Stack

Speakers

Raghuram (a16z General Partner) × Josh Gruenstein (CEO & Co-founder, Tutor Intelligence) × Martin Casado (a16z General Partner)

Bottom line

The success or failure of robotics hinges on solving AI, hardware, and customer acquisition simultaneously, not sequentially.

3-Line Summary

Tutor Intelligence is deploying general-purpose robots at scale in U.S. factories and warehouses, generating training data from paid customer work in a self-reinforcing flywheel. Nine out of 10 American factories have zero robots, the average worker is 55 years old, and turnover is three months—labor shortages are acute, and the window is open to leapfrog China's legacy automation infrastructure. Founded 5.5 years ago with 160 employees, Tutor has helped manufacture and ship over 250 million products.

3 Key Takeaways

1. The U.S. deployed only 38,000 industrial robot work cells in 2025—90% of the market is greenfield

Almost every customer Tutor sells to has never bought a robot before. The pitch is simple: "Have you considered a robot for $16/hour?" From first call to signed contract takes 40-60 days on average; from contract to robots on-site doing work, about 40 days; robots are in production within a day, often within hours. The market is wide open.

2. Ongoing costs (inference, teleoperation) now dominate robot economics, not upfront CapEx

Traditional robotics was big capital expenditure with 10-year ROI. Tutor's cost structure is "surprising and different"—more of the cost is ongoing (inference, teleoperation, maintenance) than upfront. This shifts the business model from ROI to operational expense, changing how customers budget and adopt.

3. China's legacy automation investment may backfire—the U.S. is building greenfield with general-purpose robots

China invested heavily in the previous regime of industrial automation (fast repetitive motions, custom machines). Josh argues this may be a strategic mistake: "In China, they're gonna have to tear down. We're not going to have to do that in the United States in the same way." The U.S. is deploying general-purpose robots from scratch, potentially equalizing the industrial base.

Editorial Perspective

The insight that robotics failures stem from treating AI, hardware, and go-to-market as separate layers—rather than a unified systems problem—may explain 60 years of unfulfilled promises. Tutor's structure, where paid customer work directly generates training data, mirrors the "miracle" of generative AI: training distribution equals economically viable distribution. The fact that 9 out of 10 U.S. factories have zero robots signals both market immaturity and an open window for whoever solves the full stack first to set the standard. The shift from CapEx to OpEx-dominant economics also changes the adoption curve—robots become operational expenses, not decade-long bets.

Source

a16z Podcast "The Robots Already Working in American Factories | Tutor Intelligence" (November 2, 2026)

https://a16z.simplecast.com/episodes/the-robots-already-working-in-american-factories-tutor-intelligence-331v57N3

AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.

• --