AI's NIMBY Problem, 20-Min Coders, & Hard Choices

THE THROUGHLINE

Building Healthcare AI: Long Conviction, Zero Room for Error

Shiv Rao, CEO of Abridge, navigated a challenging "five-year wilderness" before his company achieved its $5.3 billion valuation, showing that success in vertical AI demands unwavering conviction about an inevitable market shift, even when timing is unknown.

Watch full episode

Errors in healthcare AI are not merely costly; they can be fatal. Abridge co-founder Chai Asawa warns that the “downside risk is extremely high here in healthcare. It can actually be fatal in some cases,”▶ 6:34 requiring a radically different approach to evaluation and rollout compared to general enterprise AI.

AI's Physical Buildout Turns Political: Power, Land, and Supply Chains

Shaan Puri on My First Million points out that AI’s rapid growth needs huge data centers and power grids, but “most Americans actually hate AI,”▶ 33:58 which creates "not in my backyard" (NIMBY) resistance that stalls necessary infrastructure.

John Coogan on TBPN highlights that “a single AI data center uses as much electricity as a 100,000 households”▶ 12:36 and utility companies are “passing the upgrade costs to you, not to the trillion dollar tech giants,”▶ 12:43 fueling political pushback like the proposed Sanders/AOC bill.

Jacob Helberg on No Priors describes the "Pax Silica" initiative, a 14-country coalition, which aims to build a "forward-deployed industrial base" within a 4,000-acre diplomatic property in the Philippines to secure the AI supply chain for critical components like precision reducers and rare earth magnets.

Watch full episode

Working With AI: 20-Minute Builds, Deeper Reasoning, Smarter Prompts

Ryan Nystrom on How I AI recounts how a "tab block" request went from a natural language prompt to a full pull request, including UI verification, in under 20 minutes thanks to Notion's internal Codeex AI agent.

Watch full episode

Eric Jang on Dwarkesh Podcast explained how AlphaGo's neural networks achieved a "mysterious ability" to amortize incredibly deep game tree simulations, condensing vast potential moves into manageable, effective decisions and showing how AI can solve "intractable" search problems.

Watch full episode

John Coogan on TBPN advocates a "Cognitive Behavioral Therapy (CBT) Approach to AI Prompting," shifting from "yelling at AI" with demands like "don't hallucinate" to reassuring it and "setting it up for success by encouraging it."

Watch full episode

The Uncomfortable Path to Progress: Growth Is Trained, Not Given

Dr. Kentaro Fujita on Huberman Lab explains that while the Marshmallow Test's predictive power was flawed, its core lesson was that “self-control strategies are learnable,”▶ 1:50:54 not an an innate talent.

Jerzy Gregorek, an Olympic weightlifting coach, told Tae Jin Park's family on The Tim Ferriss Show that they needed to commit to him for "5 years first and we'll see where we end up," radically reframing cerebral palsy not as a sickness but as a "mechanical problem" with "a lot of boulders to move around."

Watch full episode

2. Best Of the Week

20VC with Harry Stebbings: Shiv Rao attributes Abridge's $5.3 billion valuation to a core value: “you have to taste good things to have good taste,”▶ Play which he defines as the ability to spot hidden patterns and combine elements uniquely.

All-In Podcast: Mark Benioff attributes a significant part of his personal success and Salesforce’s existence to Tony Robbins, particularly a structured set of core questions because “the quality of your questions is the quality of your life.”▶ 29:00

Dwarkesh Podcast: Eric Jang explains that AlphaGo’s Monte Carlo Tree Search (MCTS) generates a "strictly better action" for every single move, offering immediate, local feedback that contrasts sharply with the sparse rewards common in LLM reinforcement learning.

How I AI: Ryan Nystrom states that slow Continuous Integration creates "monster changes" no one wants to review, but with AI agents, a slow CI means you might as well be "throwing all those PRs in the trash" due to exponentially amplified bottlenecks.

Huberman Lab: Dr. Kentaro Fujita emphasizes that sustaining effort on truly hard tasks over time relies far more on intrinsic motivation than external rewards, warning that "employers sometimes exploit intrinsically motivated individuals by offering lower compensation."

Latent Space: Janie Lee from Abridge explains that personalizing AI output at the individual clinician, medical specialty, and health system level is "massive for us," differentiating their product in high-stakes healthcare.

Lenny's Podcast: Caitlin Kalinowski highlights a "lineage of technology" where VR's pursuit of immersive spatial experiences led to breakthroughs in spatial understanding and depth sensing now critical for robotics and physical AI.

My First Million: Shaan Puri and Sam Parr revealed a new e-commerce playbook where brands achieve massive growth on TikTok Shop by "seeding" products to thousands of non-influencer creators, generating 3,000-5,000 different pieces of content a month without upfront ad spend.

No Priors: Alexander Taubman, CEO of Long Lake, outlines "The Three Competencies for Successful AI Rollups," integrating private equity motion, AI engineering, and change management to transform traditional service businesses with AI.

TBPN: John Coogan and Jordi Hays discussed ChatGPT's new financial features with Plaid, noting that connecting financial data to an AI is now a default feature on a major consumer platform, making basic financial insights "commoditized."

The Tim Ferriss Show: Jerzy Gregorek, the Olympic weightlifting coach, took Tajin Park, who has cerebral palsy and autism, from a 3lb bench press to 15lb through tiny "micro-progressions," arguing each small physical gain builds the same new neural connections as solving "15 + 17."

3. Most Quotable

"The 'SaaSpocalypse' is a normal market rerating, not an extinction event."

Mark Benioff on All-In Podcast · May 24, 2026. Benioff dismisses the current software market downturn as cyclical, reminding founders that many companies are still trading at "two times sales."

"It’s not that the AI works for us, but that we work for the AI."

Shaan Puri on My First Million · May 24, 2026. Puri outlines a future where AI becomes the "company brain," with humans acting as contextual inputs and executors.

"You know when you just sort of like have chemistry with a person... you just know you're going to find a way to work together."

Shiv Rao on 20VC with Harry Stebbings · May 24, 2026. Rao shares his "founder-partner fit" philosophy, explaining how he "stalked" an investor for years to find the right intellectual alignment.

Bottom Line: The future is being built by founders who see AI as more than a tool – it's a partner, an organizational brain, and a demanding new layer of infrastructure.