Did OpenAI Steal the Navier-Stokes Solution? | E2335

Watch on YouTube ↗  |  September 09, 2026 at 23:41  |  1:20:42  |  This Week in Startups
Speakers
Jason Calacanis — Angel Investor / Founder, LAUNCH
Rebecca Lynn — Co-Founder, Canvas Prime
Ben Lerer — Co-founder, Lerer Hippeau

Summary

This Week in Startups roundtable covers AI doomer headlines after an Anthropic researcher quit, Meta's Muse agent launch, and the OpenAI-Navier-Stokes data dispute. The guests debate trust in frontier AI labs, enterprise AI budgets, and big tech M&A. Market implications include positive views on mega-cap tech, Meta distribution, enterprise AI spend, and AI data-security infrastructure, while AI application-layer startups and autonomous-vehicle adoption face caution.

  • Anthropic researcher resignation and 10% AI extinction odds trigger doomer debate
  • OpenAI accused of using a researcher's Navier-Stokes work to advance its model
  • Meta launches Muse agent with Stripe payments and massive distribution potential
  • Jason argues frontier AI labs cannot be trusted with proprietary data
  • Rebecca sees enterprise AI budgets growing and AI tools as underpriced
  • Jason expects mega-cap tech M&A to accelerate under Trump
  • Data security and on-prem AI vendors framed as beneficiaries of frontier-lab distrust
  • Autonomous driving adoption noted as slower than early expectations
Ideas
Jason Calacanis Angel Investor / Founder, LAUNCH 0:42
Don't trust frontier labs; data security wins.
Frontier AI labs cannot be trusted with proprietary or important data. They have terms-of-service leeway, free credits are not free, and companies need encrypted, on-prem, open-source, or verticalized AI and data-security layers to protect IP from being trained or handed to competitors.
Rebecca Lynn Co-Founder, Canvas Prime 22:44
Autonomous driving adoption slower than expected.
Autonomous driving adoption has been slower than early investors expected because public safety expectations demand near-perfect performance before removing drivers. The technology is progressing, but rollout is later than anticipated.
Jason Calacanis Angel Investor / Founder, LAUNCH 34:16
Meta distribution will make Muse work.
Meta's enormous distribution through Instagram, WhatsApp, and Threads can force-feed Muse to mainstream consumers. Even though Meta often copies or buys successes, deep integration into Instagram with prompt nudges makes Muse likely to do very well with non-tech-savvy users.
Jason Calacanis Angel Investor / Founder, LAUNCH 40:03
Trump era favors big tech M&A.
With Lena Khan's M&A freeze over and the Trump administration favoring national champions, mega-cap tech companies are entering a stronger acquisition environment. At $2-4 trillion market caps, spending $10-30 billion on multiple AI acquisitions is rational portfolio-level R&D because a single winner can be worth hundreds of billions.
Rebecca Lynn Co-Founder, Canvas Prime 62:20
AI budgets keep rising; tools underpriced.
Frontier labs are under pressure to monetize massive data-center investments and cannot maximize revenue from tokens alone, so they will move into the AI application layer. Startups built on frontier models risk having their products preempted and absorbed by the labs.
Up Next

This This Week in Startups video, published September 09, 2026, features Jason Calacanis, Rebecca Lynn discussing AI data security / privacy infrastructure, Autonomous vehicles, META, XLK, AI-SECTOR. 5 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Jason Calacanis, Rebecca Lynn  · Tickers: AI data security / privacy infrastructure, Autonomous vehicles, META, XLK, AI-SECTOR