Jul 17, 2026 · 50m · more-or-less
Open Source AI Wins. Now Labs Are Running to Washington |Thinking Machines, Johnny Ive, Realtime API · More or Less Podcast
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of the More or Less podcast, hosts Dave Morin, Brit Morin, Sam Lessin, and Jessica Lessin analyze the shifting economics of artificial intelligence, contrasting open-source model adoption against government regulatory capture and corporate valuation hype. They also discuss Silicon Valley's Hollywoodification, hardware partnerships like Jony Ive and OpenAI, and personal trends in bio-hacking and longevity.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 99.1% of the talking time here. How this is scored →
The hosts as informed peer 5.9 Guest teaching 3.9 Guest disagreement 3.9 The hosts pushing back 3.4
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 25:32 Sam derides Demis Hassabis' review committee
Sam aggressively dismisses the concept of government model review bodies as pure theater, arguing hyperbolically that only bombing global compute centers would represent a logically coherent containment strategy.
Hardest push from the hosts ▶ 3:14 Jessica defends Anthropic's financials against DaveJessica firmly rejects Dave's skepticism about AI foundation models, insisting Anthropic is a real company with legitimate GAAP metrics and revenue growth.
Biggest teaching moment ▶ 14:29 Sam reframes national security restrictions using crypto historySam dismantles Jessica's framing around restricting Chinese open source AI, educating the group on internet architecture and historical parallels to 1990s encryption export controls.
The host holds their own ▶ 36:58 Dave demonstrates technical depth on real-time voice modelsDave corrects Brit's assumptions about passive transcription hardware by articulating the architectural shift to full-duplex live streaming voice models.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Welcome to Deep Summer and Vacation Debates | 5 | 4 | 4 | 5 | Jessica and Dave clash over whether AI foundation model labs like Anthropic represent real, viable public market candidates. Jessica cites revenue growth and GAAP profitability, while Dave challenges the sustainability of their capital burn. | |
| OpenAI's 'Easy Button' Merch and Market Realities | 7 | 4 | 4 | 4 | Dave details his migration to cheap Chinese open-weights models like GLM for utility agent tasks, while Sam counters by arguing personal users will happily pay a premium to avoid errors. Both hosts demonstrate strong familiarity with LLM tiering and enterprise token budgeting. | |
| Regulatory Buzz Around Chinese Open Source AI | 7 | 6 | 6 | 4 | Jessica prompts a discussion on US national security concerns regarding open source Chinese models. Sam rejects the premise entirely, arguing that trying to restrict open source code contradicts basic internet architecture and comparing it to 1990s encryption export controls. | |
| Thinking Machines Open Weights and Tech Industry Wars | 7 | 3 | 3 | 3 | Dave introduces Mira Murati's new Thinking Machines model, citing specific technical parameters and context window specs. The group draws parallels between lab lobbying and the record industry's fight against Napster. | |
| The Policy Fight: Politics vs. Capitalism in AI | 7 | 5 | 6 | 4 | Sam forcefully condemns frontier labs for pivoting from market competition to regulatory capture and invoking apocalyptic risks. Dave questions whether legitimate biosecurity dangers exist, prompting Sam to analogize unilateral tech bans to unilateral climate policy. | |
| Demis Hassabis' 30-Day Model Review Proposal | 6 | 5 | 7 | 3 | Jessica summarizes Demis Hassabis' proposal for a 30-day pre-release review committee for frontier models. Sam mocks the proposal as ineffective political theater, satirically arguing that only military action against compute centers would logically achieve control. | |
| Jessica's Early Exit and Benson Boone Concert Banter | 1 | 1 | 1 | 1 | Jessica departs the podcast early to travel to a Benson Boone concert, prompting lighthearted teasing from Brit, Sam, and Dave. | |
| OpenAI Consumer Hardware and Jony Ive Branding | 7 | 4 | 3 | 3 | The panel discusses OpenAI's hardware rumors and partnership with Jony Ive. Sam articulates his thesis on the Hollywoodification of Silicon Valley, explaining how star talent and narrative multiples drive tech valuations over fundamentals. | |
| Voice AI, OpenAI's Realtime API, and Ambient Context | 8 | 5 | 3 | 4 | Brit inquires about passive audio recording wearables, but Dave redirects the conversation to full-duplex real-time voice streaming APIs. Dave outlines technical constraints around machine memory, physics, and multimodal context in robotics. | |
| Bio-Hacking, Peptides, and FDA Bottlenecks | 6 | 3 | 2 | 3 | Sam shares his initial experimentation with peptides, and Dave probes whether the sector is viable for venture investment. Sam breaks down compounding pharmacy business economics, pointing out that without insurance coverage, market size remains restricted. | |
| Discussion on Dating App References, Talent Recruiting, and Social Clout Scores | 6 | 4 | 5 | 3 | Brit brings up Hinge's new peer reference feature, leading Dave to cite his prior tenure on Hinge's board. The conversation moves to social scores, prompting Sam to passionately condemn Klout as a fundamental misunderstanding of social capital. | |
| Jay-Z's NYC Concert, Music Dynasties, and Millennial Nostalgia | 4 | 3 | 3 | 4 | Dave describes Jay-Z's stadium concerts as the millennial generation's classic rock nostalgia tour. Brit pushes back on whether Jay-Z and Beyonce are outdated, leading into closing remarks on biological age tests. |
Statements from this episode (30)
Brit Morin: Several portfolio companies are in active M&A processes
“I have a number of companies that are in M&A processes, so I do think there are real acquisitions that are gonna happen.”
Jessica Lessin: Anthropic will go public by September
“Anthropic will go public. I would be shocked if they're not public in September.”
Brit Morin: Anthropic and OpenAI valuations are unwarranted
“None of them are real companies that warrant the valuations they have.”
Sam Lessin: Founders are rushing to cash out before tech market collapses
“I think everyone's out securing their bags before it all falls apart.”
Sam Lessin: A real company is one you can successfully sue
“A real company is very simple. A real company is a company you can sue, and you get paid if you win the lawsuit.”
Dave Morin: Open-source AI models are less than two months behind
“And the open source models are less than two months behind.”
Dave Morin: Switched utility agent workloads to GLM-5.2 over frontier models
“I've switched out GLM five, two for almost every utilitarian use case that matters in any of my agents, because why would I pay for, you know, frontier intelligence To do all utility use cases.”
Sam Lessin: Massive AI pricing pressure coming; open source captures volume
“Obviously, there's gonna be massive pricing pressure on multiplying big numbers, and I totally agree with Dave. I've been saying this, I think, as long as anyone, I think we all have, which is like, you're just multiplying big numbers. Open source has caught u…”
Sam Lessin: AGI is a completely meaningless term
“AGI is a meaningless term at this point. It doesn't mean anything.”
Jessica Lessin: OpenAI and Anthropic fear Chinese AI models undercutting them
“Anthropic and open AI are worried about the Chinese models undercutting them.”
Sam Lessin: Tech is pursuing political regulatory capture over market capitalism
“This is what I don't like about our moment in history, is I don't think we're playing capitalism anymore. Like, I think we're playing some, like, political capture game that is way less interesting than capitalism, right?”
Sam Lessin: Cross-border embargoes on open-source AI cannot work on the internet
“It's like, you can't stop this from moving cross border. Right. And so the basic point with all this stuff, again, is like, it's one thing if you're just like, okay, like the Chinese models are bad. And like, we're going to put up a trade embargo on like copyi…”
Dave Morin: Thinking Machines released the first heavy-hitting American open-weights model
“So, you know, Mira Mirati's thinking machines appears to be a really solid Open weights model. You know, it's a almost nine hundred and seventy five billion parameter model million token context window mixture of experts. It's really good. And this thing's ope…”
Dave Morin: Thinking Machines' model beats Nvidia's Nemotron, rivals Chinese models
“Nemotron's fine. This is substantially better. This one's competitive with the Chinese models.”
Sam Lessin: Restricting global software movement is equivalent to banning books
“I just think at the end of the day, like, stopping people from moving software around the world, it's the same thing as stopping them removing, like, books around the world.”
Sam Lessin: AI labs seeking regulatory capture is cynical and anti-American
“I think it's very cynical for these companies to even try to switch the field of play from business to politics, right? Like, I think that's a very cynical, anti-American thing to do, to be like, okay, we're not going to play and compete in capitalism in a fai…”
Sam Lessin: Single-country AI regulation only protects incumbent market caps
“But, like, you taking any action as a single country does nothing, right? Like, and all it does is protect certain companies, right? And their market caps.”
Jessica Lessin: Demis Hassabis proposed 30-day government review for frontier AI
“What Demis, the AI leader, Google, Sabis, proposed was, like, It was this idea that there should be a separate government body that reviews frontier models basically for 30 days prior to their release.”
Jessica Lessin: OpenAI is expected to unveil a smart speaker this year
“We're talking about the fact that sometime this year they are expected to unveil A smart speaker.”
Sam Lessin: Jony Ive's brand makes his OpenAI acquisition effectively free
“The Johnny Ive acquisition thing, especially if it just turns out to be an Alexa, is like particularly interesting for this, because it was one step along that line of like, oh, I'm buying, I'm gonna pay a ton of money for Johnny Ive, because it will make rais…”
Lessin: Tech startups must align with dominant corporate cabals for value
“Studios, cabals rule everything, just like in Hollywood, right? You need to be able to align with Club Elon or Club OpenAI or Club Meta or Club Google to be valuable.”
Sam Lessin questions value of passive listening AI hardware
“In practice, like most of what people say isn't that valuable. So I don't know that like passive listening is actually all it's cracked up to be in the end of the day for usefulness. It's mostly a lot of ums, but look, marginally lowering the barrier to record…”
Dave Morin: Real-time streaming AI voice will supersede meeting transcripts
“I don't think that's actually how the future looks. I think the future looks much more real time and that you're just gonna be like streaming, you know, in and out of these things in real time. And they'll like tell you things and it's gonna look much more lik…”
Dave Morin: OpenAI's new voice model solves interruption and inference restart issues
“Everyone who's used the voice model knows that it's been this pretty janky experience where if you, like, say the wrong thing, it interrupts it restarts the inference, it gets confused. This is really good. It doesn't do that stuff anymore. You can just talk t…”
Sam Lessin: Peptides Won't Get Insurance Coverage Anytime Soon
“I don't think they can be covered by insurance anytime soon. So like that really cuts the market way, way, way down. Like if you don't have payers involved, like GLP ones, it's much harder to make huge businesses out of it.”
Sam Lessin: Klout was stupid and fundamentally misunderstood social capital
“No, but clout, clout was so fucking stupid. Like clout always annoyed me. It was like such a, like misunderstanding of how social capital works. Like so fundamentally, but people liked it because people liked scores.”
Sam Lessin: Biological age testing methods show a 10-year variance
“My biological age is like a variance of 10 years based on whether you Dexa or Garmin or whatever the hell, like they're all different. It's all bullshit.”