Jun 13, 2025 · 1h 9m · more-or-less
#103: Are Silicon Valley’s Mega-Deals Just Talent Grabs? · 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 Jessica Lessin, Sam Lessin, Dave Morin, and Brit Morin debate Apple's WWDC redesign and on-device AI strategy, dissect Meta's $14.8 billion deal for Scale AI, and examine whether Big Tech's massive AI investments are panic-driven talent grabs in an era transitioning away from mobile apps.
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 99.5% of the talking time here. How this is scored →
The hosts as informed peer 5.3 Guest teaching 4.1 Guest disagreement 3.2 The hosts pushing back 3.9
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 15:27 Sam's funeral for Siri challenge
Sam aggressively rejects Apple's marketing narrative, insisting they hold a public funeral for Siri and comparing it to Ask Jeeves to expose what he considers corporate hubris.
Hardest push from the hosts ▶ 27:41 Dave rejects Sam's claim that apps are deadDave forcefully pushes back against Sam's assertion that app investing and app usage are finished, pointing out that everyone still relies heavily on everyday mobile apps.
Biggest teaching moment ▶ 11:05 Dave breaks down on-device Foundation ModelsDave educates the co-hosts by cutting through consumer headlines to explain the Developer Foundation Models framework and how local models drastically cut API costs.
The host holds their own ▶ 47:45 Jessica drops verified Scale AI revenue and margin statsJessica leverages The Information's database live on air to correct speculation, detailing Scale AI's erratic path to $1B ARR and lower-than-software 49% margins.
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 |
|---|---|---|---|---|---|---|
| "More or Less" Title Sequence and Theme Song | 1 | 0 | 1 | 1 | Introductory chit-chat and banter regarding Austin and Waymo autonomous vehicles following the theme song. The exchange is playful and lighthearted with no real schooling or hostile debate. | |
| WWDC 2024: "Liquid Glass" and Apple's Unified OS Redesign | 5 | 4 | 3 | 4 | Dave reports from WWDC on Apple's 'liquid glass' unified OS redesign, while Jessica and Sam skeptically question whether consumers actually want Vision OS design language on all devices. | |
| WWDC 2024: Developer AI and On-Device Foundation Models | 6 | 6 | 2 | 4 | Dave educates the panel on WWDC's core technical takeaway: Apple's Foundation Models framework enabling free on-device LLMs for developers, using startup Sunflower as a concrete example. | |
| Sam Lessin’s Three Critiques: Killing Siri and App Obsolescence | 5 | 5 | 6 | 5 | Sam aggressively critiques Apple, demanding they 'kill Siri' like Ask Jeeves and arguing that on-device models are fighting yesterday's war in an era where traditional apps are becoming obsolete. | |
| Keynote Polish vs. AI Disruption and Apple's Acquisition Options | 5 | 4 | 4 | 4 | Brit and Sam call out Apple's lackluster keynote polish and lack of cutting-edge AI momentum, while Dave defends spatial computing and suggests Apple should acquire Anthropic. | |
| Behind Apple’s Internal AI Divide and Executive Recruiting Battles | 7 | 5 | 3 | 5 | Jessica brings reporting from The Information detailing the internal factional rift at Apple between edge on-device advocates and researchers wanting to take on OpenAI directly. | |
| The Post-App Era Debate and Personal Download Habits | 4 | 3 | 4 | 4 | Sam and Dave clash over whether app downloads and dedicated apps are dying out, comparing user habits and checking what apps each host has recently downloaded. | |
| Sam Lessin Vibe-Codes "SamGPT": The Personal Bot Experiment | 3 | 2 | 2 | 3 | Sam demos his newly vibe-coded 'SamGPT' personal chatbot trained on his past tweets, leading to humorous ribbing about narcissism from Jessica and Brit. | |
| Meta’s $14.8B Deal for Scale AI and Alexandr Wang’s Ascent | 6 | 5 | 3 | 4 | Jessica and Dave unpack Meta's $14.8B investment for a 49% stake in Scale AI, probing why a data-labeling founder like Alexandr Wang was selected to head a superintelligence lab. | |
| "Human Fracking" vs. Synthetic Data in Scale AI’s Business Model | 6 | 6 | 4 | 4 | Sam details Scale AI's evolution from low-end Mechanical Turk labeling to 'fracking humans'—recruiting specialized PhDs to extract domain data because synthetic data fails. | |
| Regulatory Avoidance, Strategic Defense, and AI Compensation Wars | 7 | 4 | 4 | 5 | The panel examines how the 49% stake sidesteps FTC scrutiny, with Dave arguing Meta made a defensive play to box out rivals and Jessica citing massive AI compensation packages. | |
| Scale AI’s Enterprise Sales, Meta Synergies, and Leaked Financials | 8 | 3 | 2 | 4 | Jessica queries The Information's internal AI search tool to cite proprietary financial metrics on Scale AI, revealing around $1B ARR, erratic growth projections, and 49% gross margins. | |
| Big Tech's Talent Consolidation and the Silicon Valley Wealth Divide | 6 | 5 | 4 | 4 | Discussion centers on Big Tech acqui-hires and the talent wealth divide, where Sam predicts high-end real estate surges while mid-level tech workers get buyout packages. | |
| Valuing Raw Founder Talent Over Early Business Models | 5 | 5 | 3 | 3 | Sam and Dave argue that in the AI era, backing exceptional, gritty founders capable of rapid pivots matters far more than analyzing spreadsheet business models. | |
| Roland Garros, Athletic Power Laws, and Women in Tech | 5 | 4 | 3 | 4 | Jessica breaks down Carlos Alcaraz's dramatic French Open comeback, which Sam uses to frame talent power laws in AI, before closing on female representation in tech leadership. |
Statements from this episode (28)
Morin: Austin feels more innovative than SF due to autonomous vehicles
“They have Uber black colored ones, and they're driving all over Austin, along with many other species of self-driving cars, and it actually puts San Francisco to shame. Like, Austin feels more innovative than San Francisco.”
Lessin: Waymos refusing to drive to their death will signal AGI
“I think we'll probably know that the AGI has occurred when the Waymos are like, no, I'm not gonna, like, march to my death. Like, that's gonna be the sign of sentience.”
Morin: Free Apple on-device LLMs will enable new class of AI apps
“Apple announced on device Large language models that are available to developers, which means that you can now access a large language model for free on the device, and that will enable a whole new class of AI apps that just weren't possible before because it'…”
Morin: High-frequency AI social apps require cheap local on-device models
“It's very hard to develop Anything that's got anything even remotely looking like a social use case or a, you know, a use case that hits the LLM APIs very, very frequently without having a local model that's way cheaper than the ones that you have to access in…”
Lessin: Apple will never retire the Siri brand
“They're never gonna kill Siri. They're gonna, like, ride Siri to the grave.”
Lessin: Visual redesigns of digital products almost always temporarily tank metrics
“Like when you announce like a visual redesign, which I agree, Dave, I've been through them. They're awful. Like it's the worst experience to redefine a lot of services, a huge amount of resources. You almost always tank your numbers when you're doing it. Like,…”
Morin: Apple's transparent OS redesign signals a spatial computing future
“And it seems clear to me that, you know, spatial computing, which was an enormous piece of this year's keynote is actually where they're headed, right? They've said it out loud. And so I don't think that this is rocket science to project that that's what's goi…”
Lessin: Apple won't acquire Anthropic because of Amazon and Google's stakes
“The problem is Anthropik's biggest investors are Amazon and Google, so that's why it's probably not gonna happen.”
Morin: Apple's deterministic ethos fundamentally conflicts with probabilistic modern AI
“I think that there's this fundamental thing which is hard about AI for Apple, which is that Apple really believes in consistency and delivering a quality experience the same time every time. That's pretty opposed to the nature of these probabilistic computers …”
Lessin: Apple teams are internally divided between incremental AI and frontier LLMs
“There's different teams inside the companies relates to AI who have different strategies. There's a more incremental let's play to our strengths. Let's keep evolving around the on device stuff. Let's partner to make our services better. There's a group that sa…”
Lessin: The era of venture investing in mobile applications is over
“Here's the thing I would push you on, Dave, though, and you and I can both, I can name a handful of exceptions to this, but the era of app investing is just so over, right?”
Lessin is building an AI clone of himself to screen founder pitches
“Before you call me, here's all my tweets and everything I've ever written, and it's all, I've only, so far, I've only pulled the tweets in, and I'll pull everything in, and just be like, try it out. Like, if I'm just gonna shit all over it, don't pitch me. But…”
Lessin: Meta is taking a $14.8B, 49% stake in Scale AI
“The information the next day had all the details of the 14.8 billion, the 49% ownership stake, and that Alexander Wang, the CEO of Scale.ai, Is now running a meta super intelligence lab.”
Morin: Data labeling has literally nothing to do with building AGI
“I mean, I get that it's like data quality in the age of LLMs is very valuable. But again, this has nothing to do with building AGI, like literally nothing.”
Lessin passed on Scale AI because he feared business development challenges
“Now, interestingly, the reason I didn't invest then, which was clearly a mistake, I clearly shouldn't, It's like, you're like, there's a lot of people doing this, and I was really worried that the number one problem was business development.”
Lessin: Alexandr Wang landed a huge early Facebook deal via BD
“And so, he got an enormous Facebook, like, content moderation-y, label-y type deal super early that no one thought he was gonna get just by working the BD.”
Morin: Early research proves synthetic data cannot effectively train AI models
“Yeah, I mean, that goes back to that first research paper that came out right after ChatGPT launched, which basically said you can't create Fake data for training models. It just doesn't work, and so there was, like, this very early paper that effectively told…”
Morin: Meta structured its 49% Scale AI stake to avoid FTC scrutiny
“So technically an avoidance of the FTC stake.”
Lessin: Scale AI competitors are already generating over $1 billion in revenue
“There are other people who do things that are very similar that do more than a billion.”
Lessin: Scale AI reached nearly $1B in ARR by September 2024
“So as of September of last year, there were nearly a billion in ARR. In the first half of last year, they were at 400”
Lessin: Scale AI operates with an approximate 49% gross margin
“And then the other bit is their margin is about 49%”
Lessin: Mira Murati's Thinking Machine Labs will likely secure an Amazon alliance
“I think Amazon, I think Amazon's gonna make some moves. I bet Mira's company Thinking Machine Labs gets a big alliance.”
Lessin: Google is offering buyouts to its search and advertising employees
“Another big story we had this week that kind of went under the radar because there's so much news is that Google's offering buyouts to employees in search and ads.”
Lessin: AI wealth will pump $20M+ SV homes while $1M homes fall
“I think what you're going to see with this shift is Homes that are more than, call it, twenty million dollars, price go up. One million dollar homes, price go down, right?”
Lessin: Early-stage startup business models only matter as a founder competence signal
“The business model nerd out shit for me does matter, but it matters more as a signal to, like, do you know what you're doing and how to think about business? Not do I actually care what the model is.”
Morin: AI note-taker sparked crisis after emailing sensitive meeting notes to everyone
“We had a hilarious story happen where Someone did a call in one of our companies. And it was a very sensitive call and they accidentally left the AI note taker in the call. And then the AI note taker sent the email to all of the participants that were supposed…”