Aug 27, 2026 · 1h 26m · news
Anthropic's $30T Assumption & OpenAI Confirms IPO | Why Customer Service & Robotics are Overinflated
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 20VC panel, venture investors Jason Lemkin, Rory O'Driscoll, and Harry Stebbings analyze the shifting economics of the AI landscape, debating frontier model competition, soaring startup valuations, and corporate compute budgeting. They evaluate the market viability of autonomous agents and open-source infrastructure while identifying overhyped venture categories.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 11.6% of the talking time here. How this is scored →
Harry as informed peer 4.2 Guest teaching 4.2 Guest disagreement 3.1 Harry pushing back 3.0
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 8:23 Jason rejects 15x venture return as sufficient for seed investing
Jason forcefully dismisses Rory's premise that a 15x return on an unsuccessful thesis represents a good seed outcome, insisting that multi-hundred-million-dollar entry prices make 15x inadequate.
Hardest push from Harry ▶ 21:12 Harry refutes Rory's low training data TAM estimateWhen Rory estimates the training data market cap at only five to ten billion, Harry bluntly interjects with 'Way more' and rattles off actual ARR figures of four competitor companies to disprove the estimate.
Biggest teaching moment ▶ 50:07 Jason and Rory explain Kospi index composition to HarryHarry admits not knowing what comprises the Kospi index, leading Jason and Rory to explain that it represents the Korean market driven heavily by memory semiconductor manufacturers.
Harry holds his own ▶ 32:37 Harry uses Vercel platform telemetry to counter Rory on open weightsHarry interrupts Rory's downplaying of open-weight competition by citing Vercel's internal data showing 68% open-weight model usage, forcing Rory to immediately retract and concede the point.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Nvidia's Poolside Deal and Venture Economics of Open Models | 5 | 5 | 6 | 3 | Harry introduces the Poolside acquisition and shares his own seed return data (15x). Jason and Rory clash over whether 15x is an acceptable seed return in modern venture economics or a mathematical shortfall. | |
| Mercor's $20B Valuation and Nvidia's Vendor Financing Play | 7 | 4 | 3 | 6 | Harry showcases deep sector knowledge as an investor in Mercor, directly challenging Rory's conservative market size calculation with precise ARR numbers from Surge, Handshake, and Micro1. | |
| OpenAI's 2027 IPO Target and Enterprise LLM Competition | 6 | 5 | 4 | 5 | Rory breaks down OpenAI's revenue growth trajectory and ranking shift against Anthropic. Harry intervenes with Vercel platform data showing 68% open-weight usage, prompting Rory to concede and realign his argument. | |
| Hugging Face Valuation Rumors and the Open-Weight Window | 3 | 3 | 2 | 2 | Jason and Rory discuss Hugging Face's rumored $13B valuation and strategic fit for enterprise IT buyers seeking counterbalances to closed frontier models. Harry largely facilitates. | |
| Citadel's Market Play, AI Stock Volatility, and SF Real Estate | 2 | 7 | 2 | 1 | The conversation covers Citadel unwinding Leopold Aschenbrenner's position, Korean market volatility, and SF rent inflation. Harry openly asks what makes up the Kospi index, prompting a brief educational explanation. | |
| Corporate Token Addiction, CFO Budgets, and Stripe's AI Surge | 5 | 5 | 5 | 4 | Jason presents his thesis on enterprise token addiction, while Rory counters using the Stripe letter's framing of intelligence allocation like capital. Harry offers counterpoints on media generation quality and legal AI adoption. | |
| AI Agents and Trust: Security, Guardrails, and Reliability | 4 | 4 | 3 | 4 | Harry challenges skepticism around AI agent security by drawing an analogy to the early days of online credit card processing. Jason counters with practical examples of unpredictable LLM behavior. | |
| Personal Productivity vs. Enterprise Automation for AI Agents | 2 | 5 | 2 | 1 | Rory educates on the limitations of personal productivity software versus high-ROI corporate workflow automation, citing historical precedents like Evernote and Ben Thompson's observations on mainstream behavior. | |
| Debating the Most Overhyped AI Investment Categories | 6 | 2 | 3 | 3 | Jason prompts the group to identify the most overhyped AI categories. Harry leads with customer service and humanoid robotics, providing concise structured arguments that both guests validate. | |
| The 'What If It Works' Mindset and Founder Creativity | 2 | 2 | 1 | 1 | The panel wraps up cooperatively, with Rory and Jason reflecting on the importance of founder creativity and maintaining a 'what if it works' mentality despite analytical skepticism. |
Statements from this episode (34)
Stebbings: Nvidia licenses Poolside factory for $6B, invests $1B at $12B
“NVIDIA is paying six billion dollars to license Poolside's model factory and investing a billion dollars more at a twelve billion dollar pre-money valuation, moving a 109 engineers over to NemoTron to help build it.”
O'Driscoll: Competing at the AI frontier is almost impossible for new entrants
“The lesson is, it's almost impossible to compete now at the frontier. And by the way, that has positive implications for OpenAI and Entropic, we should talk about in a second. So the negative lesson for everyone else is it is almost the next smartest people wh…”
Stebbings: 20VC to see roughly 15x return on seed stake in Poolside
“You know, it'll be like a 15 X for us as a seed ambassador. It's pretty great.”
Lemkin: A $9B exit is not enough for seed investing in 2026
“Nine billion doesn't clear the bar for seed investing in 2026.”
O'Driscoll: Anthropic and OpenAI exist only due to hyperscaler funding
“The only people capable of financing a state-of-the-art frontier model in the United States of America has been the hyperscalers themselves. The only reason Entropic and OpenAI exist is because Microsoft, Google, and Amazon gave them enough money to play becau…”
O'Driscoll: Only one or two trillion-dollar outcomes this decade, all AI
“I mean, there's only going to be one or two trillion dollar outcomes per decade, and they're likely going to be all concentrated for now in, you know, frontier models.”
Stebbings: Mercor Crosses $2B ARR and Raises at $20B Valuation
“I'm an investor in McCaw. I never thought this would be as big as it has got as quickly as hit, as it has done. It's crossing out two, two and a half billion ARR. They're doing a new round led by General Catalyst at twenty billion.”
Stebbings Predicts Largest AI Data Providers Will Reach $200B
“I think your largest data providers will be two hundred billion dollar companies.”
Stebbings: Surge AI at $3.5B ARR, Mercor $2.5B, Handshake $1B, Micro1 $500M
“Surge are doing three, three and a half. McCaw are doing two and a half. Handshaker doing one. Micro One are doing half.”
O'Driscoll: Anthropic Gross Margins Rose From -91% to +30% in One Year
“Anthropic went from -91 to positive 30 in a year.”
Stebbings: OpenAI CFO told employees the company will IPO in 2027
“CFO Sarah Fryer told employees at OpenAI, we will be a public company in twenty-twenty-seven.”
O'Driscoll: OpenAI will IPO later and lower than Anthropic
“The rank, the most, the big picture fact is the ranking has changed, and you're now number two. So you will go out later then and a lower price than the other guys, unless you change the trajectory massively.”
O'Driscoll: Open-weight will drive token volume while frontier models capture revenue
“Yes, I think that the vast predominance of tokens process will be open weight, and obviously, as people reiterating, the vast majority of revenue, the significant majority of revenue will still be frontier state-of-the-art because it can command more value tha…”
O'Driscoll: Low-cost competitors pressure market number twos much worse than number ones
“In a situation where a low-cost competitor With nearly equivalent functionalities attempting to enter your industry, you would far prefer to be number one than number two. Because number one can say, you got to just buy us because we're number one. But you're …”
O'Driscoll: Coding is the highest-ROI and fastest-adopting market in AI
“Coding is the fastest adopting market. It's the highest ROI market. It's the mother load, right?”
Lemkin: Selling subsidized consumer AI tokens is history's worst business model
“It's a crappy business that consumers selling 10,000 dollars worth of tokens for 200 dollars is one of the worst business models of our lifetimes, right? If that was the only business, these guys would be dead in the water. It's a pretty bad business.”
O'Driscoll: Consumer AI Could Reach Half Google's Size in a Decade
“I could totally see a business, you know, plus or minus, you know, 50% to the same size as the Google consumer business, maybe over the next decade.”
O'Driscoll: Anthropic's $30 Trillion TAM Target Is 'Delusional'
“I just say on Tropic and the Wall Street Journal, they believe their TAM is 30 trillion. And I, and then I say to myself, oh, I remember everyone when they're doing really well gets slightly delusional. Your time is, you know, the entire US GDP.”
O'Driscoll estimates Hugging Face generates roughly $150M in revenue
“Well, to a fifteen billion dollar outcome, I think it's roughly a hundred and fifty million.”
O'Driscoll: Open-weight AI enabling tech is at peak valuation moment
“If you are an enabling technology for open-weight models, now is peak moment.”
Lemkin: Any Hugging Face acquirer will leave it untouched for 24-36 months
“I'm almost confident if anyone actually spends three billion, let alone thirteen billion, they're going to 95% leave it alone for 24 to 36 months, right?”
Stebbings: Citadel unwound 80% of Leopold Aschenbrenner's forced-sale book
“Ken Griffin's Citadel Unwinds, 80% of Leopold Ashenbrenner's four sold book.”
Lemkin: Mediocre Dogpatch apartments near Y Combinator rent for $10,000 a month
“Rents in the mediocre apartments just in dogpatch are 10,000 dollars a month now. Mediocre apartments down the street from YC, the Avalon.”
O'Driscoll: California took 75% of venture funding, driven by OpenAI and Anthropic
“California didn't just outperform everywhere else. It got three quarters of the total dollars. Now, that's invention. That's skewed by the fact that Entropic and OpenAI together got probably 60% of the total dollars.”
Lemkin: Workers Will Threaten to Quit Without Continuous AI Agents
“Next year, there's going to be backlash. I can see it in my best portfolio companies where we can't go back anymore. We can't live. We can't go back in time. And I need my 10 sub agents running 24 hours a day to do my job or I quit.”
O'Driscoll: Companies Must Manage AI Like Capital, Not Software Licenses
“People are going to have to control intelligence in a way that's more like how you control money and less like how you think about software licensing.”
Lemkin: Databricks and OpenAI IPOs will be nail in coffin for SaaS
“I think it will be the nail in the coffin for almost every public software company. And what I mean is when Stripe and data, like, okay, there's OpenAI, there's Anthropic, we can put them in a different category, right? They clearly are on many levels. When Op…”
O'Driscoll: Stripe Reduced Its Share Count Over Past Four Years
“Their share count is down on three years ago, four years ago, which means they've been buying back stock.”
O'Driscoll: Silicon Valley overestimates consumer demand for personal productivity tools
“Silicon Valley forgets every three years that the average American is not trying to be efficient, right? No one wakes up in the morning and says, I need to grind down my to-do list in Harlem. They're just living life. Yeah, they're doing their job, and then th…”
Stebbings: Huge amount of money will burn in AI customer support
“I think a huge amount of money is going to get burning customer support.”
O'Driscoll: Defense tech venture returns will suffer from market consolidation
“I think an area where, despite it being amazing for America and important for the world, I think the venture returns of the March might be tough will be defense, not because we don't need all these products, but because I think there's an element of that busin…”