Jul 16, 2026 · 1h 26m · news
Zuckerberg Back on X Challenging Codex & Claude Code | SK Hynix’s $26BN IPO · 20VC with Harry Stebbings
Clips from this episode (6)
Short vertical cuts produced from the tape, captions burned in. Where a cut lands on a statement from the ledger, its card says so.
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 20VC podcast, host Harry Stebbings and venture capitalists Rory O’Driscoll and Jason Lemkin analyze the shifting landscape of venture capital, highlighting the financial pressures on legacy SaaS, the strategic rise of AI-driven tools, and the evolving ethical and structural trends of modern startups and investment funds.
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 11.8% of the talking time here. How this is scored →
Harry as informed peer 3.4 Guest teaching 6.3 Guest disagreement 3.2 Harry pushing back 2.5
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 45:57 Rory forcefully rejects Harry's cynical framing
Rory directly cuts off Harry mid-sentence with 'Stop. I'm going to push right back in your face', arguing Harry mischaracterized his structural point as a cynical old-guy cyclical comment.
Hardest push from Harry ▶ 45:44 Harry challenges Rory's framing on late-stage VCHarry explicitly pushes back on Rory's analysis by invoking perspective from cynical older VCs who claim to have seen late-stage private market enthusiasm in every prior cycle.
Biggest teaching moment ▶ 27:32 Rory educates on developer counts using BLS dataRory schools the panel on the physical limits of AI coding spend by citing Bureau of Labor Statistics figures showing only 1.8 million US developers exist, demonstrating current AI revenue is already hitting massive TAM thresholds.
Harry holds his own ▶ 1:12:45 Harry demonstrates valuation expertise across public/private SaaSHarry presents hard valuation metrics on Touch Bistro's $70M acquisition at 1x ARR alongside Wix's $2.1B market cap on $2.1B revenue to challenge the panel on the future pricing of pre-AI legacy SaaS.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Episode Agenda & Core Highlights | 3 | 7 | 2 | 2 | Harry sets up the Apple lawsuit agenda item, but Rory quickly steps in with granular details from his reading of the 41-page lawsuit, distinguishing between Tan Tang and Cheng Lu, before analyzing California's legal framework around non-competes and inevitable disclosure. | |
| The Impact of Litigation on OpenAI's Hardware Strategy | 3 | 5 | 2 | 2 | Harry asks whether the Apple lawsuit will delay OpenAI's hardware product timeline. Both Jason and Rory explain why hardware was a high-cash-burn distraction for OpenAI compared to core LLM coding value. | |
| Zuckerberg's Return to X and Meta's API Pricing Shift | 4 | 6 | 3 | 3 | Harry highlights Zuck's return to X and Meta's pricing shift for Spark 1.1. Rory banters with Harry about social platform engagement before Jason and Rory analyze token pricing tiers in enterprise CIO budgets. | |
| The Databricks Paper: Redefining AI Performance Metrics | 3 | 7 | 2 | 2 | Harry prompts a breakdown of the Databricks paper. Rory explains why cost per completed task and Pareto curves replace raw cost per token, while Jason details how developers doom-code and consume massive token volumes. | |
| The "Piranha" Threat of AI Design to SaaS Giants | 2 | 7 | 3 | 2 | Harry asks if AI design tools alter Figma usage. Jason articulates how AI tools function like piranhas eating away the bottom of the funnel for incumbent SaaS tools like Figma and Salesforce. | |
| Analyzing the TAM and Ceilings of AI Coding Spend | 2 | 8 | 3 | 2 | Harry asks Rory what aspect of the AI spend debate was overlooked. Rory drops Bureau of Labor Statistics numbers on software engineer counts and salaries to demonstrate that AI spend is rapidly approaching physical TAM ceilings. | |
| Anthropic's Shift to Biosciences & Personal Banter | 3 | 5 | 3 | 2 | Harry questions if Anthropic's move into biosciences signals TAM limits. Rory attributes it to founder noble search for meaning, leading into light banter about Anthropic's culture filter rejecting Jason and Rory. | |
| SK Hynix’s $26BN Listing and the Memory CapEx Rush | 4 | 7 | 2 | 2 | Harry introduces SK Hynix's $26B NASDAQ listing. Rory explains memory oligopolies and Korean stock market dynamics, while Jason connects IBM's earnings miss to CFOs scrambling for memory budgets. | |
| Jason Calacanis' Late-Stage Syndicate Shift and the YC Machine | 5 | 6 | 6 | 5 | Harry asks if Calacanis' syndicate shift to late stage reflects the market, adding that older VCs dismiss this as cyclical. Rory forcefully interrupts Harry to push back, distinguishing between structural growth in late-stage private rounds and traditional venture cycles. | |
| Discussion on Phoebe Gates' Startup Fia and Affiliate Marketing Scandal | 3 | 6 | 3 | 2 | Harry raises the controversy surrounding Phoebe Gates' startup Fia regarding affiliate link attribution. Jason downplays the outrage, pointing out that affiliate marketing dark patterns are ubiquitous across web tech. | |
| Ethics of "Cutting Corners" and Buying Sketchy Data in Startups | 2 | 7 | 5 | 3 | Jason asks whether buying sketchy data or cutting corners is acceptable for early startups. Rory initially expresses disapproval but concedes after reflecting on regulatory boundary-pushing by Uber, Airbnb, and frontier LLM labs. | |
| Banter on Quotes and Transition | 2 | 4 | 4 | 3 | Harry and Rory engage in brief banter regarding Rory's penchant for historical quotes versus Harry's short-form reading habits before transitioning to Carta seed valuation data. | |
| Mega-Funds, Ownership Targets, and Combining Rounds | 3 | 7 | 3 | 2 | Harry notes Carta data showing seed valuations reaching $200M. Jason explains that mega-funds deliberately combine multiple rounds into a single massive upfront check to hit 20% ownership targets. | |
| Chai Discovery's Hot Round and Biotech vs. LLM Trajectories | 4 | 6 | 3 | 2 | Harry highlights Chai Discovery's $400M mega-round. Rory differentiates biotech capital trajectories from software LLMs, while Jason advises founders against publicly announcing hot rounds to avoid attracting competitor daggers. | |
| The Three Categories of Venture Capital Investors | 3 | 7 | 3 | 2 | Rory categorizes venture investors into passive angels, board-level early stage, and late-stage growth replacements for public markets, while Jason notes how late-stage investors now receive zero information disclosure. | |
| Valuation Discrepancies and Terminal Decay of Legacy SaaS (Constellation vs. Bending Spoon | 6 | 6 | 3 | 3 | Harry introduces Touch Bistro's $70M sale and Wix's 1x revenue valuation to query legacy SaaS valuations. Rory and Jason explain how venture debt traps force fire sales and discuss the accelerating terminal decay of non-AI SaaS. | |
| Greylock's New $1.5 Billion Fund and Venture Capital Fund Discipline | 5 | 6 | 4 | 3 | Harry praises Greylock's $1.5B fundraise as a model of fund size discipline. Rory and Jason debate whether capping fund sizes reflects LP discipline or long-term GP alignment on franchise preservation. |
Statements from this episode (49)
O'Driscoll: Real social media engagement remains on X rather than Threads
“As three inveterate Twitter junkies here, we know what the action is, and it ain't on threads, okay?”
O'Driscoll: OpenAI's consumer success may have distracted the company
“To some extent, it may have been a distraction for OpenAI that they were so successful in consumer.”
O'Driscoll: Late-stage VCs are structurally forced to invest in competitors
“In the early stage, when you're on the board, you can't invest in two competitors. In the late stage, structurally, you have to.”
O'Driscoll: Employers will quickly abandon workers who steal trade secrets
“The first thing he's going to discover is there's no gratitude in litigation, right? You think you're caught helping your new employee or trying to get a job, they're going to burn you so fast your head's going to turn, right?”
O'Driscoll: 400 employees have left Apple to join OpenAI
“400 people from Apple have gone to move to OpenAI”
O'Driscoll: OpenAI has been actively discussing building a hardware device
“OpenAI has been talking about building a hardware device.”
O'Driscoll: Anthropic could not have been founded in Massachusetts due to non-competes
“I wonder if Anthropa could have been started in Massachusetts. Probably not, right? No, you're exactly right.”
Lemkin: OpenAI will likely cancel or pause its hardware project
“Listen, if I had to breathe the tea leaves, hardware is probably on the bubble. This was one of Sam's initiatives that made sense when everything was working and OpenAI had an unassailable lead over, over its competitors, right? Now everything's, you know, we …”
O'Driscoll: Economic value from AI coding dwarfs all other LLM applications
“The main thing here is that LLMs are really amazing at code, and the amount of economic value that can be created from that dwells to everything else, right?”
Lemkin: Enterprise budget caps will force adoption of cheaper AI models
“As every organization goes from burn all the tokens you want, baby, to budgets, However you do it, whether it's automated, whether it's manual, whether it's picking a dropdown, everyone's going to have to have a cheaper model.”
O'Driscoll: Active CIOs will implement model tiering to control AI spend
“When the pressure on spend is out there, the sentence you said that really resonated with me is every company is going to have an internal tiering. They might all buy it from OpenAI or Anthropic, but every company with a CIO who's half awake It's going to have…”
O'Driscoll: Cost per completed task matters more than cost per token
“I'm trying to, the big picture at the start was the basic point is, you know, cost per token is really not a useful metric. It's kind of cost per completed task was the first thing. How much does it cost to get things done?”
Lemkin: Top developers can consume 10x more AI tokens than today
“You can consume any top tier developer can assume an order of magnitude more tokens than they are now. The ideas are limitless.”
Lemkin: Figma is now tracking million-dollar enterprise software deals
“Figma's tracking million dollar deals, right?”
Lemkin: Next-generation startups may never adopt Salesforce due to AI agents
“The next generation of folks may not graduate to Salesforce. Because they never started there. They started agentically. They may never graduate there.”
Lemkin: Public SaaS companies must grow net new logos by 15% annually
“For public companies, the only thing that really matters to me is, are they growing net new logos, 15% or more a year?”
O'Driscoll: BLS data shows 1.8 million US software developers
“There were only 1.8 million developers in the United States, of which only 200,000 of them work in software companies, another 600,000 work in tech companies, you know, HP, IBM, and a million of them work in literally JP Morgan, B of A, right?”
O'Driscoll: Hypergrowth software startups decelerate gradually, never stopping suddenly
“Things in motion stay in motion. Things growing at 10 X year on year might decline to eight X or six X or four X, but they don't come to a grinding stop.”
Lemkin: Software companies will spend 10% of revenue on AI tokens
“All the software spend is going to put 10% of their top line into tokens.”
O'Driscoll: AI labs pursue life sciences for purpose over market size
“So I think actually people's interest in doing life science is one third TAM expansion and two thirds, shit, I'm putting everyone at a job. The least I can do is keep them alive. I want to do good because I want to have meaning. It's the search for meaning and…”
O'Driscoll: Daniela Amodei effectively runs Anthropic
“Based on my understanding of the entropic org structure, you in fact don't work for the man, you work for the woman, right? Daniella runs everything.”
O'Driscoll: Samsung and SK Hynix comprise 60% of South Korea's stock market
“The two memory companies dominate the Korean stock market. They are, the two companies, Samsung and Hynix, are 60% of the stock market. And roughly a 120% of the GDP in terms of market cap.”
O'Driscoll: AI memory chip supercycle will outlast normal cycles before correcting
“I think it'll last a little longer than the typical cycle, but in the end it'll, I mean, at some point it does correct. You're not going to have the memory makers Earning 70%, not net margins, not gross margins, net margins forever.”
Lemkin: when a late $100M check can 10x, why did I bother getting in early
“And when everything can exit north of twenty billion, you know, it's I, even me, like I have an investment with a hundred million dollar position that took me years and years and years to get to right now. And then just watching a growth round where someone co…”
Stebbings: Secondary Buyers Exist Every Day for Top Startup Holdings
“In all of my top names, I can get out of them today with ease and every single day I have buyers for them.”
Lemkin: Paul Graham probably is the GOAT, but be careful if you disagree
“And listen, I mean, one of the most successful investors of all time, right? Wildly successful beyond what anybody realizes. Directionally correct earlier than everybody else. And his portfolio is unmatched. So he has earned the right for everyone to follow wh…”
Stebbings: Y Combinator charges unprecedented carried interest premiums up to 50%
“The carry premiums that they charge are incredible too. People don't know or discuss that enough. Unprecedented. 40, 50% in cases.”
Lemkin: Faking affiliate attribution is standard practice across startups
“Doing basically what everyone in your industry does, doing scammy affiliate marketing when it's basically part of your thing. I mean, startups cut corners, right? It's not, this isn't Delvesque, but everyone cuts a corner. When everyone in your industry is sli…”
Lemkin: GoDaddy relies on tricking customers into unwanted add-on purchases
“I don't think GoDaddy could exist if it hadn't perfected getting you to stick stuff when you buy a domain name you never knew.”
Lemkin: Almost every AI go-to-market startup buys illegitimate data sources
“Almost every GT AI agent GTM provider is buying data sources that they probably shouldn't be buying. Almost every startup that is doing GTM, they are buying, whether it's a cleanse Chinese data center, they are taking your, like, you can't be that good with da…”
O'Driscoll: Major Startup Breakthroughs Rely on Boundary Pushing Legalized Upon Success
“So many breakthroughs rely on a little bit of pushing the boundaries. I mean, Uber did with taxi regulation. Airbnb did with regulation. You're right. Polymarket did in terms of how they're regulated. So I'm actually going to get off my high horse and say you …”
Stebbings: The top 5% of seed rounds now hit $200M valuations
“The top five percent of seed rounds hit two hundred million dollar valuation.”
O'Driscoll: Small seed rounds are pointless for capital-heavy AI neolabs
“If it's going to take you three hundred million to get something done, there's no point raising 20 at 20 pre. You're just not going to get there.”
O'Driscoll: Top decile seed valuations surged 6x from new AI categories
“Ordinary pricing has gone up 10, 20%, but top decile pricing has gone up six X or something like that, and that's really just a function of there are a class of bet that, frankly, didn't exist five years ago.”
Lemkin: Multi-billion funds raise seed valuations to meet 20% ownership targets
“If you see an S tier team, and it's in the right space, and you want to hit your number, your 20% ownership, and your fund is in billions, the math just makes sense. It's a bet. It's just a bet.”
O'Driscoll: Modern capital-intensive tech startups get far better terms than historical biotech
“At least a good portion of this technology is now doing capital intensive deals again, and they're doing it on terms that are way more favorable to the entrepreneur than you had seen in the past, and they're doing that for one reason. They have the money, and …”
Stebbings: Chai Discovery raised $400M at $3.8B valuation
“Four hundred million at a 3.8 billion dollar valuation led by Index, Kleiner, Sequoia, Dimension.”
Stebbings: Isomorphic Labs round was 8x to 10x oversubscribed
“Isomorphic was incredibly oversubscribed to the tune of eight to 10 X. The majority of people did not get anything at all when they wanted huge amounts.”
Lemkin: Startups with talent momentum shouldn't announce hot funding rounds
“The most in my career today, there are the most reasons to not announce, if you can get away with it, to not announce a hot round. Right. If you've already got enough, if you've already got enough going on to attract talent, right, for recruiting especially if…”
Lemkin: Smaller late-stage investors receive no information from companies
“In my experience for these smaller late stage investors, they get nothing from the company. Nothing.”
O'Driscoll: Heavy Venture Debt Traps Slow-Growing Companies
“It's just the real dangers of adding a big slug of venture debt on top of a slow-growing business. You just end up trapped. You put your equity at risk, and you end up with a cap table which is very misaligned.”
Lemkin: More Stalled SaaS Unicorns Will Be Wiped Out By Debt
“There's going to be a lot more of these stories coming in the coming years.”
Lemkin: Raising venture debt instead of equity is a sucker bet
“Certainly don't do debt instead of an equity round. That's the sucker bet, right? It's just, you better be the hottest thing on planet earth or it's gonna kill you.”
Lemkin: OMERS Ventures recovered only $2M from its $100M TouchBistro investment
“I think Omer's made two million on its hundred million.”
Lemkin: Stalled $70M ARR SaaS companies are now worth only 1x revenue
“It's a clean look of what's eight percent, six percent growth with deteriorating market share at seventy million ARR, where you're not reigning cash worth, it's worth one X.”
Lemkin: Legacy pre-AI SaaS companies will die when their contracts renew
“I think when the renewals come, they all just die like Marketo. They're just all gonna die.”
Lemkin: Adobe's Marketo is in terminal decay and heading to zero revenue
“It is going to zero. It is, but it is one of the slowest decays you can get, right? Because moving off a marketing automation system in the enterprise is very slow, but it is without question going to zero. There are no net new logos and everyone hates it.”