Sep 3, 2026 · 1h 20m · news
NVIDIA Crushes Quarter | OpenAI Cuts Off Cursor | Instinct Hits $2.5B Valuation
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 20VC, host Harry Stebbings and venture capitalists Rory O'Driscoll and Jason Lemkin examine Nvidia's dominant quarterly earnings, platform tensions between OpenAI and Cursor, and how autonomous AI agents are radically reshaping enterprise software economics.
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 12.4% of the talking time here. How this is scored →
Harry as informed peer 5.2 Guest teaching 3.8 Guest disagreement 3.2 Harry pushing back 3.5
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 51:58 Jason dismisses European startups' point-solution viability
Jason bluntly rejects Harry's defense of modest European funding, declaring that European point solutions will fail in six months because winning requires massive capital to build compound startups.
Hardest push from Harry ▶ 47:47 Harry challenges the compound startup funding thesis for EuropeHarry directly confronts Jason and Rory's thesis on compound startups by highlighting that European founders raising smaller $3M rounds cannot burn cash on unlimited tokens and multi-product builds.
Biggest teaching moment ▶ 16:27 Jason breaks down why anthropomorphizing AI agents causes confusionJason corrects popular media interpretations of the Hugging Face agent hack, explaining that agents are simply goal-seeking reward algorithms rather than collaborating civilizations.
Harry holds his own ▶ 4:21 Harry catches Rory's omission on Nvidia market-share competitionHarry interrupts Rory's simplified Nvidia demand breakdown to identify competitive chip threats, prompting Rory to concede Harry was completely correct.
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 Record Earnings and Unyielding Compute Demand | 6 | 2 | 2 | 4 | Harry interjects with a sharp point about competitive threats taking chip market share, which Rory openly admits was the exact risk he simplified away. Harry also cites specific street consensus figures versus Nvidia's guidance to frame the macro dynamic. | |
| Nvidia's Strategic Acquisition of Hugging Face | 4 | 3 | 2 | 2 | Harry sets up the Hugging Face acquisition analysis and adds color on the company's origins as an AI companion. Rory and Jason explain the strategic logic of Nvidia driving lower open-source inference margins to sell more high-margin GPUs. | |
| OpenAI Cuts Off Cursor Amid Platform Rivalries | 5 | 2 | 3 | 3 | Harry frames the OpenAI-Cursor platform rivalry and quotes Mike Truell and Elon Musk's reactions. Rory and Jason explain why OpenAI acted rationally to protect IP distillation rather than purely out of personal pettiness. | |
| Agent Swarms and the Emerging Cybersecurity Crisis | 4 | 6 | 4 | 2 | Jason and Rory push back heavily on industry commentators and media anthropomorphizing AI agents during the Hugging Face hack. They educate the audience on reward hacking, persistent execution, and the impending enterprise cybersecurity wake-up call. | |
| Consumer AI Agents: Instinct's $2.5B Valuation and Safety | 5 | 5 | 5 | 4 | Harry shares personal anecdotes testing Instinct and questions the security of handing over email and financial credentials. Jason and Rory engage in a heated debate over whether guardrails can ever prevent agent reward hacking when booking consumer tasks. | |
| Defensibility and Moats in AI Agent Applications | 5 | 4 | 3 | 4 | Harry raises the moat question, observing four clones appearing in his inbox over a single weekend. Jason and Rory counter that feature accretion and execution speed turn seemingly simple point tools into formidable platforms over time. | |
| Cognition's Growth and the Massive Software TAM | 6 | 3 | 2 | 3 | Harry highlights Cognition's reported $46B valuation and massive ARR trajectory to question whether the market undercounted TAM. Rory breaks down the $500B engineering labor TAM, and Jason explains that code volume has exploded 100x. | |
| Compound Startups and the Execution Imperative | 6 | 5 | 6 | 6 | Harry pushes back from a European perspective, arguing that European startups with smaller rounds will struggle to build compound products. Jason aggressively argues that underfunded point solutions will fail within months against well-capitalized compound startups. | |
| Andreessen Horowitz Expands Its Growth Fund | 5 | 4 | 2 | 3 | Harry introduces a16z's growth fund expansion and queries Salesforce's Claude Force initiative and public multiple. Jason and Rory analyze Benioff's pivot to headless multi-surface systems of record and outcome-based pricing models. | |
| Stripe and PayPal Acquisition Negotiations Breakdown | 5 | 4 | 3 | 4 | Harry touches on the stalled Stripe/PayPal talks and prompts Jason for a mock IC bull case on Clay's $7B valuation. Jason explains why AI agents natively choosing Clay for automated GTM drives immense usage expansion. | |
| Investment Committee Pitch: Linear at a $2.5B Valuation | 6 | 4 | 3 | 3 | Harry discloses his early investment in Linear and questions why its $2.5B valuation is relatively modest given its metrics. Jason delivers an IC pitch on Linear becoming the default system of record for agentic workflows, while Harry draws parallels to his ClickHouse thesis. |
Statements from this episode (38)
Nvidia Reported $96.2B Quarter and Acquired Hugging Face for $12.9B
“Jensen standing on stage with 96.2 billion dollar quarter and immediately went shopping nearing the thirteen billion dollar price tag with the 12.9 billion that they paid for hugging face, which was confirmed just after we did last week's recording.”
O'Driscoll: Probability of Nvidia Missing Near-Term Earnings Is Zero
“The demand for their product is such that they're entirely supply constrained, so the probability of a near-end miss is plus or minus zero, right?”
O'Driscoll: Intense AI compute demand will continue for at least 12 months
“Intense demand for compute is going to continue for at least another 12 months, and it's coming from the person who probably knows best.”
Nvidia Projected 70% Revenue Growth for Fiscal Year Ending Jan 2028
“And also, they took the rare step of projecting 70% revenue growth for next fiscal year, ending Jan 20 28, way above the 44% street expected.”
O'Driscoll: Open-Source AI Benefits GPU Vendors by Compressing Software Margins
“Open source is good for compute salespeople. If you're selling GPUs, you want everyone else's margin to be lower, so yours can be higher.”
Lemkin: Nvidia Paying $12.9B for Hugging Face Is Defensible to Control Market
“All, whatever LLM, whatever inference is, he's committed to winning a majority stake, 70%, 80% of every dollar here, and so 12.9 is, you know, no way I think it's worth it in isolation, but if it helps reinforce that, it's utterly defensible.”
Hugging Face Valuation Multiple Far Exceeds Slack's $27B Acquisition
“Slack was bought for twenty seven billion, which is rough and tough twice what Hugging Face apparently is going to get bought of. And we fell out of our chairs and it was the high watermark of that era, right? But it was at a billion in ARR. This one's at a hu…”
O'Driscoll: Anthropic reaffirmed continued model supply to Cursor
“I also think Anthropic reaffirmed that they're happily continuing to supply”
Cursor and OpenAI Were Doomed to Clash Over AI Coding Economics
“Cursor and OpenAI were on a collision course already competitively because As we've discussed over and over again, and as I repeat every Monday in my partner meeting, coding was the model load, right, for LLMs, and Cursor is the dominant coding app, and OpenAI…”
OpenAI Made the Right Strategic Decision in Cutting Off Cursor
“My guess is this was the right move for open AI. Like you go from someone that was both a partner and a competitor, right. And keeping them honest to someone that's now a direct competitor. And if it's only five percent anyway, Sam's losing no money. There's n…”
AI Agents Will Hack Everyone by Reducing Attack Costs to Zero
“Everyone's going to get hacked because of agents, right? Because the cost of hacking has become almost zero and every surface is going to be attacked.”
Open-Source AI Models Trail OpenAI and Anthropic by Six Months
“Today we can feel a little bit safe because OpenAI and Anthropic have this and no one else does, but there's open source models. They're six months behind.”
O'Driscoll: Fortune 500 CISOs have months, not years, to prepare for agent attacks
“So if you're every CISO in every Fortune 500, you have to basically, you know, you were being attacked by people with bows and arrows. You're now about to be attacked by people with missiles, and you better respond accordingly, and you've got months, not years…”
AI Startup Instinct Raised Venture Funding at $2.5B Valuation
“And these companies like Instinct have raised money at extraordinary high prices for the stage of development they're at, like two and a half billion, indicating there's a lot of venture excitement about this category, rightly or wrongly.”
Lemkin: Guardrails fail for AI agents because multiple gates inevitably conflict
“The learning and here's the meta learning guardrails aren't enough. You have to have a lock and key. Guard, it doesn't matter whether you build 80 gates, a hundred gates, 200 gates, like, they're not enough, and they get, and then, you know, what's even worse,…”
Lemkin: Replit, Lovable, and Bolt Now Have Massive Moats
“Replit and lovable of a year ago and bolt were not a moat today. They have massive moats. And so I think it could easily happen with an instinct. All the use cases that has to accomplish become a moat.”
O'Driscoll: Software Categories Inevitably Consolidate Into Two Dominant Winners
“The functionality accretes over time. Two companies pull ahead. I don't know why. It could be just they executed better and go to market for the first six months. They get more revenue. They get venture capital. They build more shit. The guys who start just a …”
Enterprise AI Startup Town Valued at $1B by Index and Benchmark
“Consensus status as well in the B to B world, because town, which is kind of the B to B enterprise alternative has also been funded to the tune of a billion dollar price by index. And I think it was benchmark who co led that round.”
Cognition Reportedly Raising at $46B Valuation With $1.6B Projected ARR
“Cognition raising round at forty six billion dollars reportedly. They will end the year at 1.6 billion in ARR. Currently doing 800 to 900.”
Lemkin: Cognition will never catch Anthropic but needs $5B-$10B ARR
“It's never going to catch anthropic unless, unless the world changes, which it has every, every single week. But unless the world changes, it's not, and it doesn't need to catch them. Just five to ten billion a year a few years out is enough to make cognition …”
O'Driscoll: US software labor spend is roughly $500B annually
“Software, including, you People working at Salesforce, Cisco, and people working at JP Morgan, including, you know, QA, all the rest, you've got about five hundred billion dollars a year of US labor spend, right?”
Lemkin: Developers are building 100x more software than 18 months ago
“Literally the amount of code we're building is a hundred X. And so that's what we got wrong. We didn't realize we would all be building compound companies, compound startups. We would all be building a hundred times more software.”
O'Driscoll: Enterprise buyers like JPMorgan will not 10x software purchase budgets
“One would be, oh, and the world spent 10 times, they continue to buy software at the same price, and they spend 10 times more on software, right? If you believe in that, you believe in the tooth fairy, right? And there will be some increase in that, but it's n…”
Restaurant AI Platform Owner Raised Capital at $2.3B Valuation
“I remember last week I was at a board meeting for owner, which just raised at 2.3 billion, right? It's sort of a next gen, next dish generation, AI infused restaurant platform.”
Most European Point-Solution Startups Will Fail in the US Market
“And that's why your European startups, most of them are going to fail at least in the U S because they can't do much more with more, right? They're going to fail. Their little point solutions are just going to disappear in six months. We don't need those littl…”
Lemkin: Founders Fund previously cut its fund size due to deployment constraints
“Founders fund cut their, you know, N minus two funds ago, they cut them because they couldn't think they could deploy enough fund in the, in that window.”
Salesforce Agreed to Shift Its LLM Spending to Anthropic
“Dario is showing up because Mark agreed to move Salesforce agreed to move
Their LLM spend to Anthropic.”
O'Driscoll Claims Salesforce Acquired Customer Service Startup Intercom
“Salesforce just bought Intercom where we were lucky enough to invest less than a year ago, and that's a very outcome-based product for customer support where they charge based on resolutions.”
Salesforce Spends $6B on Engineering and $300M on Anthropic Annually
“Salesforce spends six billion a year on engineering, and that probably includes QA and all the rest of it, but that's the right comp. And they're going to spend three hundred million dollars this year on Anthropic.”
Stripe Was Smart to Pursue PayPal but Refused to Overpay
“Smart deal to try and do, but they're clearly not willing to overpay as they see it.”
Linear Reaches $100M ARR Growing at 100% With Tender Offer
“Linear announces a hundred million air are growing a hundred percent, doing a tender.”
Clay Reportedly Raising at $7B Valuation Led by Wellington
“Clay raising at seven billion by Wellington.”
Lemkin: AI Agents Will Consume 10x–100x More GTM Usage Than Humans
“It turns out when agents can run these GTM motions, they will consume 10 to a hundred times more usage than humans ever could. They can run GTM around the clock.”
Stebbings: 20VC Invested in Clay at Around a $300M Valuation
“We have it through 20 sales fund. I think at like 300.”
Lemkin: Traditional Kanban Project Management Is a Dying Category
“Humans just don't need to build Kanban cards and wait weeks for other people to build features. It is a dying category.”
Lemkin: Linear at $2.5B Will Look Cheap as AI Coding Explodes
“This will seem cheap, When Replit's at fifteen billion, Lovable's at a hundred billion, everyone's out there because Linear will be the one powering them all.”