Feb 3, 2025 · 1h 15m · news
Nabeel Hyatt, GP @ Spark Capital: To Win in AI, Investors Need to Change Their Approach | E1255 · 20VC with Harry Stebbings
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 deep-dive interview, Spark Capital General Partner Nabeel Hyatt challenges the industrialized, spreadsheet-driven playbooks of modern venture capital, arguing that the unpredictable AI landscape demands a return to creative, high-conviction, and product-centric boutique investing.
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 25.3% of the talking time here. How this is scored →
Harry as informed peer 6.0 Guest teaching 6.0 Guest disagreement 3.7 Harry pushing back 5.1
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 21:29 Dismissing late-stage macro mindset
Nabeel explicitly rejects Harry's prompt about late-stage mindsets improving early investing, calling it a leftover 2021 narrative that causes investors to make wrong calls in AI.
Hardest push from Harry ▶ 14:15 Challenging non-commoditization with deployment realityHarry forcefully pushes back on Nabeel's claim that venture isn't commoditized, citing Sequoia's fund sizes and sharing real experience losing deals to deployment-focused funds.
Biggest teaching moment ▶ 7:05 Explaining principal markup incentive trapsNabeel educates Harry on how bloated partnership structures create junior VC incentive loops that prioritize quick paper markups over genuine company building.
Harry holds his own ▶ 33:58 Demonstrating deep portfolio research and product analysis frameworkHarry demonstrates high expertise by referencing interviews with Nabeel's portfolio founders, challenging product transience, and laying out his own team's CPO interview framework.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Welcoming Nabeel Hyatt and Introducing the Mindset Shift | 5 | 5 | 3 | 4 | Harry introduces the core thesis around AI investing requiring a mindset shift from B2B SaaS spreadsheets. Nabeel reframes the shift from puzzles to mysteries, while Harry tests the framing by bringing up classic SaaS success stories like Ramp and Brex. | |
| Systemic Roadblocks: Why VCs Struggle to Adapt | 6 | 4 | 2 | 5 | Harry demonstrates strong understanding of LP dynamics, pointing out that drastic firm restructuring creates LP instability. Nabeel agrees, emphasizing the need for smaller, flexible venture teams acting as sounding boards for founders. | |
| Incentive Misalignment: The Pursuit of Paper Markups | 6 | 6 | 5 | 5 | Nabeel critiques how bloated partnerships align junior VCs toward chasing paper markups rather than long-term value creation. Harry counters with Jason Lemkin's view of venture as a packaging industry, which Nabeel explicitly rejects. | |
| The Fallacy of the Coverage Game and Pattern Matching | 5 | 6 | 4 | 5 | Harry defends the need for filtering heuristics given the volume of startups. Nabeel rejects this Brita filter approach to venture, explaining how Spark intentionally stayed small to avoid transactional coverage investing. | |
| Is Venture Capital Transitioning to a Commodity? | 7 | 5 | 4 | 7 | Harry quotes Doug Leone on venture becoming a low-margin commoditized industry and cites exact Sequoia fund sizes and deployment pressure. Nabeel contends that true competition remains constrained to a small sub-segment of quality firms. | |
| The Nature of Exceptions: Spotting Generational AI Value | 6 | 6 | 5 | 5 | Nabeel explicitly rejects playbooks, emphasizing that venture is strictly an exceptions business. Harry presents a real internal investment debate about 50 competing data providers to highlight the friction between TAM logic and exception hunting. | |
| Internal Partnership Culture: Coaching and Self-Actualization | 6 | 5 | 3 | 4 | Nabeel frames mentorship as self-actualization rather than copying senior partners, and Harry validates this with a personal anecdote about trying to copy Fred Destin. They discuss fund culture differences between early and growth stage teams. | |
| Early Stage vs. Late Stage Mindset: Avoiding the Macro Trap | 6 | 7 | 6 | 6 | Harry suggests that a late-stage mindset improves early-stage investing, but Nabeel forcefully shuts down the premise as a 2021 narrative that leads to wrong calls in AI. Nabeel further dismisses macro-obsessed VC commentators. | |
| High-Service vs. High-Volume: The Limits of Obsession | 6 | 7 | 4 | 5 | Nabeel responds to Keith Raboy's hands-off VC philosophy by arguing that aiming for non-interfering mediocrity sets the bar far too low. He lays out a framework balancing execution speed against product taste using Granola as an example. | |
| Platform Risk: Defending Value Against Foundation Models | 7 | 6 | 5 | 7 | Harry raises platform risk and foundation model disruption. When Nabeel dismisses TAM calculations, Harry pushes back hard by pointing to Crumbl Cookies and Starbucks as venture-scale outcomes in non-traditional TAMs. | |
| Evaluating Founders through Product Decisions | 8 | 7 | 4 | 6 | Harry showcases deep preparation by referencing background conversations with several of Nabeel's portfolio founders. Nabeel clarifies that evaluating product is actually a window into founder execution and judgment rather than static features. | |
| Filter Calls, Personal Connections, and Sourcing | 5 | 5 | 2 | 4 | Nabeel describes his sourcing rhythm, short filter calls, and preference for long-term relational context. Harry explores whether in-person meetings remain mandatory before check writing. | |
| Valuation, Capital Efficiency, and Figma Regrets | 7 | 6 | 3 | 6 | Harry presses Nabeel on valuation boundaries during Series A deals. Nabeel notes that overcapitalization often ruins early-stage startups and names passing on Figma as his biggest missed opportunity. | |
| The Three Categories of AI Startups | 5 | 7 | 3 | 3 | Nabeel breaks down his three-part lens for AI startups: Adaptation, Evolution, and Revolution. He notes that most current AI startups are merely adapted products with fresh paint, whereas Spark targets revolution. | |
| Sustainable Value Accrual and Foundation Models | 6 | 7 | 4 | 6 | Harry questions whether value accrues to application layers or model layers, calling Spark's Anthropic investment a hedge. Harry challenges the Anthropic bull case using DeepSeek and You.com counterexamples. | |
| DeepSeek and the AI Competitive Moat | 6 | 7 | 4 | 5 | Harry quotes Grok leadership suggesting OpenAI must open source post-DeepSeek. Nabeel argues DeepSeek does not alter his thesis because capital-heavy training was never the primary moat. | |
| The Agent Economy and Avoiding Arbitrage | 7 | 7 | 4 | 5 | Harry cites a recent discussion with Manny Medina and highlights Happy Robot's voice agents in trucking. Nabeel warns that surface-level AI agent startups risk near-term arbitrage down to zero margins. | |
| Studying Successes and U.S. Economic Optimism | 7 | 6 | 6 | 7 | Nabeel rejects studying portfolio losses in favor of studying wins. Harry delivers a spirited critique comparing US economic optimism with UK fiscal stagnation, while also challenging San Francisco talent churn. | |
| Reframing the Series A Conversation Through Storytelling | 5 | 7 | 3 | 4 | Harry asks how to guide seed-stage founders toward Series A milestones. Nabeel describes his exercise of co-creating the next round narrative deck immediately after investing to focus on storytelling over raw ARR targets. | |
| The Growth Dilemma of Traditional Enterprise Software Startups | 6 | 4 | 2 | 5 | Harry brings up the plight of traditional enterprise SaaS startups growing at 2x that are overlooked by growth VCs fixated on AI hypergrowth. Nabeel candidly admits the problem is unsolved. | |
| Quick Fire: Does Wealth Enable Better Venture Investing? | 7 | 7 | 5 | 6 | Harry puts forward a strong quick-fire thesis that wealth makes better venture investors by removing downside fear. Nabeel reframes this, arguing a nothing-to-lose mentality and creative risk-taking matter far more than personal bank balances. | |
| Quick Fire: Measuring Success and finding Joy in the Game | 4 | 5 | 2 | 3 | In quick fire, Nabeel reflects on selling his company and explains why intrinsic joy in the day-to-day craft dictates success better than financial scorecards. | |
| Quick Fire: Structural Flaws in the Venture Capital Market | 5 | 5 | 3 | 4 | Nabeel outlines structural flaws in venture capital, advocating for long-term alignment over transactional term sheet mechanics. Harry wraps up the interview warmly. |
Statements from this episode (46)
Hyatt: Venture capital is now run by junior partners and associates
“The industry today is run basically by principals, associates, and junior GPs.”
Hyatt: VC principals care about internal promotions, not startup exits
“A principal is not actually waiting for an exit. They just want a promotion, man.”
Hyatt: Founders churn out short-term arbitrage projects just to raise seed
“We are in the industrialization of startups playbook land where everybody's trying to churn out some piece of ridiculous arbitrage every week in order to get through the end of their incubator and raise their seed round.”
Stebbings: Spreadsheet-driven SaaS investing will make VCs obsolete in AI
“I'm really concerned that actually the way that we've always invested Maybe more spreadsheet SaaS investing. It's gonna make us dinosaurs if we don't move with the times.”
Hyatt: B2B SaaS investing was about metrics; AI is a mystery
“Greg Treverton uses this phrase of puzzles versus mysteries, which is just like puzzles are this thing that you can like, you know, use that raw horsepower to solve. And mysteries are, you know, there you have to go on the journey. There's like fog of war and …”
Hyatt: The AI era demands rampant organizational creativity over optimization
“We were kind of like in late stage capitalism for startups. Everything was red ocean. And so it was all about optimization and speed and like minor arbitrages. We are now in the, in a world where like you need rampant creativity inside of an org.”
Hyatt: Most VC firms are failing to adapt due to LP constraints
“No, I think most, you know this cycle, because the cycle in VC evolution is like horribly, horribly slow, because it also loops back to LPs. They don't get to act on their own.”
Hyatt: AI founders need board members who actively use AI products
“And they need a sounding board of board members that are actually, actually using their products and actually have a curiosity to use the rest of AI products and get more native.”
Hyatt: Rapid-growth AI startups reaching $10M ARR will die in two years
“That will probably be dead in two years.”
Hyatt: Junior VCs prioritize quick markups from later-stage firms for promotions
“Well, that means if I want to get a promotion and I'm inside of schmooby schmooby VC firm, then I need markups. So if I need markups, how do I quickly get markups? I figure out what the guy one stage after me is interested in. And so my job is not really to go…”
Hyatt: Treating venture capital as a packaging industry is a losing strategy
“I hate that analogy deeply. Well, I to a certain extent- I understand it. I don't think it's good for startups and I don't think it's good for founders. And I also, maybe more importantly than just some kind of value judgment, I think it's a losing strategy wh…”
Hyatt: High-volume 'Brita filter' deal coverage is ineffective venture investing
“I don't know that the Brita filter version of investing is the right way to evaluate, or at least I'm not executing the way that I want to do my job and the way that I think my partner should do their jobs together in that kind of like trying to get, win the c…”
Hyatt: Spark Capital chose not to raise $5B during the SaaS boom
“We could have very easily raised five billion dollars. We could have very easily You know, tripled or quadrupled the size of the team. We didn't do that. We stayed seven people, six people partnership.”
Stebbings: 20VC lost two deals to competitors trebling valuation for common stock
“We've lost two deals in 24 months where literally they trebled the price and went to common stock.”
Hyatt: Venture capital is defined by exceptions, not playbooks
“This industry is all about exceptions. That's literally the industry we're in. And so, why are we building a bunch of playbooks if the whole thing is about exceptions?”
Hyatt: All VC measurements prior to exit are false profits and packaging
“The reason for that is that all of the measurements before exit are all false profits, right? Until the thing is actually a return back capital, and you see this wonderful, enduring institution, It's all just games and packaging”
Hyatt: I wrote only two checks during the COVID remote investing boom
“I wrote the fewest checks I've ever written in my career. I think I wrote like two checks in a year and a half.”
Hyatt: Spark Capital's early-stage deal quality dropped during COVID
“I think our quality of investing went down on the early stage team and has since recovered.”
Hyatt: Founders seeking hands-off VCs are settling for mediocrity
“They're setting their bar too low. Like that's the bottom line. Like they're settling for mediocrity because they're afraid of risk.”
Hyatt: Taste will be the main AI differentiator as execution becomes automated
“Especially in the world of AI, only taste is going to matter in the future because execution is just going to happen, right?”
Hyatt: Long-term static moats do not exist for modern AI startups
“But some kind of compounding effect where we think no matter what happens after this year, it's eBay, they just launch a product and 40 years later, the product looked basically exactly the same and it's just fine. Like that, that, those ages are not right now…”
Hyatt: Spark Capital ignores market size when evaluating investment opportunities
“We don't talk about or look at market size at all, unless it's sometimes there's a, you know, a confirmation that it's a small market.”
Hyatt: One-hour intro calls are the worst VC meeting format
“I think the one hour call is the worst call anywhere, because it's like too long to get a real read, and yet, you know, it's like, sorry, too short to get a real read, and too long to get the kind of speed dating version of the world.”
Hyatt: I passed on Discord twice before eventually investing
“Discord, Jason, I knew for seven years. We were founders together before, before investing in Discord, and also smartly passed I was twice on Discord before investing.”
Hyatt: Spark Capital won Wordware deal despite higher competing term sheets
“The last investment I did was a company called Wordware, and that was a very, very fast, very, very competitive. They had term sheets for quite a bit higher, but you know, I'm getting dinner with the founders. I'm going for walks in the morning. Like, I met wi…”
Hyatt: I passed on pre-launch Figma due to high valuation and check size
“Mine is Figma. It was quite a while ago now. It was still pre-launch, so it was trying to write like a very large check pre-launch.”
Hyatt: Most AI startups are adapted products with superficial AI branding
“So we mostly have evolved products that are not good enough, or we have adapted products with a coat of paint on top that says AI.”
Hyatt: AI models must own the direct user interface to stay ahead
“I think you want to own the interface with the customer, and if you own the interface with the customer, you have the chance to iterate on an interface with the customer and stay ahead of everybody else.”
Hyatt: Pure compute-heavy AI labs without consumer applications will fail
“So if I compare that to somebody who's spending a bunch of money on compute for an academic lab for a very large foundational model I don't think those things accrue very well over time. I think you need to be full stack.”
Hyatt: We invested in Anthropic because capital isn't the only AI moat
“If we had believed that only capital was going to win, Then we would not have invested in Anthropic, because you know, surely Sam was telling us and everybody else, like, we're going to win the capital game. The capital game is the only way to win, and so ther…”
Hyatt: Frontier AI companies must build vertical products, not API-only tools
“I think when you're at the front end of innovation, when you're moving really fast, I think you want to be vertical. I think you want to have as much connectivity layer between the model you're trying to build and The thing that the consumer is trying to wrest…”
Hyatt: OpenAI pays PhDs to manually solve math problems for model training
“OpenAI and lots of other companies are paying a bunch of PhDs to, like, by hand go, you know, figure out PhD math equations so they can internalize the model.”
Hyatt: Web 2.0 was the wisdom of crowds; AI is the wisdom of experts
“I think web two point now was very much the kind of wisdom of crowds, and we're at an age where it's the wisdom of experts.”
Hyatt: Expert data exhaust improves both AI products and underlying models
“If you are running a next gen product with AI deeply embedded, and you have the best users using that product, that will inform you to make better products for those users and inform the models that you're building.”
Hyatt: Most AI agent startups are near-term arbitrage risking market collapse
“I think most of them fall into, if you just really think through the second order effects of it, fall into kind of like near-term arbitrage, which might just take that whole market to zero, especially if you are meeting the market where it is today with the mo…”
Hyatt: AI founders should build for model capabilities six to nine months ahead
“If you are doing something at the very edge, I mean, I definitely am constantly trying to advise founders and encourage founders to think about what the models might do in six months or nine months and start to chart there.”
Hyatt: VCs should focus on analyzing successes rather than studying losses
“I don't know that you should look at your losses. I mean, maybe if you felt like the decision that you made at the time, if you look back on it and like you just were in a bad place when you made the choice and you should try to not be in the bad place the nex…”
Hyatt: AI startup founders must build in San Francisco to minimize risk
“And so if that's true, and you're trying to risk mitigate all the things that are going to kill you, and it's the age of AI, I don't know why you're not in San Francisco.”
Hyatt: I outline the next round's pitch deck immediately after investing
“About 20% of the time, 25% of the time, founders are down for it. After we invest, I immediately try and have a conversation and get a pitch deck together for the next round. Like, what would the next, just a glossary, just a table of contents. What would you …”
Hyatt: Founders default to financial numbers when they lack a real story
“And I think people default back to numbers when they have no other story to tell.”
Stebbings: VCs no longer find doubling enterprise ARR exciting due to AI
“It's not even they're stagnating. It's they're doubling. And that's not exciting enough now for VCs who are used to AI revenues”
Hyatt: Investors shouldn't evaluate AI companies by their underlying models
“I don't think you should evaluate any company by the models that are underneath it. Even model companies.”
Stebbings: Personal wealth makes venture capitalists better investors by removing fear
“I have a strong thesis that it does, because you no longer worry about downside. You no longer worry about your next fund. You no longer worry about Protection.”
Hyatt: A job protection mindset leads to the biggest VC investment mistakes
“And so I came in with a nothing to lose mentality, which allows you to sit on the front end of creative risk and being willing to take that extra risk, which of course, that's what this business is about. Versus this kind of protection is like, I just want to …”