Oct 21, 2024 · 1h 3m · news
Mamoon Hamid: AI - Where Value Accrues, Startups vs Incumbents & Scaling Laws | E1217 · 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 episode of the 20VC podcast, venture capitalist Mamoon Hamid of Kleiner Perkins shares his investment philosophy, breaking down his high-conviction thesis on the AI application layer, the mechanics of market-defining early-stage investments like Figma and Slack, and lessons learned from navigating economic cycles.
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 26.1% of the talking time here. How this is scored →
Harry as informed peer 4.2 Guest teaching 3.8 Guest disagreement 1.2 Harry pushing back 2.4
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 43:20 Mamoon explicitly disagrees on founder fundraising dogma
When Harry asks if founders should always raise as much money as possible at the highest price, Mamoon directly disagrees, emphasizing that early-stage success is about partnership alignment over maxing out valuation.
Hardest push from Harry ▶ 15:40 Harry challenges the business quality of LLM providersHarry forcefully refuses to accept that foundation models represent good businesses, citing rapid price dumping and quoting Sarah Tavel's description of LLMs as the fastest depreciating asset in history.
Biggest teaching moment ▶ 18:22 Mamoon breaks down global GDP and labor markets to reframe CapEx concernsMamoon educates Harry on macroeconomics, using $100T global GDP and its 50-60% labor composition to demonstrate how AI capturing labor value easily absorbs multi-hundred-billion-dollar CapEx numbers.
Harry holds his own ▶ 15:40 Harry demonstrates deep domain research on LLM pricing dynamicsHarry shows strong host expertise by citing specific VC commentary from Sarah Tavel and pointing out real-time token price collapses to pressure the guest on LLM unit economics.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| The AI Super Cycle and Market Dynamics | 5 | 3 | 1 | 2 | Harry sets the macro context of the AI super cycle versus the 1997 internet boom. He demonstrates background knowledge by citing Larry Ellison's claim that entering the frontier model race costs $100B. Mamoon agrees and elaborates on incumbent spend and application layer potential. | |
| AI Value Accrual and the Application Layer | 5 | 3 | 2 | 4 | Harry presses Mamoon on category crowding, asking how differentiation is possible when ten alternatives exist for every niche like medical transcription. Mamoon explains that fine-tuning model accuracy from 87% to 99% requires deep technical expertise rather than tourist founders. | |
| AI vs. Traditional SaaS and Navigating the Pricing Environment | 6 | 2 | 1 | 3 | Harry demonstrates active market tracking by referencing three pre-product deals he saw evaluated at $750M pre-money in a single week. Mamoon explains fund math constraints and the concept of an occasional 'YOLO bucket' exception. | |
| Revenue Scaling, Custom Tooling, and Proof of Concepts | 4 | 5 | 1 | 2 | Harry brings up fast revenue scaling and internal tooling trends like Klarna replacing Workday and Salesforce. Mamoon educates Harry using history, explaining how Kleiner Perkins wasted millions building an in-house CRM before adopting Affinity, showing why specialized software wins. | |
| Future AI Interactions and Middle-Layer Hype | 2 | 3 | 1 | 1 | A lighthearted segment where Mamoon critiques over-investing in middleware and vector databases as fleeting value. The conversation turns conversational with anecdotes about CEO hold music. | |
| The Viability of the LLM Layer and Scaling Laws | 7 | 3 | 2 | 7 | Harry delivers strong, well-informed pushback against the LLM layer, citing Sarah Tavel's analysis of rapidly depreciating assets and aggressive price dumping. Mamoon acknowledges the current margin challenges of token sales while arguing long-term economics will reflect cloud infrastructure margins. | |
| The Capex vs. Revenue Gap and Tech GDP Growth | 6 | 6 | 1 | 3 | Harry cites David Cahn's $600B AI CapEx question to challenge host spend assumptions. Mamoon schools Harry by breaking down the $100T global GDP and labor market share to reframe why tech market capture will absorb CapEx costs. | |
| Aaron Levie, Box, and the Lessons of Down Cycles | 5 | 4 | 2 | 4 | Mamoon shares Box's near-death experience in 2008 requiring three bridge loans. Harry challenges the wisdom of ignoring market signals and continuing to fund a company when the market is rejecting it, invoking Warren Buffett and Charlie Munger. | |
| Why Startups Plateau and the Art of Selling | 3 | 4 | 1 | 1 | Mamoon reflects on why breakout companies plateau and details his framework on selling at the 'local maxima' of perceived value, sharing the Yammer acquisition narrative. | |
| Best Performing Investments and the Truth About IRR | 4 | 6 | 1 | 2 | Mamoon explains venture return metrics, revealing that a 70x outcome over 15 years actually yielded only a 15% IRR. The two also evaluate M&A stagnation and why regulatory pressure beyond the FTC impacted deals like Adobe/Figma. | |
| The Figma Investment and the Prepared Mind | 4 | 5 | 1 | 1 | Harry asks what Mamoon saw in Figma before revenue scaled. Mamoon details tracking DAU/MAU and L28 cohort metrics showing designers using WebGL in browser 15-18 days a month. | |
| Market Creation versus Competitive Landscapes | 3 | 4 | 1 | 2 | Harry brings up founder feedback regarding Mamoon's archetype preference. Mamoon details his two favorite founder profiles: hyper-obsessed product creators in new markets and ambitious repeat founders. | |
| Valuation Discipline and Premium pricing for Experience | 4 | 3 | 2 | 2 | Harry asks about valuation premiums for experienced founders. Mamoon shares examples like co-leading Glean at $35M post with Arvind Jain to illustrate fair pricing alignment between founders and GPs. | |
| Fundraising Cadence, Market Pull, and Valuation Multiples | 5 | 4 | 1 | 3 | Harry points out that Mamoon backed Slack at $250M post when it only had $500k ARR (500x ARR multiple). Mamoon explains why revenue multiples are misleading at the early stage when engagement data proves massive market pull. | |
| VC Echo Chambers, Sourcing fallacies, and Voting Structures | 4 | 2 | 1 | 2 | Harry and Mamoon align on the flaws of VC voting structures and proprietary data platforms. Mamoon clarifies that Kleiner Perkins operates on partner conviction and open table discussions rather than formal voting. | |
| Scenario Planning, Team Dynamics, and Deployment Speed | 3 | 3 | 1 | 1 | Mamoon rejects probability-weighted scenario modeling as 'false precision'. He shares how Kleiner Perkins deployed their 2019 fund in 15 months into winners like Rippling and Glean. | |
| Stage Specialization and the Hard Lessons of Capital Loss | 3 | 5 | 1 | 2 | Mamoon discusses stage plasticity and candidly recounts losing $30M on consumer lending startup Tally due to rising interest rates, reflecting on the lessons of capital loss. | |
| Investor Evolution, the Dreamer Persona, and Board Governance | 3 | 4 | 1 | 3 | Harry confronts Mamoon with peer feedback that he is such a dreamer he sometimes holds on to struggling investments too long. Mamoon accepts the criticism as part of his core identity as a founder champion. | |
| Quick-Fire Insights on Industry, History, and Faith | 3 | 3 | 1 | 1 | A quick-fire round covering AI excitement, investor respect, and OpenAI. Mamoon discusses how personal faith fundamentally guides his humility, treatment of founders, and board participation. |
Statements from this episode (51)
Hamid: AI cycle is 10x larger than the 1997 internet boom
“We're in the midst of a super cycle like none we've seen before. The AI super cycle, as you know, and it reminds me of the time when I first came to Silicon Valley in 1997. I was 19 years old, and it was all just roses all around me. It was the rise of the int…”
Hamid: AI applications will create trillions in value over next decade
“There are so many things to build on top of this infrastructure, all these frontier models that is going to create so many trillions of value over the next decade.”
Hamid: Incumbents and VCs are over-investing in AI due to FOMO
“Everyone's over-investing right now into this cycle, and because none of us can miss, whether it's the large incumbents, Or us as venture investors back in companies.”
Kleiner Perkins focuses AI investments on co-pilots for top-earning professions
“We've invested in a lot of application layer companies, and that are solving very specific pain points. And the way we've looked at it pretty simply is, if you think about, we took actually the top 20 jobs in the US, and who makes the most? Simple. And it's do…”
Hamid: AI medical transcribers must achieve 99% accuracy to be viable
“You can't have a medical transcriber that's 87% good, ok? It has to be close to like 99% good. And that actually requires real Technical depth and adeptness.”
Stebbings: Three pre-product AI startups raised at $750M+ valuations in one week
“I saw three companies my moon last week that raised it over the seven hundred and fifty million pre-product.”
Hamid: Kleiner Perkins needs 15-20% ownership on $5M-$10M early-stage checks
“Because as you know, Harry, you know, we have to get our ownership at the early stages where you're investing five to ten million dollars for 15 to 20% for the math to work for our funds.”
Hamid: Kleiner Perkins reserves a 'YOLO bucket' for rule-breaking deals
“And in some ways we have a, what we call like a Yolo bucket in our funds and where you just have this extreme conviction around the founder and the company where you're sort of willing to break the rules.”
Hamid: AI SaaS startups command $300 to $500 monthly per seat
“What we're seeing right now is that you have seat-based pricing that was like, 30 dollars a month, 40 dollars a month, and now you're getting 300 dollars a month, 400 dollars a month, even 500 dollars a month for the software.”
Hamid: Kleiner Perkins spent millions building and maintaining an internal CRM
“We built an internal CRM at Kleiner Perkins, and I think it sort of cost us many millions of dollars, and then ongoing millions of dollars a year to just upkeep.”
Hamid: Every major enterprise CIO is funding AI proofs of concept
“We're in this era of doing a bunch of proof of concepts. You're just trying out all this cool stuff that's come into existence in the last two years and seeing, what can I do with it? And every CIO at every large company is spending real money doing POCs.”
Hamid: AI agents will replace human customer support within 10 years
“I think we will hopefully never talk to a customer support agent ever again. Like someone you call for the airlines, like, you know, help me my flight, I need to upgrade it, or I need to change my seat, or can you cancel it because I can't go? You know, the ba…”
Hamid: AI middleware is overinvested and its value will be fleeting
“There's a lot of time being spent on a lot of the middle layer between the foundation models and the applications. And I think in the middle layers, middleware, things that allow you to use those models better, faster, cheaper and build applications on top. So…”
Hamid: OpenAI will figure out gross margins and become highly profitable
“Today it's not a great business, but, you know, I think they're smart enough to figure out how they can get to a gross margin that will allow them to be a highly profitable business over time.”
Hamid: LLM token prices dropped 200x over the last 18 months
“In the last 18 months the price of a token has gone down by 200 X”
Hamid: LLM token prices will fall another 10x-20x within two years
“So, will it go down to the next two years? I don't know, but it will probably go down by 10 X or 20 X”
Hamid: Tech sector will gain $10 trillion in annual spend within decade
“So, 10 trillion of annual spend will get created for technology companies over the next decade.”
Hamid: AI revenue will justify CapEx by expanding into global labor markets
“I think the revenue will be there for, because again, we're not just tackling software, linear software, as I call it. We're tackling Like labor, and labor shortages, and things that humans can do, but it's the worst part of their job, or we don't have enough …”
Hamid: Venture capital is a business that does not scale with headcount
“It's, we think it's a business that doesn't scale, actually.”
Hamid: Half of Kleiner Perkins' growth fund backs its early-stage winners
“Because our growth fund, even, half of the dollars are allocated towards, not allocated, but half those dollars are invested in our best companies from our early stage funds.”
Hamid: Box was one week from running out of cash before IPO
“I've had companies that are one week away from cash out become public companies.”
Hamid: Box raised 2008 bridge financing at a $25M valuation
“We got to invest more dollars in the box at a twenty-five million dollar valuation. More, like, every incremental dollar or, you know, two, three million dollars in bridge that was being done was done at that valuation.”
Hamid: Aaron Levie's Box IPO dilution led VCs to re-up founder equity
“Aaron has done a great PSA for founders with his equity stake at the IPO, where I think we as investors, board members have done a way better job making sure that our founders get re-upped when, especially when they dip below a certain threshold.”
Hamid: Sell startups at the local maximum of market-perceived value
“Local maxima that I think about sometimes with companies where this is sort of the local maximum in terms of perceived value by the market for a company, and that's a great time to sell, and so now you have to figure out when is the local maximum for a company…”
Hamid: Yammer's $1.2B sale freed him to invest in Slack
“There's one particular example that I think ended up working out pretty well. This is when Yammer got acquired by Microsoft it was a 1.2 billion dollar acquisition. At the time it felt like, man, we've got so much ahead of us. Yammer can become the next Slack,…”
Hamid: John Doerr has held his Google stock for over 25 years
“You can hold on to your Google stock forever as John Doerr has done over the last 25 years.”
Hamid shares entry valuations for Slack, Figma, and Rippling
“Probably Slack, I would say, you know, even at the 250 post that's, you know, with all the dilution over time and, If you take the twenty-seven billion or some other number that's a, obviously, a good, great multiple. Figma, you know, the initial investment wa…”
Hamid: 15-year hold on a 70x multiple deal yielded only 15% IRR
“There's one investment I remember doing early days of USVP, Where we did it at, I think, TenPost. And it was a three million dollar check for like 30% of the company. And that company, about a year and a half ago, the founder CEO, Steve Flagg, still a friend, …”
Hamid: FTC and Lina Khan did not block the Adobe-Figma deal
“I don't think, ah, a lot of us put a lot of the blame on Lena Khan, poor Lena Khan because, you know, in our own experience, that's not, that was not the issue for Adobe Figma. It wasn't Lena, ah, or the FTC, ah, there were other issues. So I think it's broadl…”
Hamid: 2025 will be a strong year for tech IPOs
“I'm really hopeful. I think next year will be a good year for IPOs.”
Hamid: Early Figma data showed designers using the product 15-18 days monthly
“You just saw that designers were using the product 1516, 1718 days out of a month.”
Hamid: Sketch and InVision were significantly larger than Figma early on
“But Sketch and InVision are way bigger than Figma for a while.”
Hamid: Slack and Figma created entirely new software market categories
“Slack created a market. In some ways, Figma created a market. There wasn't a notion of collaborative design software.”
Hamid: Avoid founders who rely on top-down market mapping
“I don't like the, we looked at the landscape, and we discovered that this is a great place to build a business. Like, we did a whole market mapping exercise, And the TAM is going to be X billion dollars and, you know, we've got yeah, it is that sort of like th…”
Hamid: Kleiner Perkins co-led Glean's early round at $35M post-money
“We'll be back to Arvind Glean. We did it as a co-led it with Lightspeed at 35 post. And Arvind is a G. You know, he started Rubrik. He's like one of the, you know, Google fellow type of person. He raised a lot. He's like fifteen million at 35 post, so gave up …”
Hamid: Kleiner Perkins backed Aleph at a $35M post-money valuation
“Or another example is Syed Ali at Aleph. You know, I think we did that also at, like, 35 post. And Syed had come off of a company just sold for six billion dollars.”
Hamid: Kleiner Perkins avoids backing great founders in bad markets
“Bad markets, the structure of industries where margins are compressed and customers are bad, life's just too hard that way... Probably not.”
Hamid: Invested in Slack at a $250M valuation on $500k ARR
“It was 500 K of ARR at the time... Two hundred fifty million post.”
Hamid: Valuing early-stage startups on revenue multiples is a mistake
“Yeah, I think it's really a mistake to look at revenue multiples at like half a million.”
Hamid: Data-driven VC sourcing platforms are mostly BS
“I think it's mostly BS. Yeah. Yeah. I think many have tried and failed at using the very data oriented approach to investing in startups, when at the end of the day it's about the founders and-”
Hamid: Early-stage VC firms should not use voting structures
“Yeah, I don't believe in voting structures for early stage venture capital. I think you have to have, allow every partner to put themselves on the line with their conviction that they've built up over a year sometimes weeks sometimes, but you have to give into…”
Hamid: Outcome scenario planning in venture capital is false precision
“It's such false precision that I think it's, it feels like busy work actually.”
Hamid: KP's Figma deal was contentious at $500k ARR and $110M valuation
“You know, actually Figma was quite contentious. Yeah, it was, yeah. Because to your point, around like half a million of AR at a 110, 15 post, whatever it was at the time it was really like the same thing. There's Envision, there's Sketch. This company's been …”
Hamid: Kleiner Perkins deployed its 2019 venture fund in 15 months
“So there's a period of time in 2000 19, so we just raised our first fund as a team together, where we deployed that fund, like, within, like, 15 months.”
Hamid: VCs lack neuroplasticity to switch between early and late stages
“I think generally speaking, though it's hard to have the neuroplasticity to one day think about you know, 10 years ahead and a new infrastructure company that's building on a new open source framework, and the next day think about, like, pre-IPO stage, you kno…”
Hamid discloses $30 million loss on shutdown startup Tally
“Thirty million or so, and that's multiples of the largest loss prior to that, so you know, I don't wear this badge of honor of, like, losing a lot.”
Hamid: Consumer lending startups are exceptionally difficult venture bets
“That lending businesses are really hard. Consumer lending is very hard.”
Hamid: Effective board meetings focus deeply on only one or two topics
“I like board meetings where there's one or two things that are talked about in detail. Meaning, like, you go deep dive into one or two things, because at any given point in time in a company's juncture, that moment in time, one or two things that really matter…”
Hamid: Prefers capital-efficient startups over capital-intensive models like DoorDash
“I, I'm, those are not my kind of businesses. I like more capital efficiency but I can get behind them.”
Hamid: Kleiner Perkins will not expand into a massive capital accumulator
“I don't think we want that at all. I think we love where we are today. We really do, and I don't try to, I'm not BSing you. You're a long time friend, and I believe in venture, early stage venture being a beautiful asset class, especially if you can follow the…”