20VC: You Need $1BN To Do Seed Today: The Untold Truths of Venture | Who Wins and Who Loses in the Next Decade of VC | Why Margin and Price Matter Less in a World of AI with Sameer Dholakia, Bessemer

Fireworks AI

• Sameer Dholakia · Bessemer Venture PartnersBullish

  Leading open-source model providers like Fireworks can build very large businesses with venture returns.

  Asked whether open models are much harder to monetize, he says it absolutely can build a big business, though the exact scale is uncertain. He cites open-source precedent such as Red Hat and databases, where enterprises prefer a provider with the right talent behind open software. Bessemer is an investor in both Anthropic (closed) and Fireworks (open), and he thinks the open versus closed token mix matters less than the size of the overall market.

  “you can build a very large business and get venture returns. If you are the leading provider of a open source solution... most enterprises don't wanna be responsible for the open source.”at 13:28

• Harry Stebbings · 20VCBullish

  Fireworks will win the open-model inference race, and 20VC has money invested there.

  He says competition among AI infrastructure providers has deepened, but expects Lagora and Harvey to do well in legal, and base 10 and Modal to do well too. He picks Fireworks as the winner, which is why 20VC's money is invested there.

  “When we look at a fireworks, I think base 10 and modal will do well too. I think fireworks will win, which is why my money is there.”at 23:14

• Sameer Dholakia · Bessemer Venture PartnersBullish

  Fireworks is generational; enterprises will own their intelligence; sees 20x at $15B entry.

  Sameer calls Fireworks one of his favorite companies and says its CEO Lynn is generational. He cites a strong team: Lynn as strategic technologist and deep technologist, plus George Hu, who joined as president after building go-to-market at Twilio. His core market pitch is that enterprises will not want to rent intelligence from someone else and will want to own some of their own, so healthy demand for owned models follows. He says Bessemer and the other investor both entered at a $15B price and that he sees a 20x return on that entry. He concedes the company may not be a $100B-market-cap generational outcome in the strict sense, but says it is extraordinary and that he wishes he had backed up the truck more.

  “one of my favorite companies, period, is Fireworks. I think Lynn is generational.”at 56:35

Anthropic

Across the shows: 44 speakers in other episodes, 24 bullish and 47 bearish calls, past 30 days

• Sameer Dholakia · Bessemer Venture PartnersBullish

  Frontier model labs will consolidate like hyperscale clouds; Anthropic is a winner and no bear case exists.

  He used a cloud-redux analogy: training frontier models is massively compute- and capital-intensive, so as in cloud, only a handful of players with big balance sheets (AWS, Microsoft, Google) can compete, and hundreds of applications will sit on top. AWS's investment and compute access were pivotal to Anthropic. Bessemer led into the Series D at about $18B post-money. He sees no bear case for frontier labs because intelligence is solving problems that were previously unsolvable and humanity has many problems left. He concedes he underestimated how big a frontier business could become.

  “I think this market is gonna be a cloud redux was my analogy that in the cloud era, you had a handful of hyperscalers. ... And so it's going to consolidate to a handful of players. ... Do I think there's a bear case for the frontier labs? I really don't.”at 8:28

• Sameer Dholakia · Bessemer Venture PartnersBullish

  Anthropic is a conviction bet; Bessemer put 20% of its last growth fund into it.

  Sameer says Bessemer's last growth fund put about 20% into a single name, Anthropic, and that he feels good about that concentration. He says the firm will flex position size up when it believes a company is generational. The position size is the evidence he offers for his view that Anthropic is one of those companies.

  “our last growth fund, our century two, I think we had 20% into one name. The name was Anthropic. I feel pretty good about it.”at 36:14

Legora AB

• Sameer Dholakia · Bessemer Venture PartnersBullish

  Legora's legal AI market is huge and its growth from $20M to $200M ARR justified the $1.8B entry.

  Bessemer led Legora's round at a $1.8B valuation when ARR was about $20M; he says the company has since grown to about $200M ARR, so that entry looked insane but worked. He accepts Harry's TAM math (a $300B US legal market, a 20% software take, so $60B). He adds that global expansion adds roughly 60% more TAM, and that about $1T of legal services spend exists globally, with many legal needs unserved because prices are high. Capturing a small share could support a $5-10B revenue business worth a reasonable multiple.

  “we invested at [$1.8B valuation] leading that round, and that seemed insane at the time, you know, but the company just went, you know, 20 to 200. ... I feel pretty good about that.”at 19:21

Field Guide

• Sameer Dholakia · Bessemer Venture PartnersBullish

  Field Guide, AI for auditors, is ripping after an early Bessemer entry at sub-$10M ARR.

  Field Guide builds AI for auditors and Bessemer invested early, when it had under $10M of ARR, at an entry price of a little under $200M. Sameer says the company is growing very fast and that its inflection came after the early-2026 model leap. Once the models improved, the solution became far more compelling and uptake followed. He says Bessemer has not yet been able to back up the truck on it, so the position is still small relative to the upside he sees.

  “And I did that one early. They were, I think, they were sub 10,000,000 of ARR at the time we invested. It's probably a little under a $200,000,000 entry price, and they are just ripping. They are growing really fast.”at 38:38

Waymo

Across the shows: 3 speakers in other episodes, 2 bullish and 1 bearish calls, past 30 days

• Sameer Dholakia · Bessemer Venture PartnersBullish

  Waymo's driverless tech can become a trillion-dollar business beyond ride-hailing, via personal cars, trucking and logistics.

  Sameer argues that ride-hailing is not the only market for Waymo's autonomous technology. He says Waymo has the opportunity to sell an autonomous driver subscription for every OEM's cars, which he compares to Tesla's FSD subscription, and that families would make it a default purchase. He notes Waymo is not prioritising this near term because it is focused on ride-hailing, which he says it is winning, rolling out nationally and globally. He then says that applying the same technology to personal car ownership, trucking and logistics opens a much larger market, and concludes Waymo will be an extraordinary, roughly trillion-dollar business. No timeframe is given.

  “Waymo's gonna be an extraordinary, I think, a trillion dollar business. Easy.”at 1:11:05

Venture capital & startups

• Sameer Dholakia · Bessemer Venture Partners

  10 to 15x next-twelve-month revenue multiples are reasonable for hypergrowth AI companies.

  He values companies on next-twelve-month (NTM) revenue, looking at what the business can be in a year and whether the multiple is reasonable. Paying 10 to 15 times NTM is not crazy for businesses growing at rates never seen before, because they will grow into the price. He applies this to deals like Anthropic (around 150x current ARR at entry) and Legora (about $1.8B on roughly $20M ARR).

  “something that's in the 10 to 15 times NTM, I don't think is crazy, especially if there are these businesses that are growing at rates that have never been seen in the history of business.”at 18:14

• Sameer Dholakia · Bessemer Venture Partners

  Revenue quality matters: sticky enterprise NDR deserves higher multiples than prosumer or consumer revenue.

  Drawing on his enterprise B2B background, he prefers gross and net dollar retention among highly sticky customers for whom the product is strategic, since they keep expanding. He would pay a much lower multiple for prosumer or consumer businesses. Harry raised churn as a risk that can kill a business.

  “I prefer the gross dollar retention rates, the net dollar retention rates of highly sticky customers that for whom your thing is super strategic... So I would pay a much lower multiple for, like, a prosumer y business or a consumer business.”at 20:00

• Sameer Dholakia · Bessemer Venture Partners

  Capital is a moat: well-funded AI startups win brand and market leadership, which is self-reinforcing.

  He argues competitive breadth is no bigger than before; what has changed is the capital intensity of competition, because compute is scarce and expensive. Heavily funded players like Legora and Harvey can scale quickly, build a brand (big law firms name them as the major legal AI companies), and attract talent in a flywheel that wins customers. Capital is necessary but not sufficient; the product must also be exceptional.

  “I actually don't believe that the competitive breadth has changed. What's changed is the capital intensity behind the competition, and therefore, it leads to this competitive intensity.”at 23:34

• Sameer Dholakia · Bessemer Venture Partners

  A billion-dollar seed fund is needed today because seed valuations are factors higher and AI deals are capital intensive.

  Seed checks used to be $3-5M at $20-40M valuations. Today valuations are factors higher. To find the winners you need many shots on goal, and that is hard for a small seed fund. Many AI businesses are capital and compute intensive, so the math won't work for entrepreneurs or investors who pay less than current prices. Early-stage specialists can avoid this only where capital intensity is low, such as software without hardware or model training.

  “Do you need a billion to do seed? Seeing the prices that I'm seeing, valuations we are seeing for the early stage deals... I don't know how you would have a small seed fund and be successful. I don't.”at 28:22

• Sameer Dholakia · Bessemer Venture Partners

  Mid-sized venture firms will struggle; success requires being a giant global platform or a true specialist.

  As venture professionalizes and goes global across stages and sectors, industries consolidate. Winners are either giant global platforms that consolidate across stages, sectors and geographies, or extreme specialists who add differentiated value and pick better. Firms in between face a hard path. He cites autos (the big three and Toyota/Honda versus niche players like Porsche and Ferrari) as the pattern. He says Bessemer is a big global platform.

  “And if you're in between, good luck. I think that's a rough place to be. ... not being a giant global platform or an incredible specialist who can add differentiated value and be better at picking... I think it's a hard world to make money.”at 30:57

• Sameer Dholakia · Bessemer Venture Partners

  Companies will stay private longer; private VC is structurally bigger, so public listings are rarer.

  Sameer says it is a permanent structural shift that companies will stay private longer. He points to the size of the private venture asset class, with many more dollars and firms than before. LPs are allocating more to the asset class because the biggest gains of the AI boom will already have happened in private markets. Crossover investors will spend more time in private rounds, so public investors would miss most of the run-up.

  “I believe that it is a permanent structural shift that companies will stay private longer. The asset class of private venture capital is just that much bigger. There are that many more dollars in the market,”at 52:00

AI: overall outlook

• Sameer Dholakia · Bessemer Venture Partners

  Intelligence will be a bigger market than oil and this AI super cycle exceeds the last two combined.

  He has seen a trillion-dollar super cycle follow each prior technology moment he called magic (the internet, VMware virtualization and cloud). He calls AI bigger than the last two combined. His analogy: oil powers machines, while intelligence powers all humans (consumers, SMBs, enterprises, government), so demand for intelligence is effectively universal and the market scale is far larger than oil.

  “and this one is bigger than the last two combined, in my humble opinion. Why do you say that? The analogy I often try to draw is intelligence is gonna be bigger as a market than oil because oil powers all the machines.”at 7:06

AI pricing & unit economics

• Sameer Dholakia · Bessemer Venture Partners

  AI companies' gross margins rise meaningfully with scale, starting from much lower levels at entry.

  He says economies of scale and share matter a lot. The gross margins of businesses he invested in during the AI wave were meaningfully lower at the time of investment than they are at scale today, by factors. He is not worried about near-term gross margins, since he holds for five to seven years. Any model that assumes margins rise to 60-70% must be defended with a clear reason why that is possible.

  “you know, economies of scale really matters. Share and position really matters. ... I can remember most of the businesses I invested in, gross margins at the time we invested in this AI world were meaningfully lower than they are at scale today.”at 21:04

IPOs, M&A & deal activity

• Sameer Dholakia · Bessemer Venture Partners

  IPO window will unlock next year, later next year, and 2028, driven by demand for AI exposure.

  Sameer expects the IPO window to reopen after being shut for a long time, with listings coming next year, later next year, and in 2028. He reasons on supply and demand. Investors want pure-play AI exposure, and they want it across dozens or hundreds of companies rather than a few. That demand will be strong enough that investors pay up, and prices will move to bring supply to market. He also says companies that list become more disciplined, though he concedes the case is arguable.

  “I also believe that we will see an unlock in the IPO markets next year, later next year, and in 2028. I think a lot of companies will go public,”at 52:46

Private markets for individual investors

• Sameer Dholakia · Bessemer Venture Partners

  Retail should only buy top-brand platforms or specialist venture funds; most should stay in public markets.

  Sameer says retail investors are welcome in principle, but a random retail fund would attract adverse selection, because institutions such as hospitals and pension funds would pick the best managers first. Retail should only enter through giant global platforms with brand, leverage and power, or through very specialised managers, and only if they accept that venture is high risk. With public markets having run up strongly, he says a retail investor who is uncomfortable with that risk should simply buy public markets and avoid the liquidity hit.

  “Retail should only go into... either giant global platforms with brand and leverage and power or super, super specialized folks, and you know you're taking risk. And if that doesn't feel comfortable for the retail investor, they shouldn't play”at 56:35

Frontier AI labs & model race

• Sameer Dholakia · Bessemer Venture Partners

  Countries and regulated buyers will want sovereign models, which Bessemer is backing in India and France.

  Sameer says he buys the sovereign model argument and that Bessemer is invested in Sarvam in India and in Mistral. In India, Indic languages and multimodal data create a local case for models trained on local content, and India has deep technical talent. For government use cases and regulated industries, a sovereign model will feel more comfortable to buyers. Because the intelligence market is so large, not every buyer needs to choose sovereign, but enough of them will to build a very large business.

  “look, at the end of the day, like, if you're trying to determine whatever my military strategy for fill in the blank, there are certain use cases where if it's a government use case or a regulated industry or whatever, that having it be a sovereign model is probably gonna feel more comfortable to that buyer.”at 1:00:52

Open-source AI models

• Sameer Dholakia · Bessemer Venture Partners

  The open token economy will be far bigger than he expected a year ago.

  Sameer says the biggest change in his thinking over the last twelve months is his appreciation for how big the open token economy will be. He did not see it a year ago, and it is a large part of why Bessemer is investing as heavily as it is in companies such as Fal and Fireworks. Both businesses sit across the stack in generative media and infrastructure. He gives no quantified market size, so the view is a directional thesis.

  “In the last twelve, I would say just my deep appreciation for how big the open token economy is gonna be. I just didn't see it twelve months ago.”at 1:07:56

Robotics & autonomous vehicles

• Sameer Dholakia · Bessemer Venture Partners

  Embodied AI and autonomy will change work and life; robotics is about where gen AI was in 2021.

  Sameer says he is excited about robotics and embodied AI. Bessemer has invested nearly $200M in Waymo, and he expects autonomy to change how people work and live. He compares embodied robotics today to generative AI around 2021, which he calls a starting point, and says the field has a couple of years to go. He thinks it will move faster than most people expect. He is also looking at autonomy in construction, mining and agriculture.

  “I'll say two. One is I'm wildly fired up about all things robotics and embodied AI. Like, we invested nearly $200,000,000 into Waymo, and I just think that, you know, autonomy and what that changes and how that changes the way we work and live, I think, is awesome.”at 1:08:38

Pharma & biotech

• Sameer Dholakia · Bessemer Venture Partners

  AI will drive the fastest drug discovery and biology progress, compressing decades of therapeutics work into a decade.

  Sameer argues that using AI to 'drug the undruggable' and find cures is by far the greatest potential and impact of the technology. His mechanism is that wet labs will become automatable, and an AI system that can see and reason over massive datasets in ways a single human cannot will speed up discovery. He expects the pace of discovery in biology, therapeutics and drug development over the next decade to deliver what would otherwise have taken half a century, and says he believes Dario Amodei's view on this.

  “I think we're gonna see in the next decade what would have taken us half a century. I really believe that.”at 1:11:48