20VC: Trillion or Bust: The New VC Game | Coreweave Sell-Off Creates Panic | Town at $1B, Anduril at $90B: Revenue Multiples Out of Control | General Atlantic Eyes IPO & Modal Raises $750M at $15.75B

Modal Labs

• Jason Lemkin · SaaStrBullish

  Modal is a better business than CoreWeave because it adds value rather than just renting GPUs.

  Jason says he likes Modal's $15.75B round more than CoreWeave, because Modal adds value to the ecosystem rather than being an outsourced rack of GPUs. He acknowledges the competition is infinite and limitless, so the company needs strong conviction from its growth investors. He also says the round is a strong one and that Modal is one of the most interesting rounds he has seen.

  “I like it better than CoreWeave. I like adding value to the universe rather than just being some outsourced racks of GPUs.”at 32:17

• Rory O'Driscoll · Scale Venture PartnersBullish

  Modal's growth has accelerated sharply, and inference demand is exploding, supporting its rise.

  Rory says Modal was always seen as strong by developers, but ran at roughly $60-100M run rate in the middle to late part of last year. The acceleration to a $1B run rate is stunning, which shows that inference is exploding. Modal is developer friendly and has a range of products. He points to the big market opportunity in inference, where the top providers together are only around $5-6B in run rate revenue, while open-weight models are served across many platforms.

  “The interesting thing about Modal is they were always perceived strong by developers... And that level of acceleration is just amazing, right?”at 32:45

• Jason Lemkin · SaaStrBullish

  Modal's round is cheap; the deal makes money even if margins never reach the 40s.

  Lemkin argues Modal's valuation is cheap relative to its fast revenue growth, which he sketches from roughly $1B to $3B and onward to about $5B, with a further step up expected next year. He concludes the deal pays off even if Modal's gross margins never reach the 40s, because the growth carries the return.

  “Modal was, you know, by all intents, it was cheap. ... Even if the margins never hit the 40s, we make money on this deal.”at 36:19

CoreWeave

Across the shows: 34 speakers in other episodes, 8 bullish and 27 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersUncertain

  Neoclouds are inherently risky, but CoreWeave should recover if end-user AI demand holds up.

  Rory says CoreWeave fell 15-20% last week, and bond prices rose, which implied higher default risk. Neoclouds are always riskier because they underwrite high growth with large capex and lots of debt, so any slowdown in AI growth hits them hard. He argues that if end-user demand is there, CoreWeave will true up, since its main customers are Microsoft and OpenAI. The open question is whether demand can keep growing fast enough to fill rising capex plans, which he ties to the combined growth of OpenAI and Anthropic.

  “If the end user demand is there, then everything is going to work out, including CoreWeave.”at 14:44

• Jason Lemkin · SaaStrBearish

  CoreWeave is a B-tier neocloud with no unique moat and heavy debt, so it is shock-sensitive and not worth holding.

  Jason compares CoreWeave with Nvidia and calls it B-tier. He argues CoreWeave exists only because Nvidia and its partners need it. It produces no singular silicon, GPU or software that no one else can make, and it rents out excess capacity while buying GPUs. With massive debt loads and risk, he expects it to be more sensitive to shocks than A-tier assets. He says B-tier assets should not command a premium, and he would not hold CoreWeave long term.

  “CoreWeave exists because Nvidia and Friends need it to exist for now. It has no inherent reason to exist.”at 16:01

Anthropic

Across the shows: 37 speakers in other episodes, 22 bullish and 41 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersBullish

  Anthropic's IPO valuation looks reasonable: 12-13x forward revenue on linearly slowing growth is not crazy.

  Rory says Anthropic's growth is clearly slowing from an unsustainably high 10x pace, and it is highly unlikely to 10x again. If revenue adds about $5B per quarter linearly, from $16.8B in Q3 to about $20-21B by Q4 and then $105-110B next year, a valuation of 12-13x forward revenue for the most exciting company in the world does not feel crazy. It does not take much squinting to make that work, so the IPO price looks plausible.

  “And now you're saying it's 12, 13 times forward revenues for the most exciting company in the world. It doesn't feel crazy. It doesn't take a lot of squinting to make that work.”at 20:13

• Jason Lemkin · SaaStrBearish

  Anthropic's market share will likely fall about 10% by end of next year, though still strong in absolute terms.

  Jason says developers are not loyal, and market share has shifted quickly among Anthropic, OpenAI, open-weight models and Kimi since the show began. Replit's switch of its default model to OpenAI, because it is cheaper at roughly peer quality, shows the pattern. He expects Anthropic's share to fall about 10% by the end of next year, while revenue stays large in absolute terms. Investors building top-line models should allow for materially worse outcomes than the best case, and expect choppy trading after the IPO.

  “It wouldn't surprise me if at the end of next year, [Anthropic's] market share is down 10% from where it is today. And it still could be great on an absolute basis.”at 21:21

Anduril Industries

• Jason Lemkin · SaaStrBearish

  Anduril at $90B, about 40x revenue, looks too rich given its margins.

  Lemkin calls Anduril a generational company that got in early on next-generation defense, with deep contracts. He still questions whether a roughly 40x revenue multiple, given its margins, is justified. He ties this to a broader worry that private rounds have drifted too far from public comparables.

  “[Anduril] is clearly a generational company. Right? Got in early on next generation defense, clearly deep, has the contracts, has everything. ... But is this thing worth 40 times revenue with the margins and everything it has?”at 38:35

• Rory O'Driscoll · Scale Venture PartnersUncertain

  Anduril's valuation only works if its growth continues for years; not impossible.

  O'Driscoll compares a mature defense contractor with $2B of revenue, worth about $8B at best, with a $2B-revenue firm that has doubled for a couple of years and has a line to the Pentagon. For that second firm, a path of 2, 4, 8, 20 could justify the valuation. He says this is not impossible but leans heavily on the future.

  “And the only thing that kind of makes that plausible is the growth rate. ... then maybe you go two, four, eight, 20, and you justify the valuation. So it's not impossible.”at 39:59

Oracle

Across the shows: 14 speakers in other episodes, 0 bullish and 13 bearish calls, past 30 days

• Jason Lemkin · SaaStrBearish

  Oracle is a B-tier asset with massive debt, weak enterprise software sales, and reliance on renting capacity.

  Jason puts Oracle in the B tier alongside CoreWeave. Oracle carries massive debt loads and risk and is highly sensitive to shocks. He argues Oracle cannot sell enough enterprise software, because customers do not want to install enough of its products on their servers. As a result, it is stuck renting out excess capacity, buying GPUs and doing Stargate, which he sees as B-tier activity.

  “Oracle just can't sell enough enterprise software. People don't want to install enough of this crap on their servers. So they're stuck renting out their excess capacity and buying GPUs and doing Stargate.”at 16:22

OpenAI

Across the shows: 19 speakers in other episodes, 7 bullish and 17 bearish calls, past 30 days

• Jason Lemkin · SaaStrBullish

  OpenAI models gain share because they deliver similar quality at 30-40% lower cost, as Replit's routing shows.

  Jason explains that Replit made OpenAI its default model because OpenAI's models are cheaper for roughly the same output quality. Replit's own routing layer makes this switch. Teams like Replit, Lovable and Base44 have a strong incentive to get the quality-price balance right. Their routing shows that quality is roughly peer level and the price is 30-40% better. Jason notes this could change, since Replit's choices may shift again.

  “the quality is sufficiently peer level here. And the price is 30 to 40% better for us today.”at 24:04

Fireworks AI

• Jason Lemkin · SaaStrBullish

  Fireworks' margins have risen to the mid-30s and should reach about 50% soon.

  Lemkin says Fireworks' margins have crept up this year, and are now in the mid-30s, with a path to around 50 pretty soon. He argues that even mid-30s margins are tolerable for public markets if they hold, and that the multiple matters more than the margin level.

  “Fireworks are mid thirties now, and they'll be, you know, hitting 50 pretty soon. Even mid thirties, if it stays there, I mean, the public markets can tolerate that.”at 35:07

Town

• Jason Lemkin · SaaStrNeutral

  Town at $1B is a bet he would pass on; not enough data to get excited.

  Lemkin says Town has no data and is a classic series A/B bet with a bow tie at a $1B valuation, so it is not compelling on its own. He notes the team expects only about $500M of inference spend over twelve months, far less than a peer's $5B, which limits downside. He still says he would pass and that there is not enough here to get excited either way.

  “This is just a bet. This is just an old classic series A series B with a bow tie at a billion. ... But there's just not enough here for me to get excited either way.”at 37:02

Plaid

• Jason Lemkin · SaaStrBearish

  Plaid is a good company, but he doesn't see a $25B exit, so he would not take it to IC.

  Lemkin calls Plaid a great company, but says his firm would not even do the deal today. He does not see a $25B exit for it, so he would not bring it to the investment committee. He says it is too early to predict where it will go.

  “Bunch of guys from Plaid... it's a great one, but we wouldn't even do Plaid today. Don't even see a $25,000,000,000 exit for Plaid, so I wouldn't even bring it to the IC.”at 37:12

Bending Spoons S.p.A.

Across the shows: 2 speakers in other episodes, 1 bullish and 0 bearish calls, past 30 days

• Jason Lemkin · SaaStrBullish

  Bending Spoons is Berkshire's software model: buy slow-growth assets and hold them for a decade.

  Lemkin compares Bending Spoons to Berkshire, which took a stake after Berkshire bought $330M of new shares. Bending Spoons buys challenged software assets such as Vimeo, Evernote, Eventbrite and Airtable, raises prices to make them profitable, and holds them long term. He expects these assets to endure for a decade or more, as See's Candies has done, and argues the company's premium valuation is earned.

  “I think Bending Spoons is like the Berkshire version of software, right? It's like [See's] candy or Coke.”at 42:45

General Atlantic

• Rory O'Driscoll · Scale Venture PartnersUncertain

  Only the very largest firms like General Atlantic can go public, and IPO would change its culture.

  O'Driscoll says public investors value management fee income, not carry, and that a firm needs roughly $800M of fee income, plus or minus $1B, to reach a market cap that matters. Listed alternative managers such as KKR, Carlyle, Apollo and TPG trade at 4-5x revenue at best. He concludes that only the very largest firms will do it. He also argues an IPO would make a venture firm more risk averse and turn it into a different, corporate-style business.

  “The binding constraint is the fee income... you're probably looking at having to have 800,000,000 in fee income... So the very largest companies will do it, but not many is the answer.”at 49:37

Tesla

Across the shows: 12 speakers in other episodes, 12 bullish and 4 bearish calls, past 30 days

• Jason Lemkin · SaaStrNeutral

  Tesla will merge into SpaceX within six months; this is its last standalone event.

  Lemkin reads Tesla's recent stunts as the last standalone promotion before a merger with SpaceX that he expects within six months. He says keeping the companies split makes no sense for Musk, who wants one company. He credits Musk's personal premium for supporting Tesla's high valuation, and sees the stunts as sustaining that for another year.

  “I think it's the last standalone big hurrah before it becomes one big company over the next six months, which was always Elon's goal. Having this company split up makes no sense for him.”at 54:42

Venture capital & startups

• Jason Lemkin · SaaStr

  Large funds need billion-dollar positions, so seed investors at high entry prices struggle to make money.

  Jason argues that today's multi-billion dollar funds need billion-dollar positions in nearly every company to produce returns. Mid-sized outcomes like a $2B company do not suffice, since reaching a billion-dollar position requires major ownership that founders rarely give up. He adds that seed funds entering at $50-100M valuations cannot make money, and that top-decile companies used to earn an automatic term sheet but no longer do.

  “But you gotta have billion dollar positions in everything in Big Boy Venture today. Its billion dollar positions are nothing. There's no point in investing otherwise with large funds.”at 0:00

• Harry Stebbings · 20VC

  Venture capital is concentrating in a few names, leaving most companies unfundable, worse than ever before.

  Harry says venture is broken because he keeps meeting companies that fit a profile of strong growth and profitability but cannot raise money. He cites a $20M ARR company growing 600-700% and profitable, yet VCs are not interested. He ties this to capital concentrating in fewer and fewer names, leaving everything else wholly unfundable, which he has not seen more of before in venture. Jason and Rory agree that strong-looking companies in the mid-tier struggle to raise.

  “I'm seeing that more and more with this concentration of capital going to fewer and fewer names and everything else being wholly unfundable, which more so than I've seen ever in venture before.”at 5:47

• Rory O'Driscoll · Scale Venture Partners

  Late-stage investors now capture the value, so early-stage returns are inverted versus growth-stage returns.

  Rory argues that in a rational capital world, early-stage venture should earn a higher return multiple than growth investing, or no one would do it. He thinks the last five years are clearly inverted. He cites the combined market cap of SpaceX, Anthropic and OpenAI rising from about $4 trillion to about $5 trillion over three years. He says that created over $4 trillion of value in late-stage businesses, and if the next year brings another run of similar growth, late-stage bets alone are what matter.

  “the return multiple from early stage investing should be higher than growth, just logically over an extended period of time... I think the last five years are gonna be clearly inverted.”at 12:28

• Jason Lemkin · SaaStr

  Private revenue multiples are disconnected from public comps and will likely correct.

  Lemkin points to public SaaS comps such as HubSpot, Monday and Klaviyo trading near 2x revenue, and Braze, once the best exit in its category at 6x, now near 3x. He argues that the rounds being done, for example Anduril at about 40x revenue, are disconnected from where public markets would value them. He calls that disconnect almost certainly true, and says a public-market correction, or lower exit multiples, would be needed to reconcile them.

  “If the multiples are disconnected from where they will land, the multiples, and that's just venture... I think that's almost certainly true. The multiples are disconnected.”at 39:39

• Harry Stebbings · 20VC

  Venture is about to heat up as SpaceX cash flows back to LPs and recycles into startups.

  Stebbings says LPs are receiving cash now from SpaceX and are seeing Anthropic on the horizon, with OpenAI hoped for in Q1 or Q2. He expects that cash to recycle back into the venture ecosystem, with LPs committing more capital. The result, in his view, is a hotter venture market.

  “And I think the venture market's about to get a lot hotter, by the way, because a lot of people are starting to get cash back now from SpaceX.”at 53:04

Enterprise AI adoption & AI software

• Rory O'Driscoll · Scale Venture Partners

  AI apps closest to foundation models grow fastest but face constant risk of being overtaken by model labs.

  Rory says the standalone AI apps with the highest growth rates have been closest to the foundation models, such as coding, code review and legal. Each use case is marquee and superb for the technology, but every six months brings a terror moment when the labs might move into it. He argues you cannot want infinitely fast growth while being undifferentiated from the foundation models, because the growth exists only since those models unlocked new capabilities. Investors have to accept this risk.

  “They're all very close to the foundation models. And every six months, you have the terror moment of Oh my god, they're going to do it.”at 28:45

• Jason Lemkin · SaaStr

  Code review is the most important software category today, since AI-written code is never production-ready.

  Jason argues code review is the most important category in software by far, even if it does not produce the largest fortunes. LLMs generate code quickly and it is good, but it is never production-intent. Reviewing and making code production-ready still takes an order of magnitude more time than building it. He says the biggest category is build-to-prod, and the money will come from getting AI code into production, not from building it.

  “I actually think code review is the single most important category today in software. By far the most important category.”at 29:31

• Rory O'Driscoll · Scale Venture Partners

  Coding is the mother lode of LLM value, creating ancillary opportunities in testing, memory and sandboxes.

  Rory argues coding is where most LLM value is created. Because the software development life cycle is changing, with far more software being made, the process of deploying it successfully will look different in an agentic world. That creates a range of related opportunities beyond code review, including end-to-end memory management, testing and sandboxes, where he and Harry have an investment. He concedes that some of these opportunities may be too big for the investors on the show.

  “I said it last week, I said, again, coding is the mother lode of LLM value, right? It's where it all happens. And therefore, there's going to be a whole bunch of ancillary opportunities around that.”at 31:21

AI pricing & unit economics

• Jason Lemkin · SaaStr

  AI product buyers are highly cost-sensitive, pushing routing toward cheaper models over top-tier intelligence.

  Jason says Replit users are cost sensitive, and enterprise customers are too, because they do not want massive variable bills from product teams building prototypes. Replit has a model selector, and the routing choice shows that customers push to cheaper models when quality is similar. He argues this makes the market a super cost-sensitive model. Replit, with its large user base, is a useful lab for watching how quality and cost get optimized.

  “But the enterprise customers are very cost sensitive because they don't want these massive variable bills for the product teams going off and building prototypes. It's a super cost sensitive model, right?”at 25:21

• Rory O'Driscoll · Scale Venture Partners

  Model labs earn about 70% gross margins while inference hosts earn about 30%, shaping where developers spend.

  Rory argues that if a team like Replit can buy tokens on Modal at a 30% gross margin rather than from Claude Code at a 70% margin, and the tokens are roughly the same, it will buy on Modal. He sees the inference market as expanding, but the big labs still take more than 95% of the revenue, since OpenAI and Anthropic together exceed $100B while open-weight inference providers are below 5%. He warns that if demand slows, price competition could get tougher, since everyone wants to hold market share.

  “Do I want to give money to [Claude] Code and give them a 70% gross margin for this token? Or do I wanna buy it on Modal and give them a 30% gross margin for this token?”at 33:40

AI infrastructure & data centers

• Rory O'Driscoll · Scale Venture Partners

  Neoclouds will face more competition and may not justify their valuations, though growth covers flaws.

  O'Driscoll says neoclouds do the building, buy compute, borrow capital, and then resellers buy from them and sell to developers at a large margin. That invites a lot more competition, which makes him doubt the valuations. He adds that very high growth can cover many problems, so the valuation question depends on how fast revenue grows.

  “All the Neo Clouds are saying... We do all the building work. We buy all the compute. We borrow all the capital... you can imagine a lot more competition here, right? ... On the other hand, when you have this kind of growth rate, nothing else matters.”at 34:39

US stock market outlook

• Rory O'Driscoll · Scale Venture Partners

  Big multiple expansion from current US market levels is unlikely to save valuations.

  O'Driscoll argues that the US market already trades at high multiples, so a large re-rating from here is unlikely to rescue overpriced private or public assets. He thinks the risk is that companies priced for a future that may not arrive will see multiples compress. He agrees with Lemkin that repricing is likely as large sums must be moved.

  “It's highly unlikely that the US market... you're not gonna get huge multiple expansion from here that saves you, I would guess.”at 40:30

Software & SaaS

• Jason Lemkin · SaaStr

  AI-era marketing software is being punished too hard; its multiples may be too low.

  Lemkin notes HubSpot, Klaviyo and similar marketing software companies trade below 3x revenue, which he calls the worst public category. He argues they are not dying, and that agentic AI players trading near these levels could also be underpriced if their growth slows to 60-70%. He is uncertain whether the AI versions will settle too early into their peers' multiples.

  “I do wonder if the AI versions of low multiple companies will settle too early into the multiples of their peers. ... I think the whole marketing category is the worst public category without question. ... They're not dying.”at 41:45

Autos & EVs

• Jason Lemkin · SaaStr

  China will dominate global EVs and autonomy; no one can match its cost position.

  Lemkin cites Elon Musk's view that China will take over car making and that the last car a human will ever drive is near. He says China's cost position in electric and other vehicles is so strong that no one can compete. He adds that in a few years almost no one will drive their own car.

  “China is going to dominate global EV and autonomy. No one can compete with their cost proposition in electric and otherwise.”at 54:38