20VC: OpenAI Cuts Valuation to $1.25TRN | Public Market Meltdown: GOOG Loses $500BN in 12 Hours | DoorDash Buys Instacart for $17BN | Kalshi Raises $1BN and Cognition Eyes $80BN Valuation

Cognition AI

• Jason Lemkin · SaaStrBullish

  Cognition's ~$80B valuation looks market-correct; coding agents are the AI category still working.

  Jason says Cognition at about $80B seems market-correct. He worries it owns limited IP and lucked into the Windsurf team and open-weight models. Still, coding agents are the one thing that keeps working in the latest Anthropic quarter, and Cognition has much of that field open. He says the momentum is too strong to sit out, so he wants to put the fund's last ~$2B into it, noting the fund is already up about 5x on paper.

  “I do get worried that fundamentally it only owns so much of its own IP, but geez, I don't know if you saw the last Q3 of Anthropic. Like the one thing that is working that never stops working is coding agents. And Cognition has a bunch of the field open to it.”at 42:06

• Harry Stebbings · 20VCBearish

  At ~$80B, Cognition's 10x chance is limited; the multiple is clearly better at Serval.

  Harry says the chance of a 10x at Cognition is much lower than at Serval, because Cognition is already near a trillion-dollar-equivalent outcome after dilution. He sees the multiple as clearly more evident at Serval. He also flags that Cognition is competing in a crowded field of coding agents, positioned fourth in the market, which limits how much of the market it can take.

  “The chance to get a 10x at Cognition, to Jason's point, gosh, we're nearly hitting a trillion there. So, yeah, I think the multiple is clearly much more evident there.”at 45:09

• Rory O'Driscoll · Scale Venture PartnersBullish

  Cognition can command far more dollars per enterprise than Serval; coding agents offer the fastest revenue velocity.

  Rory argues the coding-agent model is that Cognition goes into large enterprises and offers to rewrite their codebases and enable agentic coding, which commands huge contracts. He says that on money velocity available right now, Cognition can command many more dollars than Serval's slow, multi-year ServiceNow displacement. He sees Serval as a more compounding bet but capped, and frames Cognition as the one with near-term revenue velocity.

  “they're coming into large enterprises and basically saying, we will help you effectively rewrite your entire code base... So in terms of money velocity on the table right now, cognition has just an ability to command many more dollars.”at 47:28

Serval

• Harry Stebbings · 20VCBullish

  Serval has the best enterprise adoption he's seen and far more multiple upside than Cognition.

  Harry praises the Serval team and product and says enterprise adoption is better than he has ever seen. He argues the upside is a pure multiple on cash: at a $3B valuation with a 10x outcome very likely, the multiple is better than at $30B Base10 (where he says you have to squint) or at Cognition near $80B. His caveat is that Serval raised only about $150M, so a large fund check would be hard to deploy there.

  “Jake and Tatiana at [Serval] are incredible. The product's amazing. The enterprise adoption is better than I've ever seen.”at 44:48

• Jason Lemkin · SaaStrBearish

  Serval may only take a slice of ServiceNow's TAM; he's not excited to invest.

  Jason asks whether Serval expands the total addressable market or stays bounded by a subset of ServiceNow's TAM. He says every founder draws concentric circles, but for this deal he doesn't think it's expanding the TAM. Disrupting incumbents without expanding the market, he says, is not an attractive bet, so he is personally not excited.

  “a subset of ServiceNow's TAM today, right? Now every founder will expand the markets and add concentric circles, but is this really expanding the TAM here?”at 45:59

• Rory O'Driscoll · Scale Venture PartnersBullish

  Serval can expand ITSM's TAM by doing the work itself, so customers pay more than for ServiceNow.

  Rory's bull case is that ServiceNow is a list of IT tasks, and Serval performs those tasks, so it can command more value per customer. The test is like-for-like spend: a mid-market company that spends about $200k on ServiceNow would need to spend $300-400k on Serval to show TAM expansion. He argues ITSM is a much bigger market than it looks, since ServiceNow became the general to-do list for back-office functions. He sees Serval as a compounding bet against a decade-plus of ServiceNow displacement, and puts the chance it becomes the next-generation ServiceNow at 70-80% or more. Its money velocity is slower than Cognition's.

  “And [Serval] is actually a product that does those things for you. So you can command more value from it.”at 46:10

OpenAI

Across the shows: 46 speakers in other episodes, 29 bullish and 38 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersNeutral

  OpenAI's valuation cut reflects relative liquidity, growth and profitability versus public peers, not an AI wave cracking.

  Rory rejects the 'AI wave cracking' narrative. He compares OpenAI with a bigger, faster-growing, profitable company about to IPO that can be bought liquid at $1.5–2T. A smaller, slower-growing, loss-making, illiquid company would be valued below that. The lower OpenAI round is therefore a relative pricing outcome, not a verdict on AI demand.

  “I would dispense with the whole AI wave cracking narrative... It's as simple as the other company is bigger, growing faster, profitable, and about to go public, so you can buy it liquid at 1.5 to 2,000,000,000,000.”at 4:53

• Jason Lemkin · SaaStrBearish

  OpenAI is discounting near-equivalent models about 30%, likely on a higher cost basis, which signals weaker economics.

  Jason says OpenAI is selling somewhat equivalent products 30% cheaper. Other players, including Chinese open-weight models, are also adding price pressure. He suspects OpenAI's underlying cost basis is higher, so its numbers would look worse if visible. The discount appears to be working, which he says is significant because it approaches the roughly 40% discount he targets for open-weight models.

  “OpenAI is basically running around selling somewhat equivalent products for 30% off.”at 10:39

Alphabet

Across the shows: 61 speakers in other episodes, 56 bullish and 40 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersUncertain

  Alphabet's sell-off reflects search and capex angst; search cannibalization is a real but slow risk.

  Rory says search is probably facing some cannibalization, a risk that has been true for two or three years, yet search still grew about 12%. He attributes the selloff to a mix of search angst, general market angst and capex angst, and calls it nothing more than that. Google Cloud growth is real, with customers able to buy compute as fast as it is built. He still treats search cannibalization as a live long-term worry.

  “On the search side, it's been true for two or three years that there's probably gonna be some element of cannibalization of search. And, it still grew pretty nice. I think it grew 12% or so on search. But I think it was a combination of search angst and general angst”at 8:34

• Jason Lemkin · SaaStrBullish

  Google Cloud's roughly $120B run rate growing near 100% is best-in-class; the selloff is nervousness.

  Jason points to Google Cloud at about a $120B run rate growing nearly 100%, which he calls about as good as it gets in cloud. The stock fell because the market is nervous, and any metric looks short of expectations. Shopify fell 15% despite raising guidance, which he cites as evidence of that mood.

  “think it's at a $120,000,000,000 run rate growing to 100%. That's about as good as it gets for the folks in the cloud. The cloud group should feel pretty proud of themselves.”at 9:33

Anthropic

Across the shows: 102 speakers in other episodes, 128 bullish and 85 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersUncertain

  Anthropic's revenue growth decelerated sharply but remains very high; the slowdown is a matter of framing.

  Using GAAP figures, Rory estimates quarterly revenue of roughly $5B in Q1, $11B in Q2 and $17B in Q3. Sequential growth fell from about 100% to about 50%, which still compounds to roughly 5x a year. Full-year revenue was $4.5B in 2025 with a $9B exit run rate, and he expects about $56B in 2026 and about $120B in 2027, which is 12x growth falling to about 2x. Bookings growth also decelerated, though absolute bookings rose about $5B from Q2 to Q3. Both the bull and bear readings are equally true, so he stays neutral on the outcome.

  “So the growth rate naturally, the Q on Q growth rate has plummeted to a roughly 50% Q on Q growth rate... If they stick at 50%, that's still 5x year on year growth rate, that's the positive spin... Those are equally true statements of the same facts.”at 12:14

• Jason Lemkin · SaaStrBearish

  Anthropic's pricing power may have peaked as cheaper rivals and implicit discounts pressure its premium.

  Anthropic's net new bookings dipped in the latest quarter. Jason links this to pricing pressure: OpenAI is discounting near-equivalent products, Chinese open-weight models are cheaper, and Anthropic has added implicit discounts through Haiku 5.5. Demand does not appear to have fallen, since OpenRouter usage is steady, so the problem is that customers can pay less. He presents the thesis that Anthropic's ability to command a premium has hit a local maximum as one open possibility, not a certainty.

  “maybe we've reached the point where anthropic ability to command a price premium has hit a local maximum. That's one potential thesis. It is open.”at 14:40

DoorDash

Across the shows: 3 speakers in other episodes, 1 bullish and 0 bearish calls, past 30 days

• Jason Lemkin · SaaStrBullish

  DoorDash's Instacart deal makes sense: combined they would hold about 80% share, and such deals appear accretive.

  Jason says Instacart has about 20% of the market and DoorDash about 60% of delivery, so combining them could reach roughly 80% share, which is how networks are dominated. He notes that past deals like the Postmates acquisition appear accretive when well architected, contrary to his expectation. Antitrust is the main hurdle, and he thinks it is a real question given the combined share.

  “if Instacart's got 20% of the market and DoorDash has 60 of the delivery market and you can get through antitrust... And you can get to 80% market share. This is how you dominate networks.”at 28:02

• Jason Lemkin · SaaStrBearish

  AI agents are an existential risk to DoorDash because its number three competitor is nobody.

  Jason argues all app platforms face a slight disruption risk from agents that can order food or find cheaper delivery directly. He calls DoorDash's claim that agents placed only several hundred orders against 3 million daily deliveries self-serving. Agents could bypass DoorDash by ordering from restaurants that self-deliver or offer cheaper pickup. Because DoorDash's third-place rival is nobody, he sees the risk as existential.

  “I think agents are an existential risk to DoorDash because their number three competitor is nobody.”at 31:11

Cohere

• Rory O'Driscoll · Scale Venture PartnersBullish

  Cohere is a sovereign AI play and one of the few surviving frontier LLM companies, deserving credit.

  Rory says Cohere's raise is less a momentum play than a sovereign AI play for Canada. Cohere is one of the original four or five LLM developers from 2023, and most others, such as those acquired by Microsoft and Amazon, have folded. He gives it huge credit for surviving among the frontier-model companies.

  “It's almost not a momentum play... we need our sovereign AI play in Canada... give Cohere a huge credit. They are one of the survivors.”at 33:51

• Rory O'Driscoll · Scale Venture PartnersBearish

  Cohere's government-backed sovereign AI story can't force private-sector demand; a 3x IPO looks doubtful.

  Rory accepts that government money, as with Canada's support for Cohere, can keep a model company alive for a while, since governments can subsidise regardless of economics. He argues that governments cannot make private companies use the product, and says Cohere's non-disclosure of revenue doesn't help. He is not convinced it becomes a compelling business that could go public at 3x its current round valuation. He compares it to Mistral, which faces the same demand limit.

  “Governments can make things happen for a fair while, but what they can't do is make the private sector use the product. So you see it with Mistral too.”at 36:02

Binance

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

• Jason Lemkin · SaaStrBearish

  Binance barely survived the $6.6B loss; its opaque balance sheet leaves it one step from an FTX-style collapse.

  Jason argues Binance barely survived the roughly $6.6B crypto theft, which was attributed to North Korea. Binance has no transparent balance sheet, so no one knows how much cash it had, and he asks what would have happened if the loss had been $13B. He thinks the company was one step away from an FTX-type situation. He also says many firms claimed their balance sheet was customer crypto held as custodian, which is not true.

  “I think Binance barely survived this 6,600,000,000. They don't have a transparent balance sheet. We don't know how much money they had. What if it was 13,000,000,000?”at 37:25

• Rory O'Driscoll · Scale Venture PartnersBullish

  Binance covering the $6.6B crypto loss shows an excellent business, if it wasn't customer money.

  Rory argues that Binance paying back the roughly $6B from its own resources implies it generates enough profit to cover such a loss, which makes it a very strong business. He hedges that this holds only if it isn't an FTX-style case of taking money from someone else. He says he is not a crypto specialist and hasn't studied it in detail, and disagrees with Jason's view that the firm couldn't have covered a larger loss.

  “I mean, assuming that you're not doing an FTX and swiping it from someone else. ... Binance were able to make good the crypto loss, which implicitly means Binance makes enough money to be able to make good the crypto loss, which means it's an amazing freaking business.”at 38:16

Crosby

• Jason Lemkin · SaaStrBearish

  Crosby's human-plus-AI law model looks unstable; top lawyers won't trade profits for startup equity.

  Jason argues the compensation math doesn't work for Crosby: a lawyer earning about $170k plus 0.01% of an IPO that may never happen, versus about $250k at Weil or up to $10M for a top rainmaker. Rainmakers will want to keep their money, so he doubts the business is stable. He hasn't spoken with Crosby, and he is not convinced it will be as stable as VCs hope.

  “My only point is, and I'm not being critical, I haven't talked to them, I'm not convinced this is as stable as a lot of VCs might hope it would be.”at 55:09

• Rory O'Driscoll · Scale Venture PartnersUncertain

  Crosby will likely thrive in transactional law, but a $10B outcome is doubtful.

  Rory thinks routine transactional work, like NDAs and Series B purchase agreements, doesn't need expensive rainmakers, so an AI-enabled firm can serve that niche profitably. He expects a coterie of such firms in five years, with Crosby among them. He is doubtful about a $10B outcome, since it would have to pull billions of partner profits away and rainmaker work is hard for AI to replace. He says the low-margin remainder limits how big it can get, but a small business is clearly possible.

  “I think that there will be a coterie of law firms five years from now that are focused on the Crosby kind of work. Crosby will obviously be one of them. They'll do really well.”at 56:24

CoreWeave

Across the shows: 81 speakers in other episodes, 23 bullish and 87 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersUncertain

  If AI demand disappoints, neoclouds like CoreWeave will suffer first, making their debt risky.

  Rory argues that if AI demand stays strong, every GPU and data center will be needed, and CoreWeave's bonds will be repaid because a larger buyer would step in. If demand falls short, the first casualties will be smaller, highly levered operators such as CoreWeave rather than Google or Amazon with large balance sheets. That is why its bonds carry real risk, and it makes CoreWeave the test case for demand.

  “If the demand is not there, then the first person is gonna suffer when the demand is not there. It's not gonna be Google or Amazon, whoever, big ass balance sheet. It's gonna be the guys like Core Weave, and that's where the problem will come.”at 7:55

Canva Pty

• Jason Lemkin · SaaStrBearish

  Canva's growth has decelerated about 20% and agents do not use it, a sign of AI disruption to SaaS.

  Jason says Canva was still accelerating at the start of the year, but recent reports suggest growth has decelerated about 20%, with some reports even saying growth was zero and layoffs were being prepared. He turned off Canva himself because his agents do not log into it. He is surprised that core consumer and SaaS businesses like Squarespace, which announced 20% layoffs, and Wix are struggling, given AI tools that the real world isn't buying much of.

  “I just think at the start of this year, Canva was still accelerating... It at least seems to have decelerated 20%.”at 18:42

Anysphere Inc. (Cursor)

Across the shows: 3 speakers in other episodes, 3 bullish and 0 bearish calls, past 30 days

• Jason Lemkin · SaaStrBullish

  Cursor is a win at a $60B valuation despite thin margins, because great teams figure out the economics.

  Jason says Cursor barely made it, and that its V2 product would have died if it had only resold Anthropic tokens at a loss. He still regards a $60B outcome as a win, arguing that great teams find a way through margin pressure. He says he would be comfortable investing at a $50–100B valuation despite the business-model risk.

  “Cursor barely pulled it out. The Cursor V2 would have died reselling Anthropic tokens. ... I'd take the 60,000,000,000 and call it win myself. ... Great teams figure it out.”at 23:43

Uber Technologies

Across the shows: 4 speakers in other episodes, 0 bullish and 4 bearish calls, past 30 days

• Harry Stebbings · 20VCBullish

  Uber is a strong buy because it is cheap, despite agentic disruption worries.

  Harry says he agrees with Bill Ackman that Uber is a strong buy because it is very cheap. The conversation argues that agentic disruption is unlikely in ride-hailing, since booking a car through a chat assistant is a worse user experience than opening the app. Harry frames the valuation as the attraction, and the discussion does not dispute that the stock is cheap.

  “I'm with Bill Ackman. I think Uber's a banger of a buy right now, though. I mean, it's so cheap.”at 31:54

Kalshi

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

• Rory O'Driscoll · Scale Venture PartnersBullish

  Kalshi is a fast-growing money-maker and a classic momentum play, despite state regulatory fights over sports betting.

  Rory says the business is working and growing fast, though he is cynical about its origins. It began as a prediction market with social utility but has shifted largely to sports betting. It is pushing through local regulation on a national basis, which is why Utah is suing. VCs back what is working, and its valuation rose to about $40B from $22B only months earlier, so this is a momentum play.

  “Kaltry is a money deal. Let's be really clear here. The thing is working. It's growing like a weed.”at 33:26

Weil, Gotshal & Manges LLP

• Rory O'Driscoll · Scale Venture PartnersBearish

  Weil's collapse reflects losing key rainmakers, not technology, and such firms can unravel quickly.

  Rory reads the accounts and says Weil depends on four or five key lawyers, each running a key part of the business. A clash over a couple of them, and the exit of the head of its PE practice (perhaps 20-30% of the firm), can make things go wrong quickly. He calls it an American law story, not a tech story. He links it to AI: if about 70% of billings come from rainmaker work that AI can't replace, the rest is low-margin.

  “As I said, it's four or five key lawyers, each of whom run a key part of the business... I think the guy who ran all the PE practice, which is probably twenty-thirty percent of a company like this, left... And, yeah, it can go wrong pretty quickly.”at 52:47

Physical Intelligence

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

• Rory O'Driscoll · Scale Venture PartnersUncertain

  Robot model tech is impressive, but a $20B valuation needs far more robots deployed than exist today.

  Rory finds the dexterity of new robot foundation models a real breakthrough. He says robotics deployment takes much longer than LLM rollout, and that the business model still needs development. He notes only about 300,000 robots operate in the US and that pricing around $3 an hour per robot means a roughly $20B valuation requires a far larger installed base. He expects the big foundation-model companies to buy or build robotics units, since robotics doubles their addressable market. He says revenue multiples don't mean much for companies where the future drives value.

  “So two other comments though, I think deployment of robotics is just way longer than deployment of classic LLMs, and people forget that.”at 58:25

Salesforce

Across the shows: 5 speakers in other episodes, 5 bullish and 0 bearish calls, past 30 days

• Rory O'Driscoll · Scale Venture PartnersBullish

  Salesforce's roughly 10x-revenue Cresta purchase is a smart deal that adds critical mass.

  Rory thinks the Cresta deal is pretty good for Salesforce because Cresta is a more mature company with scale and enterprise deployments, and its ARR is well over $200M. Paying about 10x revenue is reasonable compared with the 40x-type multiples on hot new AI startups. He says it's a smart deal for Salesforce because it gives critical mass and moves its service business forward.

  “I think the Cresta deal is actually pretty good. You're paying 10 x.”at 1:04:27

AI pricing & unit economics

• Jason Lemkin · SaaStr

  Same-task AI costs fell about 72% after switching models, the first real deflation he has seen.

  Jason moved his agentic B2B workflows, which run constantly, to Haiku 5.5 and a cheaper model. Total costs fell about 70–72% for the same task at equal or better quality. He says that in his experience since 2015 nothing got cheaper on a usage basis until this switch. He still expects some costs to rise as agent usage grows, but he sees mature applications achieving this kind of deflation from better token quality and pricing pressure.

  “On an as used basis, nothing got cheaper. No matter what anybody said on evals and token price, nothing got cheaper until Haiku 5.5 plus [DeepSeek]. Then I saw 72% decline for B2B workflow running constantly.”at 17:24

• Rory O'Driscoll · Scale Venture Partners

  Digital product prices fall 50–60% a year, so unprofitable AI products will improve economics as they scale.

  Rory argues that digital goods historically fall in price 50–60% a year, and that a product people get hooked on will see the math improve over two or three years, as with Facebook's early server bills and YouTube. Venture markets are too aggressive to wait for perfect economics before launching, because a funded rival will launch first and then ride the cost curve down.

  “the bet you take over and over again is digital goods get reduced in price 50, 60% year on year. If people like the shit and they get hooked on it, over two or three years, the math gets better.”at 23:25

• Jason Lemkin · SaaStr

  Always-on AI agents likely cost more to serve than subscriptions charge; subsidies cannot last.

  Jason estimates persistent agent products could cost around $20 or more per user per month to run the way they are meant to, against $40–50 subscriptions. He calls that a poor model if true. He cites Replit's view that free agent subsidies are ending and says Cursor's V2 would have died reselling Anthropic tokens at a loss. He concludes subsidies cannot last, though he says great teams may still win.

  “We've also seen this in AI. The subsidies just can't last. It's too expensive, right?”at 23:43

Cybersecurity

• Jason Lemkin · SaaStr

  Open-source pen-testing tools became attack weapons; LLM agents let breaches run at scale across targets.

  Jason describes a recently built open-source Chinese pen-testing tool, meant to test your own defences and easy to hook up to LLMs. It was easily turned around to attack others, and with LLM agents, attacks can be run at scale against anyone. He says we're all being breached, pointing to Binance and other recent incidents. He compares it to guns: built for good people, now in bad hands.

  “...it became very easy to turn it around and just attack other folks with agents. ... but now you can do it with LMs at scale.”at 39:45

• Rory O'Driscoll · Scale Venture Partners

  Cyber risk is underrated; a weird systemic breach, like the Fed wire going down, is likely eventually.

  Rory says cyber dynamics over the next twelve months are hard to overstate. The story gets little daily coverage, but he expects something weird to happen at some point, such as the entire US Fed wire going down for a day. He cites the European refrigerator hack as an example of an incident that could hit mission-critical systems like traffic lights or heating facilities.

  “You never cover it on a daily basis, but you are. At some point, something really weird is going to happen.”at 40:39

Venture capital & startups

• Rory O'Driscoll · Scale Venture Partners

  Revenue multiples do not explain private valuations: Physical Intelligence is worth $20B on little revenue.

  Rory contrasts Physical Intelligence, a company with relatively little revenue valued at about $20B, with Dropbox, a company with billions in revenue that he says is not worth much. He concludes that revenue multiples do not work as a valuation guide, because the market prices growth and perceived future position rather than current revenue.

  “Physical intelligence, company with relatively little revenue is worth 20,000,000,000. Dropbox, company with billions of revenue ain't worth a lot there. It just shows revenue multiples don't get you anywhere.”at 0:06

AI: overall outlook

• Rory O'Driscoll · Scale Venture Partners

  Recent AI sell-offs are price discovery; if demand holds, these concerns will be forgotten within six months.

  Rory argues the recent AI-linked selloff and debt-market wobbles reflect price discovery on the supply side, such as what lenders need for CoreWeave or what investors pay for OpenAI. The decisive variable is demand. If demand for AI compute holds, as bulls expect, these concerns will be forgotten in six months. If demand falters, the first to suffer will be the smaller, more leveraged players, which would turn these signals into a significant change.

  “if the other side of the equation, the demand side holds up, all these things will be forgotten in six months. And if there's a problem on the demand side, on the growth side, then all these things will be straws in the wind of a very significant change.”at 7:18

Software & SaaS

• Rory O'Driscoll · Scale Venture Partners

  Software companies that refuse to adopt AI will face quiet obsolescence, living off installed bases.

  Rory compares software companies that stay private or stop investing to mainframe companies in the 1990s and early 2000s that lived off twenty-year-old contracts while their assets became terminally valued. Companies can cut costs and milk the installed base, but that is hard for investors to value. Those that do not join the AI train, even late, face a slow decline to obsolescence over the medium term.

  “But I think if you don't do that, if you decide to just not play that game, I think over the medium term, you're just heading to quiet obsolescence.”at 20:05

AI bubble & tech valuations

• Jason Lemkin · SaaStr

  Trillion-dollar valuations for OpenAI and Anthropic make a 10x after dilution hard; he's getting nervous.

  Jason says getting a 10x after dilution from a trillion-dollar valuation is a big ask, and he is starting to get nervous about the latest OpenAI and Anthropic rounds at that level. He reasons that these labs own only so much of their own IP, so the return math is harder at those prices.

  “Now to get to a 10x with dilution, a trillion dollar valuation, I start to get nervous with the latest round from OpenAI and Anthropic.”at 41:57

Enterprise AI adoption & AI software

• Jason Lemkin · SaaStr

  Most vertical AI startups won't expand TAM; they'll fight incumbents for slices and end up as tiers.

  Jason expects 2027 to show that many vertical AI plays are small startups that didn't expand the TAM. They compete for slices of ServiceNow, Salesforce and Atlassian budgets rather than growing the market. Investors assume vertical AI can charge 3-4x more, but he expects competitors to multiply, with leaders still alive but growing more slowly. He says he personally won't invest in these companies because he expects them to become tiers within five years. Rory pushes back that Serval is an exception.

  “I think [Serval] aside, 2027 will be a year where we find out a lot of vertical AI plays are nifty little startups, but didn't in the end expand TAM enough. They didn't expand TAM.”at 48:02

IPOs, M&A & deal activity

• Jason Lemkin · SaaStr

  Venture needs frequent $1B+ acquisitions for liquidity; a failed Hugging Face deal would hurt exits.

  Jason says venture depends on M&A working like clockwork to liquidate portfolios, since IPOs alone leave firms stressed. He wants serial $1B+ acquisitions from Salesforce, ServiceNow and Microsoft. He worries that the roughly $12.9B Hugging Face deal could fail antitrust review, which would force a return to convoluted team and IP deals. He notes the antitrust environment is better than two years ago, which helps deals.

  “We need M and A to work like clockwork. Otherwise, we're just stressed over these goddamn IPOs.”at 1:01:31