The Twenty Minute VC · 8 Oct 2026 · From the week of 5 October
20VC: Cognition vs Factory: Vinod Khosla Creates a Storm | OpenAI Nears $70B Run Rate: Anthropic Under Threat | ElevenLabs Doubles Its Valuation to $22B & Salesforce Buys Listen Labs for $2B
Listen on Apple Podcasts Show on Spotify Episode page
In brief
Harry Stebbings hosts regulars Jason Lemkin and Rory O'Driscoll, with MongoDB CEO Dev Ittycheria as guest, for a weekly tech-news roundtable. Topics include OpenAI's reported re-acceleration against Anthropic, the ethics of a former Snowflake sales leader moving from a Factory board-observer role to CRO at competitor Cognition, Vinod Khosla publicly criticising his own portfolio company, US open-weight models (Reflection's Beam), ElevenLabs at $22B, Salesforce buying Listen Labs for $2B, Vercel getting half its new business from agents, the lawsuit over Nvidia's Groq licensing deal, and Oura pulling its IPO. The most actionable thread for operators is the argument that AI agents are becoming buyers and recommenders of software. Dev and Jason argue vendors, especially startups facing incumbents, must deliberately make themselves visible and usable to agents, and Jason suggests regularly polling live production agents on what they would recommend. Other recurring themes are falling loyalty among developers and executives, enterprise pressure on token costs, and when founders should take an early exit.
For founders
- Dev Ittycheria argues agents are increasingly making buying decisions, and startups are disadvantaged because agents see far more data about incumbents, so founders need documentation, APIs and infrastructure that make them a first-class choice for an agent.
- Jason Lemkin suggests asking around 10 trusted people with agents in production, every two weeks, what their agents would recommend in your category, because agents are already steering purchases (he cites Resend going from 106,000 MCP calls in April to 3 million in September).
- Dev Ittycheria distinguishes AI apps that are 'a feature masking as a company' from durable franchises built on a proprietary data loop, and expects many of the former to sell to platforms with distribution.
- Jason Lemkin argues that if you can exit for around $2B within the first five years and are not building a generational company, taking the deal is reasonable; Rory O'Driscoll frames the decision as how likely you are to reach 10x the offer.
- Dev Ittycheria distinguishes light-touch advisors from advisors 'inside the tent' on product, board and financial plans; Rory O'Driscoll notes the expectations around such advisors were probably left implicit, not made concrete.
For revenue leaders
- Dev Ittycheria says marketers who spent 25 years optimising for page one of Google now have to make sure agents can find them and hold current, accurate information about them, because missing from an AI-generated answer is like being invisible on Google.
- Rory O'Driscoll predicts CFOs will push CIOs to cut token bills by routing non-frontier workloads to cheaper models, chipping away at frontier-lab revenue.
- Jason Lemkin argues most CROs see moving to a competitor as normal career movement, while Dev Ittycheria says he has never seen a CRO go to a direct competitor and warns of long-term reputational damage with the people they recruited.
- Jason Lemkin says ElevenLabs shows quality can beat price in voice AI, with relatively modest companies happily paying hundreds of thousands of dollars a year because the call has to work.
- Dev Ittycheria, citing a recent conversation with MongoDB's CIO, says token cost and budgets plus IP and data rights are important factors every enterprise weighs when choosing AI models and partners.
What was said 50, most useful first
OpenAI's reported Q3 growth would be a sharp re-acceleration after it slowed to roughly 18% quarter-on-quarter in Q2.
Rory O'Driscoll said OpenAI grew 18% quarter-on-quarter from Q1 to Q2, implying just under 100% annual growth. Over the same period Anthropic was growing about 10x, and in Q2 Anthropic's GAAP revenue ($11B) exceeded OpenAI's ($6.8B) for the first time. A 70% (or even 60%) Q3 figure, if accurate, would restore OpenAI's narrative, which he said matters given its forward commitments on compute, data centres and land. He said the open question is whether that growth came at Anthropic's expense.
“the gap wherever you grow from Q1 to Q2 was 18 % quarter on quarter. And you did that math and it was that implied, you know, just under 100 % year on year growth rate, which was massive deceleration Anthropic was still going 10x”
An advisor who joins a competitor matters very differently depending on whether they were a narrow advisor or 'inside the tent' on plans.
Discussing a former Snowflake sales leader who was a board observer at Factory and then became CRO of competitor Cognition, Dev Ittycheria distinguished two kinds of advisor. One is consulted on a specific function, such as vetting a European hire. The other sees product roadmaps, board plans, financials and win/loss data against competitors. A move to a competitor by the first is unwelcome but acceptable to him; by the second it would be 'very, very different'. He disclosed ties to both companies, said he did not know the facts, and said an advisor with confidential access should be clear with the founder when considering a competitive move.
“That's very different than someone who's inside the tent looking at your product plans, your board plans, your financial plans”
Rory O'Driscoll predicts CFOs will force CIOs to cut token bills by routing non-frontier work to cheaper models.
Rory O'Driscoll said enterprises have so far defaulted to frontier models because people are lazy and bills were not astronomical. If OpenAI and Anthropic each reach about $70B in ARR, that $140B becomes a meaningful chunk of US corporate profits. He expects CFOs to push for cheaper models on non-frontier tasks, keeping Anthropic for the hardest problems, and to slowly take 'slugs of revenue' from the frontier labs.
“We need to shave three million off this token bill.”
Frontier models on OpenRouter take only about 20–30% of tokens but about 90% of dollars, so even a modest open-weight share is significant.
Rory O'Driscoll said he would not guess 50% for US open-weight share, but that even 10–20% of tokens would be extremely significant. He reasoned that this traction would come from the more enterprise-centric, safety-conscious customers. He noted it would be risky for inference providers.
“the foundation models on OpenRouter I think get a 20 30 percent of tokens and get 90 percent of the dollars.”
Listen Labs' $2B sale to Salesforce returned about $850M on $96M invested, with Sequoia making about 25x within roughly three years.
Harry Stebbings said Listen Labs reportedly had earlier offers at $1.5B before accepting $2B. Sequoia led the seed and Series A and made about 25x, while Ribbit led the B and made about 4x in eight months. The company was described as about three years old.
“about eight hundred and fifty million dollars back to investors on ninety six invested.”
Listen on Apple Podcasts Fundraising & investors Link to this
Rory O'Driscoll suggests perhaps 10% of today's AI app companies will sell early at around 30x revenue while the rest grind out ~6x outcomes.
Comparing Listen Labs' three-year path with Intercom's long grind to a comparable price, Rory O'Driscoll said venture outcomes are very random and timing and the cycle matter enormously. Intercom could have achieved that valuation had it sold in 2018–19. He floated, as a 'maybe', that in five years a small share of current AI app companies will have sold early at about 30x revenue.
“maybe 10 % of these apps companies will have sold early. They'll have got their 30 X revenue.”
If you can sell for ~$2B in the first five years and aren't building a generational company, you should probably take it.
Jason Lemkin said company-building comes in five-year chunks that 'take it out of you', and exiting early is much easier. He said he wasn't saying it's a reason to sell. He said that this year he sat in a board meeting where every VC told a company with an offer at about this price not to sell. In his view the decision hinges on whether you're building a 'Mongo or better'.
“You can get out for two billion in the first five years and you're not building Mongo or better. I don't know, man. I would take it and enjoy my life at Salesforce.”
BladeLogic turned down acquisition approaches at several stages before selling to BMC for about $900M on roughly $100M revenue.
Dev Ittycheria said the best companies always get offers. BladeLogic was approached at every stage, including a face-to-face meeting with John Chambers in his office about Cisco buying it, and EMC came to the table. It IPO'd in July 2007 and sold to BMC about a year later for roughly $900M, about 9x revenue, which he called the highest acquisition price paid in 2008. He said the founder ultimately has to decide whether they feel good continuing to build, and that not selling earlier was the right call.
“the best companies always got offers to sell”
Companies whose pre-AI infrastructure suits agents, like Vercel, are seeing agents drive a rapidly growing share of new business.
Harry Stebbings cited Vercel at $600M ARR with agents driving 50% of new business, up from 3% at the start of the year. Jason Lemkin said founder Guillermo could not have foreseen this when founding Vercel in 2020, and pointed to Replit (founded 2016) as another example. From memory, he cited Vercel growing about 170% with roughly a 20% increase in customers, implying huge expansion revenue. From his own heavy building he said agents have opinions that are hard to argue with.
“when agents pick you, it's a force of nature right now.”
Resend grew from 106,000 MCP calls in April to 3 million in September as agents recommended it, after Jason Lemkin's agent steered him away from SendGrid.
Jason Lemkin said his first agent-driven purchase was Resend for email. Agents kept recommending SendGrid, but he couldn't get it working: free access had been deprecated, he couldn't figure out the key and it kept breaking. When he asked his agent what to use, it said Resend. He cited the MCP call growth as evidence of agents choosing products.
“Resend April, 106 ,000 MCP calls to September, 3 million. That's agents saying to use products.”
Listen on Apple Podcasts Pipeline & demand generation Link to this
Jason Lemkin tells portfolio companies to poll about 10 trusted people's production agents every two weeks on what they would recommend in their category.
Jason Lemkin advises finding 10 trusted people with agents in production doing different things and asking every two weeks what their agents recommend. He distinguished this from tools that guess what LLMs say. He demonstrated live: asking one of his app's agents about hosting returned Render as 'my pick', then Railway, Fly.io and others, with Vercel 'not suitable'. He said he barely knew Render but would probably use it, so 'Render just got a lead'.
“find 10 folks you know that have agents in production doing different things, 10 folks you trust, and ask them every two weeks, what would you recommend in this category?”
Listen on Apple Podcasts Positioning & marketing Link to this
Acquirers knowingly accept legal risk in license-and-hire deals to close overnight instead of waiting months for regulatory review.
Rory O'Driscoll offered a hypothetical: lawyers tell the acquirer they can buy the company on a Sunday night and have employees start Monday, versus filing with the US government and closing in nine months. The CEO accepts some risk. Even if the acquirer later has to settle and add, say, $200M, it may judge the risk worth it.
“I found a way that we can buy this company on a Sunday night and all the employees start on Monday morning”
Rory O'Driscoll warns that picking IPO bankers who pitch the highest price can backfire when real investor demand comes in lower.
Building on Dev Ittycheria's view that price likely sank Oura's IPO, Rory O'Driscoll said bankers tell you what you want to hear while competing for the mandate, then investors offer less. Choosing the banker with the highest number risks telegraphing a deal that then fails, which he called a far worse situation. Harry Stebbings argued Oura should have accepted the lower price and listed.
“will you just go for the person who says you're going to get the highest price but if he's whispering bullshit in your ear and it turns out not to be true then you end up in this far worse situation”
Listen on Apple Podcasts Fundraising & investors Link to this
Developer buyers show little loyalty and switch quickly to new tools or run several at once.
Drawing on 12+ years selling MongoDB to developers, Dev Ittycheria said developers are quick to jump to the 'shiny new toy'. He used this to explain teams shifting meaningful usage from Anthropic to OpenAI's Codex. He framed the open question as what Anthropic ships next and what alternatives emerge.
“having sold to developers the last 12 plus years is that there's not a lot of loyalty. Developers are very quick to use to switch from one tool to another or frankly use multiple tools”
Enterprise AI deployments may be stickier than developer or consumer usage because of contracts, training and workflows.
Dev Ittycheria said ChatGPT's roughly billion weekly users are a great distribution channel. He suggested the moat may 'potentially' sit in enterprise, where vendors must sign contracts, train thousands of employees and build workflows. He presented this as an initial reaction, not a firm conclusion.
“You have to sign contracts. You have to train thousands of employees. You have to build workflows and it may be harder to switch”
Token cost and budgets plus IP and data rights are important factors in which AI models enterprises choose.
Dev Ittycheria said he had just been discussing token costs and token budgets with MongoDB's CIO, noting no enterprise has unlimited budgets. The second factor he named was IP rights: how models are trained, who gets the data and how proprietary it is. He said these two are pretty important points when it comes to which models you choose and who you work with.
“I was just talking to our CIO and we're talking about token costs and token budgets, right?”
Given a choice between a private stake at $1.4T and a newly public stake at $2T, Rory O'Driscoll would lean towards liquidity.
Asked hypothetically whether a pension fund should buy OpenAI privately at $1.4T or Anthropic at its IPO at $2T, Rory O'Driscoll said valuing the two companies is a separate exercise. He would separately weigh liquidity against illiquidity, and prefers the liquid stock while so much information is unknown. Holding the private stake means accepting a three-to-four-year hold.
“I would much prefer to be in the liquid stock when the amount of information is as unknown as it is.”
Listen on Apple Podcasts Fundraising & investors Link to this
About 95% of CROs would see moving from advising one company to running sales at its competitor as normal.
Jason Lemkin said the CRO role has high turnover, hot jobs move fast, and CROs often treat competition as a game and are friendly with rival CROs. He acknowledged founders find it shocking. He stressed that the person was a board observer and advisor, not a full-time employee or board member, and said it would have been nicer to have 'a little more temporal separation'.
“I think 95 % of CROs is saying, this is how I found my last job.”
Listen on Apple Podcasts Sales team, hiring & comp Link to this
He has never seen a CRO move to a direct competitor and warns it does long-term reputational damage.
Dev Ittycheria said CROs recruit people by selling them on the company and how it will be transformative for their families, so leaving for a direct rival makes those recruits ask whether they were sold 'a bill of goods'. He said he had seen CROs move to companies in other spaces, but not to direct competitors. He would ask a departing executive whether no other company valued their skills, and said such moves come back to bite people.
“I think the reputational damage that you can do in a long term is going to come back and bite you.”
Listen on Apple Podcasts Sales team, hiring & comp Link to this
A full-time executive can leave for a competitor with as much information as an advisor, and advisor expectations are often left implicit.
Rory O'Driscoll argued that any full-time direct report can take a competitor's offer and walk out knowing everything up to their notice, and cannot easily warn the CEO until they have the job. A part-time advisor who is neither employed nor on the board has no fiduciary duty. He said there was probably an implied understanding in the Factory case that 'was never made concrete'. Dev Ittycheria disagreed, arguing employees see siloed information while board-level advisors see the whole strategy.
“There was probably some implicit understanding about the advisor that wasn't fully fleshed out on either side.”
Executive loyalty has permanently changed in the AI era and companies should expect people to move to competitors.
Jason Lemkin pointed to frontier-lab staff rotating between OpenAI, Anthropic and others, carrying knowledge with them. He also cited faster moves and very different compensation among top executives he knows. He said it can bother you, but the culture has changed and has to be accepted.
“I think you just have to expect people will move from competitors.”
Listen on Apple Podcasts Hiring & team building Link to this
The informal 'rule of two', under which departing executives took at most two people with them, appears to have died.
Jason Lemkin said that, in his experience, a departing leader could take one person (with some annoyance), asked permission for a second, and faced friction beyond that. He now sees people he regards as highly ethical ignore this. He attributed it partly to pressure to move fast in the AI era. Dev Ittycheria said he had seen people leave and 'carpet bomb' the organisation trying to recruit, which led to stern words.
“I think it's a symptom of change that the rule of two seems to have died.”
Listen on Apple Podcasts Hiring & team building Link to this
A VC publicly disparaging its own portfolio company hands rival firms a weapon in competitive deals.
Commenting on Vinod Khosla calling Factory, which Khosla Ventures led in its latest round, a 'struggling second-tier competitor', the speaker said they were flummoxed that he did it publicly. They said rival firms can now ask founders whether this is the partner they want when things go badly.
“And basically handed every other firm a weapon when they're competing on a deal saying, is this the partner you want when things go bad?”
Listen on Apple Podcasts Fundraising & investors Link to this
A VC backing two directly competing companies must never comment publicly on one versus the other.
Rory O'Driscoll said such situations are governed by non-legal but implied rules. The only way a firm gets through two competitive investments is to argue it loves its 'children equally' and say nothing public comparing them. He said those situations are especially fraught when the firm is actively involved, and that the Khosla episode was unhelpful for both companies.
“you love your children equally. You don't say anything in public about one about the other.”
Listen on Apple Podcasts Fundraising & investors Link to this
A near-frontier US open-weight model that is 3–4x cheaper lets enterprises cut cost without using Chinese models.
Dev Ittycheria (disclosing that Sequoia is an investor in Reflection) said Reflection's Beam gives enterprises near-frontier intelligence at three to four times lower cost. He said enterprises always hesitate over Chinese open-source models, especially in regulated industries or ones handling sensitive data. He called the launch very noteworthy for the enterprise.
“One, you get a US open source model that's near frontier intelligence. That's three to four times more cost effective, i .e. cheaper.”
Enterprise executives at Dreamforce did not want to use Chinese open-weight models and did so only under cost pressure.
Jason Lemkin said he had a surprisingly large number of executive conversations at Dreamforce and nobody really wanted a Chinese-sourced model, 'right or wrong, fair or not'. Those using them felt they were doing so under duress for cost. He said a US model that 'nails' it could see 'a torrent of demand'.
“nobody I talked to really wanted to use a Chinese source model, right or wrong, fair or not.”
Enterprises don't need frontier-level intelligence for every workload.
Dev Ittycheria said a model within about six months of the frontier that covers 90% of use cases would be attractive, especially for reasoning workloads like coding and agents. He said he didn't see why enterprises would not want to talk to such vendors immediately.
“this also reinforces the point that you don't need the frontier level intelligence for every workload.”
Dev Ittycheria expects Jevons Paradox to expand enterprise AI usage as costs fall, making the model market non-zero-sum.
Dev Ittycheria said there is a big gap between AI capabilities and enterprise skill sets. As costs drop and staff become more conversant, companies will deploy AI for more and more use cases. He said he sees this at MongoDB, and that the market is massive enough for both frontier labs and open-source providers to succeed.
“I think we're going to see Jevons Paradox kind of come into play”
Jason Lemkin estimates enterprise demand for tokens is probably an order of magnitude more than companies have figured out how to surface.
Jason Lemkin said everyone he spoke to at Dreamforce was overloaded with internal demand for tokens. He said this may lead to suboptimal model use and waste, but everyone has to manage it. Combined with reluctance to use Chinese models, he sees it as a big opening for US open-weight providers.
“the demand is probably an order of magnitude more than they've figured out how to surface”
Jason Lemkin is increasingly skeptical of published model evals.
Jason Lemkin said any eval can be made to look great. What matters to him is speed, real outputs, and performance on real and mission-critical workflows, such as correctly stating whether an item is in stock. He said evals focus on things like 'chess championships', so he would want to try Beam himself, for example on OpenRouter, before believing the claims.
“I found that every single eval needs like three asterisks and four daggers next to”
Jason Lemkin found that switching LLM providers is real work, even if easier than swapping a database.
Having recently swapped models himself, Jason Lemkin said you have to requalify prompts and redo workflows, contrary to what 'the internet says'. He still called it much easier than swapping out a database.
“You have to qualify your prompt. You have to redo your workflows.”
Jason Lemkin could see US models taking half of open-weight token flow within 12 months, but is unsure parity is real.
Asked what share of open-source tokens will run through US versus Chinese models in 12 months, Jason Lemkin said he could see it being half, given developer willingness to switch and the importance of sovereignty. He explicitly said he was not predicting it would happen, because he is skeptical whether parity is real or 'pretend'.
“But I could see it being half in 12 months.”
Dev Ittycheria expects model safety, measured by some benchmark-like mechanism, to become a major enterprise buying criterion alongside cost.
Dev Ittycheria said enterprises will care a lot about whether they can trust these systems. Following a recent Washington conference, he expects a self-regulating approach to emerge. He said the hard part is defining safety, and that a benchmark-like mechanism is needed even though benchmarks can be gamed.
“The more safe these models are with whatever mechanism that plays to prevent people from doing bad things, the more widely usable they'll become.”
Voice AI is not a fungible, price-driven category today because quality failures break the product.
In a mock investment committee on ElevenLabs at $22B, Jason Lemkin argued for putting 10% of the fund in. He said ElevenLabs' lead keeps widening, its margins are surprisingly strong and it could approach $1B in revenue. He had earlier predicted cost-driven substitution in voice, but now said a phone agent (e.g. for a flower shop) must answer in seconds and get the order right, which limits price pressure. He was shocked that relatively modest companies happily pay hundreds of thousands of dollars a year, though he said 12 months out is hard to predict.
“I'm shocked at the number of folks that are relatively modest companies paying hundreds and hundreds of thousands of dollars a year happily.”
Rory O'Driscoll sees large, fragmented, pre-technical categories that LLMs do much better, like market research, as top AI app opportunities.
Rory O'Driscoll said market research is a big existing spend category served by consulting firms and the last-generation players Qualtrics, Medallia and SurveyMonkey. He said each reached high-single to low-double-digit billions in market cap. LLMs fit the work well because voice is a solved modality and questions can adapt to answers, unlike canned surveys. He called it a top-five or top-ten app-layer use case, and said Listen Labs executed well and reached real revenue.
“It's an existing category with a big -ass spend, and LLMs just do it so much better”
Jason Lemkin questions whether a ~$20M-revenue AI app can move the needle for a Salesforce-scale acquirer.
Jason Lemkin said he generally likes Salesforce's acquisitions. With Salesforce near $50B and targeting $60B, however, he doubted a $20M next-generation survey product moves the needle. At that scale an acquisition must either tuck in or be big, and he wondered if this one 'will just be kind of forgotten in a couple of years'.
“I don't know how 20 million of revenue of next generation AI survey moves the needle.”
Dev Ittycheria splits AI apps into 'features masking as companies', which will sell, and franchises built on proprietary data loops.
Dev Ittycheria said a clever product on someone else's platform is frankly a feature, and he expects many such AI app companies to sell to larger companies for their distribution. Durable companies create a loop: usage generates data no one else has, that data improves the product, and the product attracts more usage. Competitors can then copy features but not interactions.
“are you building a feature that's masking as a company or a franchise, right?”
Rory O'Driscoll frames an acquisition offer as a cold-blooded test of whether you can reach 10x the offer.
Rory O'Driscoll said founders evaluating an offer must be cold-blooded about the likely trajectory. The question is how they feel about 'running the tape', and how likely they are to reach not just three times the offer but 10x or more.
“how likely are you to get not just three times this amount, but 10X this amount and more?”
Vendors must now market to AI agents, because being missing from an AI-generated answer is like being invisible on Google.
Dev Ittycheria said that for 25 years marketers optimised to show up on page one of Google and educate human buyers. Now they must ensure agents can find them, understand what they sell and see current, accurate information across the sources agents read. He said this extends beyond tech to e-commerce and retail, where failing to educate agents could send a business 'to zero'. He disclosed that Sequoia backs Profound, which works on this problem.
“if you're missing from an AI generated answer, it's like being invisible on Google.”
Listen on Apple Podcasts Positioning & marketing Link to this
Startups are disadvantaged with agents because agents see far more data about incumbents, so they must deliberately educate agents.
Dev Ittycheria said an agent knows your whole stack and application and makes very specific recommendations. However, the corpus of data about incumbents is far larger than about startups, so agents easily default to the incumbent. He said the onus is on startups to make their documentation, APIs and surrounding infrastructure show they are a first-class experience for an agent.
“it's actually incumbent upon the startup to figure out how they basically educate agents on what they do, the documentation, the APIs”
Listen on Apple Podcasts Positioning & marketing Link to this
Dev Ittycheria sees the bigger opportunity for AI-visibility vendors in shaping and creating content, not just analytics.
Asked whether companies like Profound and Peak are just a next-generation SEMrush, Dev Ittycheria said the bigger opportunity is to help shape and create the content that fixes the problem. Telling customers where they are missing is not enough. He said closing that loop makes the solution far more compelling.
“If they can close the loop, that becomes a very compelling solution versus just telling you, you're here, you're not here”
Listen on Apple Podcasts Positioning & marketing Link to this
Rory O'Driscoll expects AI-visibility tools to expand from showing up for humans in chatbots to showing up for agents, and to be a deeper market than SEO analytics.
Rory O'Driscoll, whose firm wrote a term sheet on Profound's Series A, said it is a natural next step for these companies to move from how you show up in ChatGPT for humans to how you show up for agents. The buyer is the marketer who must ensure the company appears correctly, for example someone at MongoDB checking how the company shows up when an agent picks a database. He expects more depth here than in the SEMrush space.
“I think it's going to have more depth than the SEMrush space, which ended up commodified and boring for a whole bunch of reasons.”
Listen on Apple Podcasts Positioning & marketing Link to this
Marketing platforms grow by solving the immediate pain point and then each adjacent one, as HubSpot did after starting in SEO.
Citing HubSpot, which began as an SEO company and added much more on top, the speaker said every marketer has a three-year list of pain around showing up in the new LLM world. Companies that solve these problems in sequence can make money, which is why the speaker called AI visibility a very good market.
“You solve the customers immediate pain point and then they realize they have an adjacent pain point and you solve that”
Listen on Apple Podcasts Positioning & marketing Link to this
'license-and-hire' deals like Nvidia–Groq may conflict with Delaware's principle that all common shareholders are treated alike, and expects the ex-employee suit to settle.
Rory O'Driscoll explained that these structures let the buyer reallocate consideration: staff who move across get a lot, while those left behind get little. Former employees who had already left were likely not considered at all. Because the class is probably small, he said he wouldn't be surprised to see it settle, though he thinks the plaintiffs probably have some legal argument.
“But Delaware Law says everyone who owns a common sharehold is exactly the same”
Founders should expect employees to ask what happens to their equity if the company's tech and part of the team are acquired in a licensing deal.
The speaker said investors long ago learned to negotiate protections for unusual exits, but employees never thought to ask. As these deals spread and people switch companies more often, the speaker said founders need to be able to answer what happens to an employee's equity if the tech and some of the team are bought but the company stays behind.
“what happens to my equity if the company's tech and some members of the team get bought”
Listen on Apple Podcasts Hiring & team building Link to this
Jason Lemkin predicts license-and-hire deals will die if one of these lawsuits succeeds, because they already carry double taxation.
Jason Lemkin said the structures are already horrific from a double-taxation perspective. Adding litigation risk and carve-backs for left-behind employees would make them unworkable. He argued many are mergers in substance despite documents saying otherwise, noted the structure is fading as quick deals have become easier under the current administration, and said even if one suit fails another may win.
“I think these deals will die because they already have double taxation.”
Rory O'Driscoll credits Meta's Muse launch to shipping a well-resourced product, which he says 'cures everything'.
Rory O'Driscoll, previously cynical about Meta's AI efforts and acquisition, said that whatever the internal gear-grinding, Zuckerberg and Alex Wang put their heads down and shipped. He said they launched Muse full on and resourced it with compute, producing something as compelling as any startup.
“they put their heads down and they shipped product. And as the CEO on this call, both CEOs, this call no better than me, that cures everything.”
It is very hard to be developer-first, enterprise-first and consumer-first at the same time.
Discussing OpenAI's underwhelming consumer reception for Dots, Rory O'Driscoll said doing one thing well is hard and doing two at once is super hard. He said OpenAI struggles with this constantly. Jason Lemkin suggested judging Dots as a persistent coding agent for Codex developers over 30–90 days instead of as a consumer product.
“It's just super hard to be developer forward, enterprise forward and consumer forward at the same time.”
Jason Lemkin sees Oura pulling its IPO despite about $1.2B and 74% growth as a bad signal for venture liquidity.
Jason Lemkin said that, despite talk of easy liquidity and frequent M&A, a company at this scale and growth with attractive recurring revenue failed to complete its IPO. He acknowledged it may have been overpriced and that Forerunner selling its entire stake was a large part of the offering. Dev Ittycheria speculated, without inside information, that price was the cause and that the optics of a holder selling its full position hurt.
“a company growing 1 .2 billion, growing 74 percent with pretty attractive recurring revenue The IPO got pulled.”
Listen on Apple Podcasts Fundraising & investors Link to this
The fundamentals of managing people have not changed in the AI era.
Dev Ittycheria pushed back on claims that 'everything's changed' with AI. He said recruiting, developing and holding people accountable work as they always have. Young founders he advises who have never hired a go-to-market or finance leader face the same questions he faced founding his first company in 1998, such as how to evaluate someone with skills very different from their own.
“I think how you manage people, how you recruit them, how you develop them, how you hold them accountable.”
Listen on Apple Podcasts Hiring & team building Link to this