Nico Wittenborn runs Adjacent by himself, raised a $40 million first fund on the back of a $10 million first close, and is now investing out of fund four.
When he started in 2019 nobody was doing it that way. Two years later the solo general partner was the thing everybody in venture was talking about, and he says the majority of new funds raised last year were solo-GP funds.
"Most of the venture bets do not turn out and they are orphaned at one point."
He was an early investor in Oura and in Bending Spoons, wrote the second check out of his first fund into Speechify, and spent five years at Point Nine and three at Insight Partners before going out alone.
The full segment is covered here so you can skip it. 56 minutes of audio, 18 minutes of reading.
Here are the 13 lessons that matter.
👤 Guest: Nico Wittenborn, founder of Adjacent, a solo-GP venture firm
🎙️ Hosts: John Coogan and Jordi Hays, who run TBPN's daily live tech show
👥 Also on: Scott Keogh of Scout Motors, Mitchell Green of Lead Edge Capital, Ben Gilbert and David Rosenthal of Acquired, and Faraj Aalaei of Cognichip, in separate segments of the same episode
📰 Published: 14 September 2026 on YouTube (TBPN)
🔴 YouTube | 🟣 Apple Podcasts | 🔗 Episode page | ⏱️ 56 min | ✅ Time saved: 38 min
Key Takeaways
Bending Spoons buys software companies at a tenth of what the market once paid for them, and it is often the only bidder in the room
At three times revenue, Wittenborn said, an acquirer is close to stealing the business
Consumer subscription businesses are more churn-prone than software sold to companies, and they hit a ceiling
His answer was to back the buyer of those businesses rather than only the builders
One payments company now sits behind 60% of newly launched apps, and it shows the launch rate exploding
The general-purpose chat assistants have already absorbed most small consumer app ideas
His example is checking the surf: no new app needed
A trillion-dollar company publicly mocking an eleven-month-old startup is new behavior
Meta stayed quiet while it cloned TikTok and Snapchat
Europe should stop trying to build its own large language model and build the thing nobody has yet
The European win last cycle was Spotify, which was built on top of Facebook rather than against it
The show's hosts argued that if AI development is to be paced, electricity generation should not be
1. The Pacing Timeline
The hosts opened on the weekend's argument about slowing AI development down, and laid out how it got here.
The idea has a seven-week paper trail. A "Pacing the Frontier" open letter on 28 July, signed by employees and leaders from OpenAI, Anthropic, DeepMind, Meta and other places; Sam Altman saying publicly the same day that the industry may have to pace the rate of AI development; OpenAI telling Axios two days later that it had helped shape the petition's language and that Altman had discussed pacing with White House officials
OpenAI then published its own version. On 18 August it put out a document on pacing model development in an era of cyber-critical capabilities, and said it had slowed scaling and paused a major reinforcement-learning run, a two-week training pause
Anthropic escalated it on 31 August, with a post that explicitly discussed pacing the frontier and called for coordinating pacing mechanisms
The essay that set the weekend off landed at 7AM Pacific on Saturday 12 September. Dario Amodei's "We Must Pace the Frontier." The hosts tracked the reaction hour by hour: Elon Musk engaging by 8AM and endorsing it by 9:30, Altman endorsing at 4PM, Demis Hassabis agreeing in general while arguing it needs work
The three-step plan is what everyone argued about. Independent third-party evaluators given access to AI companies to verify safety practices, with METR named as a possibility; frontier labs working with governments to set AI safety standards; and the United States and other democratic countries trying to coordinate with authoritarian governments, which in practice means coordinating with China
The hosts' objection to step one was that the evaluators are not outsiders. METR employees have worked at OpenAI and at Anthropic, and Anthropic staff have worked at METR — a revolving door rather than an accounting firm with no view on the AI future
Step two runs into antitrust, which is why Amodei asked for a Sherman Act waiver. One host's analogy: if American, Delta and United all agreed to cut flights for safety reasons, margins would rise, consumers would pay more, and that is precisely what the Sherman Act exists to stop
The risk they flagged sits below the frontier. A small team like Instinct, which is not training its own models, cannot afford a year of negotiating with third-party evaluators, embedded staff and a government review process. "You can't have a breakout consumer product."
2. Sacks, Trump And The Grid
The pushback the hosts read out came from the White House AI adviser, the president and an investor, and then they made an argument of their own.
David Sacks' post accused the two leading labs of asking permission they do not need. The host read it out: "You've also claimed the lead is widening because of recursive self improvement." And: "Stop pretending antitrust law has to be suspended so you can form a cartel." And: "Stop pretending METR is independent when it is intertwined with Anthropic's investors and staff."
Donald Trump posted twice over the weekend and the hosts read both. The first said the only controls or guardrails AI needs is a strong and smart high-IQ president; the second called the idea that AI is destroying humanity a hoax, and said President Xi had announced China will do nothing to stand in the way of AI
They noted the contradiction in accusing the labs of a conspiracy against AI, and the oddity of a president complaining that Google wants to build a plant in Finland because US permitting is too difficult
Gavin Baker's roundup was the hosts' anchor for what actually changed. In his account the only tangible new fact is that OpenAI and Anthropic will have embedded third-party evaluators from unknown organizations, and having them is smart because there is no Section 230-style liability shield for model outputs, so showing a duty of care will matter in future litigation
One host's own proposal was to pace the frontier and not the power. If capital and talent are not going into blitzscaling data centers, put them into nuclear, solar and grid capacity instead — the worst case is a country that cannot build artificial general intelligence but has a lot of cheap energy
The counter-argument from the other host was about how diffuse power is. Chips have a narrow set of uses; electricity can do anything, so restricting it is a blunter instrument
On whether slowing down is good business, the hosts cited Ramp's economist. Ara Kharazian had found AI spending declined a little in August, and that this was mostly price cutting rather than people using less AI — so a slower frontier with more competitors could compress the leaders' margins rather than protect them
Both labs are public benefit corporations, which the hosts said gives Altman and Amodei room to tell investors a decision will slow revenue growth and have them wear it
3. Selling Unlocked iPhones
Wittenborn's first business was arbitraging Apple's German carrier deal, and the hosts took him back to the start of it.
He grew up in Tübingen, a small town in southern Germany, and spent his junior year of high school on an exchange in Candor, New York, three and a half hours north of the city
The letter told him he was going to New York. He had the idea of New York in his mind and ended up in a town he had to look up. He said it gave him exposure to both the coastal, urban side of the country and to small-town life
The iPhone launched in Germany locked to a two-year contract and was expensive, so he bought a refurbished handset, unlocked it with a tool called Red Snow, and started reselling them while he was studying
He remembers the tooling by name. Red Snow, and the other jailbreak tools of the period, which the host said sounded sketchy
4. Point Nine And Insight
His way into venture was a cold email to a Berlin incubator in 2010, and it cost him half his salary.
The incubator was building companies and had a €6 million fund on the side, and invited him to work on it during his studies
He turned down a consulting offer that paid twice as much. "So it was really an important decision back then."
The fund spun out and became Point Nine, which he described as the first institutional venture fund in Berlin, and as historically the best performing early-stage firm in Europe. He was there five years
The firm called its approach remote VC — based in Berlin, looking for the best companies anywhere that fit the thesis. It was focused on software sold to businesses, and was early in Zendesk and later Loom
Insight Partners recruited him in 2016 and he moved to New York. "So the founder of Insight, Jeff Horing, started the firm when he was 26." A $20 million fund then; $100 billion now, with Horing still running it
Insight's edge is outbound sourcing, and he thinks it survives AI. Analysts who once cold-called and now email, carrying a price in mind for every company. "Once there's signal and press and more data, then the AI picks it up." What the analysts do that software does not is discover companies while they are still unknown, and build relationships over two or three years before investing
"So they're reaching out to someone and you still want a human to human interaction with a founder because they don't want work with a computer, not yet, at least."
5. Solo GP Before It Was Cool
He agreed his separation from Insight in 2018, left in early 2019, and started Adjacent as a firm of one.
Doing it alone was not normal then. Two years later it was the main thing people were talking about, and he said the majority of new funds last year were solo-GP funds
He now backs other solo general partners, having invested in a number of those funds and anchored a first solo-GP fund out of Adjacent
He deliberately did not use the new fund-administration services. AngelList and Carta were just starting to offer them. "And the reason for that was mostly that I didn't feel like they were institutional grade yet." He wanted institutional investors and found the flexibility limited, so he hired law firms, accountants, a fund administrator and an auditor instead
He thinks that has since changed, which makes starting a firm easier now than it was for him
Asked what supercharges a one-person firm today, he named AI — automation, agents, research, diligence and sourcing
6. The 18-Month First Fund
The first fund took a year and a half, and two other things were happening at the same time.
COVID arrived in the middle of it and his wife was pregnant with their first child. "And what made it harder was COVID started at the same time."
The first close came from founders he had already made money for. Revolut's founder was one of his first limited partners, along with the founders of Calm, whom he had backed at Insight, and a group of general partners including the Point Nine founders
Two established funds helped him get the first fund away — Thrive and Founders Fund — and the first family office in was SES, which had been early in both of them
He started investing off a $10 million first close, and his second investment was Speechify, which did very well
With that proof he went to institutions and closed a $40 million first fund, and has increased the size with each one since, adding roughly one limited partner per fund. He is now investing out of fund four
7. The Consumer Ceiling
Adjacent started as a consumer-subscription fund because everyone else was chasing software sold to businesses, and then ran into the limits of the category.
In 2019 and 2020 every new fund was doing enterprise software or crypto, so a consumer subscription thesis was differentiated. The first two funds were focused on it almost entirely
The problem is churn. Consumer subscription businesses lose customers faster than business software does, and depending on how deep the market is they hit a ceiling — at millions, tens of millions or hundreds of millions of revenue
To get a venture-scale outcome from there you have to keep growing fast at a very large revenue base, which he said ChatGPT managed: in the beginning 75% of its revenue was prosumer and consumer
"So the thesis was good, but I'm not in OpenAI."
The compensating virtue is the cost structure. These companies reach revenue quickly, are cash efficient and need far fewer people, operating expenses and distribution spending than a software-for-business company — tens or hundreds of millions of revenue, but very profitable
That is what led him to the aggregation idea: if the assets are profitable and capped, there is a business in buying them and running them together
8. Calm's Lesson, Oura's Deal
Two investments taught him the same thing, that a hardware company is worth more once it sells a subscription.
He invested in Oura around the time of his first close, before he had a fund, and did it personally through a special purpose vehicle with friends — the team he had led at Insight, who went on to found Left Lane
The reason he understood Oura was Calm. "So a lot of the growth actually came from people not wanting to meditate but needing to calm down before sleeping." Calm's growth inflected when it launched Sleep Stories
Oura was a hardware device with no subscription at the time, and he was among those pitching the move to one. Hardware plus subscription became a thesis of its own
Tractive was the same shape and probably the largest exit Austria has produced. A pet tracker built in a small Austrian town — he called it Oura for pets, plus location — sold this year, to Bending Spoons
The pet version has better retention than the human one, and he explained why. "Because your pet doesn't come to you and say, hey, I want a churn from this dog food." A person takes the wearable off when the fad passes
9. What Bending Spoons Pays
The host asked about the gap between what Bending Spoons will pay for a software business and what venture funds pay to own the same kind of company earlier.
Wittenborn expects the acquirer to keep winning those auctions. He said he would not be surprised if Bending Spoons buys a lot of these companies, and that he feels very strongly it will happen
The price is set by the absence of other bidders. Looking at the recent acquisitions, he said, you can imagine Bending Spoons was the only serious bidder for some of these companies, so it gets to set its price
His anchor for what that price is: a tenth of the peak. He said Bending Spoons buys at 10% of peak market capitalization
On the multiple, he said the buyer may be getting a bargain. "Buying a software company at like three times revenue is like kind of could be stealing it."
Two things could close the gap. If the thesis plays out and the world gets comfortable with the risk, more buyers with capital appear and prices rise
The likelier path is that the mismatch persists, because he keeps seeing companies with $100 million of revenue valued in the billions
His explanation is structural, not a claim that AI is a bubble. Venture is momentum-driven and consensus-driven, so hype cycles form; he said plainly that he is not calling AI a hype, and that the technology is changing the world. But money chasing that always funds things that do not work out
"Most of the venture bets do not turn out and they are orphaned at one point."
"The founders move on, the VC moves on, people just want to have some money back to invest into the next hot thing."
10. Chat Apps Ate The Niches
Asked why the consumer AI explosion people predicted a year ago has not arrived, he said the assistants took the ground first.
The host's framing was that the app charts are full of chat apps and short AI drama stories, not the wave of new application types people expected
Wittenborn's answer is that a general assistant covers too much. His example was checking the surf. "I don't need like the AI for checking the surf app. Just need ChatGPT or whatever."
The number he watches for launch volume comes from a portfolio company. "One is a company called RevenueCat, which is powering 60% of the new launched apps with their subscriptions today." Through it he can see an immense increase in new apps being launched
More apps means the bar for a venture outcome went up, not down. Lower barriers serve more niches, and make a venture-scale result harder to reach — which is why he moved on from consumer for a while
Where he still invests in consumer is where the technology goes deeper. He named Popcorn, a next-generation telecommunications company building its own core network infrastructure, with an eSIM, an app, an international number and AI features such as a call assistant that screens calls instead of taking a voicemail
He uses it himself to solve a problem he created. His number leaked through years of customer-service forms, so he keeps one number for the internet and a Popcorn number almost nobody has
11. Meta Versus Instinct
The personal-agent race is where he says an early-stage investor now struggles to get a position.
He expects things to work in personal agents and cannot price them. It is difficult to know what the labs will absorb, Meta is playing for the category, and it is so crowded that an investor who wants to come in early cannot — the entry points are at $500 million or a billion dollars of valuation
The incumbents are not sitting still either. Scaled companies with engaged founders watch model releases daily and have teams sprinting to ship the new feature on day one, which removes the opening for a startup selling software-plus-AI into the same problem
What is genuinely new is the public hostility. Meta launched Muse against the eleven-month-old Instinct, and a Meta executive the host named only by first name has been taking direct shots at it, including a nickname
"But it rare to see a trillion dollar company like hackling"
He contrasted it with how Facebook behaved last time. "They didn't really do that with TikTok or Snapchat. Like they were pretty quiet." The line then was that they were excited to launch stories, and nothing more
He is unsure when Muse was actually started. Meta tried to buy Instinct, he said, but that does not mean it was not already building
12. Europe Runs Its Own Race
He has made money in Europe — Revolut, Bending Spoons, Oura — and he does not think the answer to AI there is a European model.
His pitch to a friend in Austria was to go and build a neo cloud. "I know people in The US that are dumb as rocks" who are going to be billionaires because they picked a hard problem with enormous demand and worked on it for years
The European companies he backed were in categories that were not yet hot in the United States, which is part of why they were available
He is blunt about the causes of the lag. Regulation, bureaucracy and labor law, with startups tied into the same regimes as large established companies, which stops them moving fast
He does not accept the sovereign-model argument. "When I see these things where it's like Europe needs its own LLM, I'm not entirely sure about that." China firewalled itself and got a Chinese Google and a Chinese Facebook; Europe has not gone nearly far enough for that to work
The precedent he prefers is Spotify. Europe never got a European Facebook, but it got Spotify — built on top of Facebook's network and algorithm, compatible with it rather than a clone of it
Energy and defense are where he sees the dependency problem, and the ecosystem effect takes time: Bending Spoons is listed in the United States because that is where the markets are, and still describes itself as an Italian company
13. Hypersonics In LA
The last question was why he was in Los Angeles, and the answer was a company a long way from consumer subscriptions.
Inversion Space is the first investment that broke the thesis. It is doing hypersonic delivery from space, has partnered with Unreal on the Golden Dome and has signed a contract with NASA
The other stop was a fund, not a company. He is a small limited partner in Kantos in San Francisco and was in town for an event it was hosting
He also has a Ukrainian defense company based in Estonia, which he said has become something of a defense hub, though the region is not one of his core geographies
Wittenborn's bottom line is that the money to be made in software right now is in buying finished companies cheaply rather than funding new ones expensively, and that the consumer opportunity he built a firm on has been narrowed by assistants that already do most of what a small app would.
Bonus Insights
One host did the show recovering from rhabdomyolysis, which he said he got from working out too much without hydrating, and which he described as the dumbest illness you can possibly get. He turned down a doctor's note on the grounds that the job can be done sitting in a chair having a conversation
The tennis anecdote that opened the show was about attention. Alexander Zverev won the US Open point and started setting up for the next one, realizing only from the crowd that the match was over — which the hosts said venture capitalists would recognize as focusing on the next investment rather than the last exit
They admired the camera work more than the tennis. With seven seconds on the clock the director cut to Pierce Brosnan
Wittenborn's read on diamonds, prompted by a presidential post, was that the industry is in trouble because buyers are moving aggressively to lab-grown stones
The show closed on a breaking story: the president telephoned Jensen Huang live on stage at the All-In summit, and Huang put him on speakerphone. Trump's line, played back on air: "The great thing about life is that Jensen can develop the most complex computer chip in the world that nobody can copy for ten years. But he can't figure out how to put me on speakerphone."
Products, Companies & Tools Mentioned
Adjacent (Wittenborn's solo-GP firm, now investing out of its fourth fund)
Bending Spoons (The acquirer he says buys at a tenth of peak market value and is often the only serious bidder)
Oura and Calm (The pair that produced his hardware-plus-subscription thesis; Calm's growth inflected on Sleep Stories)
Speechify (The second investment out of his first fund, and the one that proved he could pick alone)
Point Nine and Insight Partners (Where he learned the business; Insight's founder started it at 26 with a $20 million fund)
RevenueCat (Handles subscriptions for 60% of newly launched apps, which is how he tracks the launch rate)
Popcorn (Next-generation telecom company with its own core network, an eSIM and an AI call screener)
Tractive (Pet tracker he called Oura for pets; sold to Bending Spoons in what he said was probably Austria's biggest exit)
Revolut and Blinkist (European companies from his earlier career; Revolut's founder was one of his first backers)
Inversion Space (Hypersonic delivery from space, partnered with Unreal on the Golden Dome and contracted with NASA)
Left Lane and Thrive and Founders Fund (The funds around his start: Left Lane's team did the Oura vehicle with him, the other two backed fund one)
Meta (Launched Muse into the personal-agent category and has been taunting an eleven-month-old competitor)
OpenAI and Anthropic (The two labs at the center of the pacing argument, both public benefit corporations)
METR (Named as a possible third-party evaluator; the hosts questioned how independent it is)
Ramp (Its economist's card-spending data was the hosts' evidence that August's AI spending dip was price cutting)
Books & Resources Mentioned
We Must Pace the Frontier (Dario Amodei's Saturday essay, and the three-step plan the whole segment argues with)
David Sacks' post on X (The reply the host read out in full — permission, cartels and METR's independence)
Nico Wittenborn - Finding the Adjacent Possible (The earlier Invest Like the Best interview with Wittenborn that the hosts pointed listeners to)
The Pacing the Frontier letter (The 28 July open letter signed by lab employees and leaders, where the phrase started)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: