The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 135 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 4 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Vollset predicts Anthropic will inevitably add advertisements to Claude
“Not only are they going to do it, but so is Claude. I guarantee it. You know, they might figure out something clever and do the thing, but like, there's no, there's just not a universe unless Claude get bought by Apple or something. That, that Claude aren't go…”
Vollset: Windsurf Deal Was a Massive Founder Payday Structured to Screw Employees
“The original deal, which was like a successful, holy crap, that's a big number deal was three billion. The Google aqua hire is 2.4. That's ballpark same number. So like, you got screwed as if this was a soft landing for the founders, but it was a massive payda…”
Vollset: Media Outrage Over Twitter Stems From Journalists Losing Privileged Status
“My honest to God view on this, and it may be deeply cynical, is that I think what has happened, the reason why there's so much heat and so much drama around this is because Two journalists, Twitter's very, very important, and they had a very special status und…”
Vollset: Public SaaS companies report zero AI disruption and no added churn
“Most of the like public SaaS companies are reporting, like, you know, they're coming like, we're not seeing any disruption whatsoever from AI, no additional churn, but in the public markets, the sky is falling.”
Vollset: SaaS founders frequently leave 60% to 80% of exit value behind
“It's not unusual that I hear about situations where I'm like, okay, well, yeah, but you left like you left like 60 or 70 or 80% of the value of the company on the table because you didn't know what was going on here. Like you didn't understand like the univers…”
Vollset: VCs only care about unsolicited, high-priced strategic acquisitions
“The only kind of M&A that I am interested in as an adventure capitalist is if a strategic comes to you and begs to buy it and is gaga for it and will pay a crazy price. That is the only kind of M&A venture capitalist will care about”
Vollset: Break-even SaaS with teams are worth more than solo-founder profitable SaaS
“So it's much more valuable to you as an acquirer if those two founders have sat down and actually put like a team in place and like made themselves a little bit of a less of a key person risk for the acquirer. And that's why like it's, yeah, you might've been …”
Vollset: Apple can win consumer AI interface without owning frontier models
“And I feel like that's actually where Apple could come in, and without, even without necessarily owning, like, a fantastic foundation model, Could come in and just be like the interface by which normal people interact with AI day in and day out. For sure, I th…”
Vollset: Autonomous AI agents cannot be truly useful without major security trade-offs
“You're almost never gonna get, like, a hundred percent security, because it's not helpful, unless it sees what I see. And if it doesn't see what I see, then it's just like, yeah, ok, like, can you do some web research for me? I'm like, yeah, that's, So can cha…”
Vollset: LLM Capability Gains Are Approaching an Asymptote
“The data shows which it's really reaching, looks to be reaching some kind of asymptote. Like this asymptote can be very high, but It's still, to me, it seems like each new model that comes out, whether it's from ChatGPT or Google or Grok or whatever, it seems …”
Vollset: Linear AI Capability Increases Require Exponential Compute Growth
“As you make a, like another step in capability, like the amount of compute, compute that you need to add, to make it smarter, as it were, just grows exponentially. So you get like a linear increase in capability with an exponential growth in compute. And that …”
Vollset: Vertical SaaS can match frontier LLMs with fine-tuned open-source models
“If it becomes so expensive that, you know, your model doesn't work, well then given the specialization you're in, and particularly if it's something like a very specialized vertical SaaS thing that you're doing, okay, you can probably get as good, if not good …”
Vollset: Low-valuation $75M exits yield identical 40x returns to $1B unicorns
“If you're B to B SaaS and as an investor, you can put money in that a couple of million and then they sell for 50 to a hundred million dollars. That's as good. Like, it doesn't matter to you. If you get 40 X your money, what do you care? Whether it, 40 exit me…”
Volset: Selling a SaaS at 5x ARR equals 15 years of dividends
“So tip, it's not unusual for a pretty well-run profitable scale, you know, at scale SaaS business to be doing thirty-ish percent free cash flow that could be kicked out of dividends. Now, if you do the math on that, if that same business is growing well enough…”
ChatGPT is a bigger shift than the iPhone, rivaling the internet
“And particularly with ChatGPT, it became very evident to me that this was like probably one of the bigger shifts in terms of a technical capability that would impact startups.
[189] Einar Volset: I think at the very least, the biggest comparative change that I…”
Dismissing AI as an NFT-like hype cycle is fundamentally wrong
“It's completely the wrong approach to think that this is just NFTs and stick your head in the sand. Like that is fundamentally wrong. Like I'll take an argument with anybody who thinks that.”
Einar Vollset says ChatGPT could genuinely hurt Google's search business.
“It's genuinely one of the first things I've seen. I've been like, truly, this could hurt Google, like, you know, not immediately, but something like that really could.”
Einar Vollset predicts AI will hit a trough in 18 months.
“It's probably overhyped at the moment. It usually tends to be, there'll be like a trough of sorrow in like, 18 months when everything is in the crapper. But I think like two, three years from now, we'll look back and we'll be like, oh yeah, this underlying tec…”
Vollset: AI startups cannot build moats without proprietary models or data
“I think it'll be hard to have a competitive moat if you don't have your own language, if you don't have your own training sets, if you don't have your own model. I think that would be true.”
Vollset: Pre-revenue VC funding effectively eliminates $20M to $50M exit optionality
“Like you can raise with just an idea or a deck or just a It's a team or whatever, but then the vision has to be big, and that does usually exclude you or preclude you from actually doing the kind of exit where like at 20 or fifty million bucks, because just th…”
Vollset: Early-stage startup success is almost impossible for investors to predict
“I think a lot of the time, I think just think as early stage investors, it's despite what some investors will tell you, I think it's almost impossible to really, really have a good sense of like, what's going to really Work. There's just a lot of randomness an…”
Vollset: Pre-launch YC companies raise at $12M to $20M pre-money
“Last I heard now, pre-product launch on demo day, it's like 12 to twenty million pre.”
Vollset: Early-stage funds cannot allow equity buybacks due to portfolio math
“Given how early we invest, and so Indy usually invested later than us, given how early we invest, you just kind of need to be an equity investor, and you need to sort of allow, basically to not allow founders to buy back equity. And, you know, we get, as you g…”
Volset: COVID-19 lockdowns will not last 18 months to two years
“Realistically, it's not like we're going to be in lockdown for 18 months to two years. Like, that doesn't, that just won't happen. Like, there are too many things that are likely to work out in one way, shape, or form.”
Vollset: SaaS PE Acquirers Now Require $2M Minimum ARR
“Yeah, there's a reason my book is, and Discretion Capital focuses on, like, two to 20. It's because, like, right now, yeah, it's probably more like two million than it is like one million.”
Vollset: Public SaaS Price-to-Revenue Multiples Are Below Industrials
“Despite growing significantly faster and being, you know, significantly higher profits, the sort of price to revenue multiples of SaaS companies is actually less than industrials now”
Vollset: Bearishness on SaaS assumes top software deployers will fail at AI
“If you're very bearish on SaaS companies, then you're sort of saying that, like, The companies that are the best in the world at deploying software are gonna be bad at deploying this particular kind of software”
Vollset: Standard software integrations are no longer defensible moats
“Integrations in and of themselves are not no longer a moat just because like, It's so much easier to build them, but to your point, like, if they're harder to get, maybe because everyone is protecting themselves from AI, then it could definitely be a moat, for…”
Vollset: Custom hardware integration creates defensible moats against AI replication
“If you have a hardware component that, like, is tightly integrated into your thing, and it's a key part of what you deliver, trying to think off the top of my head of what examples of this might be, but like, you know, a custom piece for some custom sensor, or…”
Vollset: Crunchbase lacks a proprietary data moat and is replaceable
“No, I don't think Crunchbase has it, not really. That's more like, I'm in there by myself, and I'm looking at data, and like, yeah, that's a little bit more, sort of, replaceable.”
Vollset: Downside risk of vibe-coded tools is too high for commercial businesses
“Now apply that to a business doing millions in revenue and have employees and to pay and all this stuff. Like it just, why would I want a vibe coded thing, you know, to save like an inconsequential, the downside risk is just way, way too high.”
Vollset: PE buyers hesitate on AI SaaS due to downside risk focus
“The thing, the key, key thing to understand here is that founders are sometimes a little frustrated with private equity because they think of them more like venture capitalists, and like venture capitalists typically will take, very gung-ho, will take extremel…”
Vollset: Private equity drives 70% of $2M–$20M ARR SaaS acquisitions
“The fact of the matter is, like, in this revenue range, about 70% of the deals are done by private equity, either directly, like, they are buying it because they want to use what, build it around it, which is basically what's called a platform acquisition, or …”
Vollset: PE tuck-in acquisitions often outbid strategic corporate buyers
“Tuck-ins often outcompete, like, strategic bids, because it's so valuable to them.”
Vollset personally bought public SaaS stocks because they are oversold
“I'm on record to say that I mean, I think even public SaaS companies are oversold. I actually personally rotated into some more public SaaS in the last couple of weeks, which, you know, may come back to bite me, but fundamentally.”
Vollset: AI model progress is reaching the top of an S-curve
“I think like the models are incrementally getting better, but not as much as fast as they were, relatively speaking, is sort of my view. So it feels a little bit like we're sort of at the, you know, at least they're starting on the top end of like an S-curve o…”
Vollset is bullish on Tesla for its proprietary real-world training data
“You're just gonna run out of training data, which is actually why I'm slightly bullish, quite bullish, actually, on Tesla, despite their enormous thing, because like, one of the interesting things about Tesla is like, someone told me this, which stats probably…”
Vollset: Superintelligence requires new breakthroughs, not just more compute and synthetic data
“There has to be, I think, and I think there's, I think most people now, if you look, talk to like leading edge research and stuff, sort of agree with this. They're like, look, there has to be another breakthrough to really start that, like, To get to the next …”
Vollset: Scaling vertical SaaS beyond $3M ARR requires enterprise sales
“You can get to, I would say, 1,000,002 1,000,003 million of ARR, somewhere in there, for a lot of vertical SaaS businesses with that sort of self-serve model. Like, you can differentiate the pricing, and like, you know, you gotta figure out, there's a lot of s…”
Vollset: 100x fund returns are impossible; investors require concentrated angel bets
“It is impossible to get a thousand X in a venture fund or a hundred, even a hundred X. If you're wanting to do that, then you should put all your money into single bets. Like you should be investing in individual companies and like concentrate your position as…”
Vollset predicts Bluesky will fail to gain lasting traction within three years
“No. I think it's just a hissy fit to a degree. Like, and I feel like, you know, it's been that, like there's been threads was kind of the same.”
Vollset notes zero-uncertainty startup opportunities are already captured by incumbents
“If there was a market opportunity that had zero uncertainty about it, how to do it, or whether it was worth doing, then someone with more better resource than you would have already done it.”
Vollset claims Australia's social media ban requires ID for internet access
“And the answer is apparently that they're going to require all Australians to provide identification, to go, to access the internet.”
Vollset: Gymdesk can displace "hated" incumbent Mindbody and eventually go public
“The big player in the space is publicly traded MindBody. Everyone in the space hates MindBody, and I think there's a path to where GymDesk basically becomes the better version of MindBody and potentially is public down the line.”
Vollset: SaaS Growth Stalls at $1M–$3M ARR Without New Products or Channels
“And I think the biggest mindset change that I've observed repeatedly is people, once they get past about a million in ARR two or three, they get to the stage. They're like, oh, like my growth is stalling out now. Like what I was working, like my market saturat…”
Vollset: Founders need experiment volume more than overanalyzing results
“The main problem isn't trying to figure out what's working. It's doing enough experiments so that something is working. And like, look, if they make a bunch of, if they do. 25 things and then two or three of them are working, but they're not totally sure which…”
Volset: One-time software pricing is not viable for average bootstrapped startups
“I don't think it's a winning formula for your average startup, you know, indie hacker, bootstrapper, tiny seed founder to be like, oh, screw it. Like, let's not charge 50 bucks a month or whatever. Let's just do 300 dollars once.”
Volset: Selling one-time software effectively builds in 100% monthly churn
“Basically, the way that I think about it, it's basically, like, you're basically, like, building in a hundred percent monthly churn, which, like, I spent so much time thinking about how to reduce churn, like, the fact that you would design your software busine…”