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 734 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 0 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
Feig: AI agents cannot be trusted without hard data-leak guardrails
“We don't trust agents. We say, Hey, we can try to set rules, but there needs to be hard guardrails and blocks for things like sensitive data being sent across.”
Ding: Enterprise SaaS sales are harder because building in-house is cheap
“I mean, I think enterprise sales for these large companies is much harder because the time to build the same exact product in house is just significantly lower. There's also just less leverage. Like when you're doing a negotiation against the customer for why …”
Adcock: Real-world interaction data is the missing piece to true AGI
“Because we're able to, like, this interaction data of, like, touching the world and seeing what happens through trial and error is, like it basically most of human intelligence is built this way, and I think this is, like, the last missing piece to get the tru…”
Adcock: Physical AGI starts with solving robotic hands
“I think this is extremely important to get, if we want to solve, like AGI and get to, like, human intelligence in the physical world, like, it's all going to start here with the hands for us.”
Adcock: Humanoid Robotics Is an Intelligence Problem, Not Manufacturing
“In my mind, this is not a manufacturing problem. This is an intelligence problem.”
Harrison: Pacifism Is a Privilege of the Protected
“Pacifism really is a privilege of the protected. Like, there is somebody else who is willing to do those things and build those capabilities to protect our right to say, I don't want to be involved in that, and that's fine.”
Taneja: AI rollups will onshore labor previously offshored for productivity
“We've sort of led the wave around these AI rollups where the thesis was everywhere we offshored for labor productivity, we're now onshoring back with AI.”
Abraham: Bottom 75% can't compound wealth, forcing fintechs to monetize via speculation
“If in the US, sadly, if you're in the sub 75%, you live paycheck to paycheck to credit card, you do not have money to compound the markets. You just don't. And that's also why you see the people who have, who acquire those and build products for those and, you…”
Long-Term Tech Founders Should Not Fear Down-Round IPO Repricing
“If you're prepared for that journey, if you're ready to take it through the next 1015, 20 years, then it's time to go public. And it doesn't matter if on day one you're below your last private round, the question is where do you end up five years from now, 10 …”
Del Buono: Putting initial founder liquidity into friends' startups almost always fails
“I'll see someone with their very first liquidity, the very, very first liquidity, they take it and instead of doing something a little bit safe in case there's a rainy day, they turn around and they go put it in a bunch of like very early stage startups.
Often…”
Huffman: Pricing an IPO at a discount is best marketing spend
“If you go out at a lower price, and this is what I, the advice I generally give is go out at a lower price so you can have that momentum carry your price up. Everybody's happy. Your new investors are happy. Your employees will be happy. It costs you a little b…”
Multi-floor offices function effectively as remote companies, according to Bouaziz
“Most companies are remote. As soon as you start being on two floors, three floors, you don't know the people in your company. Whether they're sitting on the third floor or they're at home, it doesn't really make a difference to you, right?”
Raanan: Seed-stage product ideas and technology are noise compared to founder quality
“The idea and the technology at this stage, at the seed stage, is, is noise. It's part of the entropy. It's actually not helping me make the right decision. The best decision is to just watch the individual and decide if that's the individual you want to partne…”
Raanan: True company importance is measured by practitioner dependence, not ARR
“An important company is not necessarily measured by revenues or ARR or valuation. It's about, it's measured by how much the market, how much practitioners depend on you. Do they consider you as a vendor, one of their trusted partners? And there are many compan…”
Alfred Lin: AI market is not zero-sum because intelligence generates revenue
“Our view is that it's not a zero sum game. And part of that is, there is really no price on intelligence. You, if you're more intelligent, if your system is more intelligent, if your product is more intelligent, is your company more intelligent and makes bette…”
Laffont: AI threat hits SaaS terminal value, not current business operations
“The threat of AI for these companies isn't actually related to their current valuation or frankly, even their current business. It's more related to the terminal value of, okay, maybe some of these companies are not benefiting from AI today. They haven't re-ac…”
Laffont: Initial TAM size is irrelevant compared to market expansion potential
“I have a personal view that I've developed, which is actually the size of the TAM is irrelevant. So I never listened to whenever an entrepreneur will pitch me a TAM, I really don't think about the TAM. I think about two things. I think about number one, Whatev…”
Blania: Combating bots is a lost cause; humans must prove humanity instead
“I think it's not about combating the bots because that will be a lost cause, but rather every human on these platforms will need some version of a proof of human and actually prove that their content is authentic.”
Rao: True hardware innovation requires exploiting non-linear silicon physics
“What's really interesting is if something is linear and predictable, it actually is very easy to simulate. If you can simulate in a digital system, it's less interesting. It turns out, like, silicon has all these beautiful physics that are associated with it. …”
Das: Almost no iconic startup has originated from a VC's investment thesis
“Almost no iconic company in the history of venture capital has come from anyone's thesis area.”
Das: Enterprise products capable of PLG must adopt it or get destroyed
“If your enterprise product can be PLG, then it must be, otherwise a PLG company will absolutely eat your lunch and destroy you.”
Franceschi: Private fundraising valuations differ fundamentally from 100% full-liquidity acquisitions
“When someone fundraises a private company, you're selling one, two percent of the company, right? When you're selling, when you're doing a transaction like this one, you're selling 100% of the company. So there's a very big difference between putting in a hund…”
Mignano: Foundation models resemble general code rather than competing business lines
“Lightspeed From a very early on in, in sort of the AI kind of revolution took this position that actually AI and foundation models are more akin to software and code than they are to business lines and strategies.”
Luckey: Optical camouflage is not useful in modern defense
“The biggest problem with optical camouflage, and the main reason you don't see a lot of money being put towards it in the modern defense base is it's not actually that useful. It's useful against human adversaries who do not have any other technology to aid th…”
Stephens: Defense primes ignore startups until market share is threatened
“I think like broadly across the defense industry this is a pattern that you saw happen with both Palantir and SpaceX as well, is that the big players, the primes, they don't really pay attention at all until their lunch is being eaten. And then the moment they…”
Schimpf: Critical defense supply chains require commercially viable factories
“You have to solve this at a commercially viable level, not at a defense specific level. It will always then be too expensive and unproducible. If you just have a defense exquisite magnet factory, like that doesn't work. You need a commercially viable magnet fa…”
Stephens: Pacifism requires a governing system with a monopoly on violence
“Pacifism as an idea is only possible inside of a governing system with a monopoly on violence. Like, you can't be a pacifist without a government monopoly on violence.”
Luckey: Optical camouflage is easy to build but useless in modern defense
“The point is, visible spectrum optic camouflage is actually quite easy, but something like this would be trivially easily seen By you know, an infrared camera, thermal camera, you know, lidar, radar. So the biggest problem with optical camouflage, and the main…”
Tusk: High-quality startups generally do not need an investor's cash
“Like, any really great startup to get into typically doesn't need your money in the first place, right? Anyone who desperately needs your money There might be a reason why. I mean, occasionally there's a diamond in the rough that no one believed in, but genera…”
Tusk: Institutional LPs want familiar strategies, not truly differentiated ones
“What it turns out that institutional investors want to hear, they think they want a different, they say they want, they think they want a differentiated strategy. They don't. What they want is This is exactly what you already know and are comfortable with this…”
Kalanick Mobilizing Uber Users Was a Historic Political Innovation
“If you have a product that people are really passionate about and you have the right means to mobilize them. You can turn your customers into political advocates and be big, powerful entrenched interests, right? Like I might've deployed that really well, and I…”
Clark: AI-native youth excel at frontier model experimentation
“The people who almost hold the model weights in their brains because they've been brought up thinking about AI natively are really, really good at the creative thinking and experiments and hypothesis testing that is necessary to push the performance of these m…”
Ulevitch: US should build better anti-ship missiles rather than match China's shipbuilding volume
“I'm not sure that the right strategy for America is to say, oh, they can build a hundred ships a year, so we need to be able to build a hundred ships a year. Like, what I would say a better strategy is if China's able to build a hundred ships a year, what if w…”
Karp: Unconventional, stubborn visions are the only way to change the world
“The experts are like, you know, like, this isn't safe because in almost all cases this doesn't work. Yeah, they're right, but in the one case where you change the world, it's the only way it can work.”
Gil: Early AI founders should pursue easy ideas over complex ones
“Early in technology waves, that's when you have all the low hanging fruit. And so one of my pieces of advice to founders is often who are doing really hard things right now in AI. I'm like, why are you doing the really hard thing? Just go do the dumb, easy thi…”
Siddharth: ASI means automating 90% of computer tasks for 90% of humans
“And I think of ASI as us having automated 90% of the tasks that 90% of humans do today in front of a computer, right?”
Siddharth: Reasoning models ended demand for simple AI data factories
“This industry had a huge shift after the reasoning models came out late last year, O-one being the first. The shift is, before the reasoning models, this industry needed simple data. Gobs and gobs of simple data. You needed a data factory. Or, ah, this industr…”
Siddharth: Enterprise AI performs best on 0.5B-to-10B parameter custom models
“It'll probably probably be a smaller model. Oftentimes we see a half a billion to a ten billion parameter model. That's the regime, not a trillion parameter model. And this will be faster, more accurate.”
Rabois: Exceptional talent means top 0.1% in one trait or rare combinations
“There's only, there's two formulations. One is you are in the top call one to 10 basis points on some specific trait. And that can, they can be very different. You can be the smartest person I've ever met. You can be the most tenacious person, persistent best …”
Adding employees without project leaders decreases productivity and increases frustration
“It's because you don't have any more barrels, you're just stacking ammunition behind the same number of barrels, so you're not going to get any more done, and you're just going to frustrate people.”
Rabois: A company's CEO determines 99% of its success or failure
“99% of everything that matters is the CEO. The most important lesson for me over the last 25 years is if you have the right CEO, things that people believe to be impossible, Or very possible, sometimes probable. If you have the wrong CEO, almost nothing else m…”
Shadbolt: Quantum will parallel AI's sudden breakthrough via hardware and architecture
“It was really the convergence of two things to my mind the availability of hardware that was built for another purpose in this case, gaming GPUs, and then also progress on the architecture where I see very strong parallels between what we've done and what has …”
Shadbolt: Basic qubit physics is a small part of quantum scale
“That stuff is really important, but it's a small piece of the puzzle of building a data center sized machine.”
Shadbolt: Venture capital pressure for early revenue warps quantum startups
“We get coached that there is a certain way to do a venture backed company. And the coaching is very often you got to get some kind of revenue by any means necessary really early. And that's certainly true if you're a SaaS company, right? Like if I'm building a…”
Shadbolt: $1B in revenue marks the starting point for useful quantum computing
“When you're making a ton of revenue you're probably doing something that, you know, within reason you could call useful and like a billion dollars of revenue would be a nice starting point for a useful machine like this. Yeah, definitely what is not useful is …”
Electric boat startups face execution risk, not tech risk, thanks to EVs
“There's also not a lot of technology risk in what we're doing because the automotive industry has existed for the past 10 years and like has been plowing money into electrification and high voltage power electronics and battery supply chains for 10 plus years …”
Strachman: Unconventional paths trigger defensiveness because observers infer superiority
“When you're doing something different, What other people hear or see or imply about that difference is that what you're doing is better than what they're doing. There's always this comparison.”
Agrawal: Fast shipping velocity is OpenAI's primary defensibility
“Keeping up with OpenAI product updates is a full-time job, and that's the strategy, is ship fast. And I feel like that done at a consistent pace, I mean, just in, just this year alone, you've had Operator and Deep Research and ChatGPT agents and Codex, and the…”