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 1,444 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 13 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
Evans: AI will not fundamentally alter enterprise software adoption dynamics
“This will be more software, and people will use this to make tools the way they did with no code, and the way they did with Excel, and the way they did with all those previous things, but it doesn't actually fundamentally change the dynamic. It just shifts tho…”
Evans: AI Model Inference Alone Generates 40% to 50% Gross Margins
“And meanwhile, we sort of know that you have positive gross margins on inference alone of sort of 40, 50%, but you've got the cost of building the next model, and you don't know where the cost The cost will move, or the cost of the next model will move, and yo…”
Evans: DeepSeek Proves Anyone With Two Billion Can Build Foundation Models
“Like anyone who's got a couple of billion dollars can make a foundation model which is what DeepSeq showed us as well.”
Evans: AI foundation models become commodities if frontier progress slows
“If the frontier slows down, then this stuff is definitely a commodity.”
Evans: AI Model Makers Lack Value Capture Due to Identical Technologies
“Because right now there isn't a path to value capture because right now you've got a lot of people doing basically the same thing with basically the same technology and the same data and the same chips.”
Evans: OpenAI's new ChatGPT work product is a confusing disaster
“I was playing with a new chat GPT work product, which is clearly not that it's a complete disaster. It's chaotic, confusing mess”
Evans: LLMs lack broad product-market fit beyond software developers
“The LLM itself Is not a great product for most people. Usage is a mile wide and an inch deep, and you have this kind of polarization between people where this really, really works, and they really, really have product market fit, which is basically software de…”
Evans: Model makers will lose pricing power within five years
“And the paradox is like right now they can name their price, but that isn't where we're going to be in five years. You can argue about how quickly the infrastructure gets built out and how fast the GPUs arrive, blah, blah, blah, blah. Fine. But that's a supply…”
Evans: AI foundation models will likely become commodity infrastructure
“Are the model companies going to have come some kind of, you know, oligopolistic power over the tech industry, or is this going to be commodity infrastructure with,
The value being somewhere else to which I think the answer is it's commodity infrastructure, bu…”
Evans: Chatbots are a terrible user interface for AI
“I feel like chatbots are terrible interface and users need product and interface and tooling and company and go to market around that.
And so that we all have thousands of companies that go out and build that.”
Evans: External customers will spend $100B–$150B on AI coding this year
“This year, people outside of those labs will spend, clearly will spend something in the order of hundred, hundred and fifty billion dollars to use AI coding. Just to use AI coding. And that's not AI people spending money from NVIDIA. That's real money. And tha…”
Evans: Only idiots believe people will use vibe coding to rebuild Stripe
“Like, no, people are not going to vibe code to their own stripe, but that's kind of a straw man. Like it's, I'd say it's a straw man, except there are people who say this, but only idiots.”
Evans: A software company's moat is go-to-market, not writing code
“I don't feel like the reason it's hard to compete with X or Y or Z company is how long it would take you to write the code to replicate the product. It's almost never the problem. It's the problem. The hard part is everything else. It's working out what the pr…”
Evans: There are no network effects in building AI models yet
“There are no network effects in building the models yet. There may be in the future, but at the moment we don't know.”
Evans: AI model infrastructure likely settles into 3 to 6 company oligopoly
“So it seems quite likely that all of that will settle out at some point with some kind of oligopoly, where like the laws of gravity and the laws of financial gravity kick in, and while this is the amount of money that the revenue that you can get, and this is …”
Evans: OpenAI and Anthropic lack distribution and cannot out-innovate startup ecosystem
“You don't have the existing feature set that Google and Meta and Microsoft and Apple and Amazon and everybody else have where they can add it and Salesforce and, you know, Workday and, you know, Figma and everyone else, some of them, some of which will survive…”
Evans: User data does not currently make AI models better
“At the moment, they don't. At the moment, the companies all say, well, indeed, we don't train, we don't use your data to make the models better, partly because the amount of user data involved isn't enough to make the, relative to the broader scale of the trai…”
Evans: AI product heads cannot set roadmaps because research breakthroughs dictate capabilities
“What that's saying is the head of product isn't setting the product strategy and has no idea what the product is going to be in a month's time. Because there's going to be stuff coming out of the research lab that they don't know about, that the research lab d…”
Evans: Undifferentiated tech with divergent market shares creates market instability
“When the technology is basically undifferentiated, the product is undifferentiated, and you've got radically different market shares, that tends to be an unstable situation.”
Evans: OpenAI lacks a technology lead, lock-in, or network effects
“If you're open AI, you've got, don't fundamentally have a technology lead. You're first among sequels at best. You've got this giant user base. That's not, that has very shallow usage and engagement. And it isn't really locked. It isn't locked in. It doesn't h…”
Evans: The delusion that AI will turn non-coders into programmers is wrong
“There's a recurring delusion throughout the history of software that firstly, people always think there's going to be a general purpose abstraction layer that will just do everything. And secondly, people always think that everyone who doesn't write code will …”
Evans: AI foundation models lack network effects and sustainable moats
“There's no network effect that we know of yet. And so unlike previous kinds of software platform shift, there's not any Obvious strategy in building the models where you can make yours better than everybody else's by doing that, following that plan. Well, you …”
Evans: OpenAI will struggle to make data centers better than rivals
“In the end, A, it's going to be very hard for particularly OpenAI to, for their data center to be better than everybody else's or cheaper or more efficient than anybody else's.”
Brown: Silicon Valley is building incredible AI capabilities that nobody needs
“And maybe this is the first time that the disconnect is just so obvious that we have this insane capability of building these things that are incredible and yet no one actually needs it right now.”
Evans: Tech companies will spend over $400B on AI infrastructure
“These companies will collectively spend something over four hundred billion dollars building infrastructure.”
Evans: AI models are commodities outside specialist use cases
“It's been very clear for like the last year that these models are commodities outside of specialist use cases and outside of very, people who are very, very deeply into using them all the time every day. Unless you're doing image, for image generation or codin…”
Evans: Sora, AI app platforms, and AI browsers will likely fail
“The Sora, the app platform, the browser, it's quite more than likely that none of those will work.”
Evans: No competing AI app has more than 5% of ChatGPT's usage
“Go look at App Store ranking, or Google Trends, or usage data of any kind that you can get, and you see that basically it's ChatGPT, and then none of the others are more than five percent of that.”
Evans: Lack of network effects makes ChatGPT's early lead fragile
“Given that there's no network effects, or don't, or not yet any network effects in LLMs, that feels very fragile. Like, you know, if you remember, you know, I cannot, you know, remember, like, everyone used Yahoo, and then everyone switched to Google, because …”
Evans: AI assistant usage remains tiny compared to Google search
“Use of all of these AI assistants is still tiny compared to Google. And there is this sort of bubble talk in the valley that, oh my god, no one uses Google anymore, which is just not true unless you're in the valley.”
Evans: ChatGPT and Claude are thin wrappers over models
“The only actual real thin GPT wrappers are the apps from these companies like chat GPT is a thin GPT wrapper. Claude is a thin GPT wrapper. Like Claude, the website, Claude, the app, is a thin GPT wrapper. It's an input box and an output box, and a little spri…”
Evans: Mainstream users have never heard of Claude despite Silicon Valley hype
“This disconnect between how everyone in the Valley is obsessed with like Claude is three percent better than ChatGPT this week and next week Gemini is overtaking Claude. But then you look at the user numbers. I know users have never heard of Gemini or Claude. …”
Evans: Apple Acts as 'Dumb Money' in TV Streaming Strategy
“It's like all they're doing is being what in the value we used to call dumb money. You know, they go to LA with a big checkbook, and they write checks to people in LA, and people in LA give them a TV show. And they go, and Apple goes and runs it on their infra…”
Evans: AI foundation models are undifferentiated except by icon color
“What you have as the model is completely undifferentiated, except at the level, as I said, of what color is the icon.”
Evans: Cheaper AI and Jevons paradox will not guarantee pricing power
“Just because if you make something much cheaper, people will probably use much more of it up to a point. Although interestingly, there's a debate now as to whether mobile data traffic is actually slowing down. But that doesn't mean you'll be able to charge muc…”
Evans: Big enterprises are running AI trials, but few use it broadly
“Every big company's got a trial. No one's really using it very much, or not many people are using it very much. There's a small number of fields where you had this immediate obvious use case, like it's great for coding, it's great for marketing, it's great for…”
Evans: Testing LLMs on complex logic is like testing Apple II for five-nines uptime
“This is, you know, classic disruption that the new thing tends to be crap at the stuff that was important for the old thing, but then it does something different that's new. And if you test an LLM by asking it to do complex logic problems or masquerade problem…”
Google paid $25 billion last year in device revenue share
“Google is paying a revenue share of 25 to 35% of the revenue they get from each search back to the device, and that was twenty-five billion dollars last year.”
The US court cannot compel Apple to add Safari choice screens
“The court can't make Apple put a choice screen in Safari. Like, Apple isn't in court. Like, they haven't been convicted of anything.”
Google pays Apple $20 billion annually to prevent vertical query routing
“And Apple is being paid twenty billion dollars a year not to do anything else here. So they're being paid twenty billion dollars a year not to say this is a hotel's query, maybe we'll send you straight to booking.com. They're being paid twenty billion dollars …”
General web search is a natural monopoly comparable to utility companies
“The general web search engine is a natural monopoly because you need a lot of money, and then once you spend the money, no one else can do it. So it's like a water company or phone company.”
LLMs are poorly suited for web search due to invisible error rates
“I'm really skeptical that you can use an LAM for general web search, just because I think there's a combination of, like, the error rate is real, it matters, and you can't see it.”
Apple is more likely to replace ChatGPT than build a search engine
“It's much more likely, I think, that Apple will shift questions from ChatGPT to its own foundation model than it would shift questions from Google to its own general search engine, I think.”
Apple will maintain Google defaults while routing specialized queries to AI
“Apple will carry on treating Google as the default general search engine. They'll just peel more stuff out into their own foundation model or into chat TPT or into booking.com.”
Divesting Android and Chrome leaves Google's text ad monopoly completely intact
“Breaking up Android and Chrome doesn't change Google's control of text ads.”
Evans: High inference costs prevent free 100-million-user consumer AI apps
“Because at the moment you can't make a consumer app that's free and have a hundred million users because you can't afford the inference cost.”
Evans: 99% of crypto and NFTs were useless scams
“It wasn't all a scam. Just 99% of it was a scam and useless.”
Evans: Probabilistic AI is a feature, not a standalone product
“It seems to me like the Occam's razor is, if you can't guarantee it's right, then it's features. It has to be wrapped in something that can manage or control or shape expectations or something around what it is that this is doing, because you can't just use it…”