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 450 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 9 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
Gurley: China's Politburo Comprises Engineers While Washington Employs Lawyers
“The vast majority of the political borough are former engineers. So this is the ruling party within China and that the vast majority of the people in Washington DC are former lawyers.”
Gurley: Shenzhen entrepreneurial energy remains high despite reports of startup slump
“A lot of the locals heavily dispute that Financial Times graph that was going around about the number of startups. They said it just mismeasured the whole thing. And so, no, I don't sense that there's any lack of enthusiasm from entrepreneurs in towns like She…”
Gerstner: Only children born after Jan 1, 2025 receive $1,000 Treasury seed
“All kids under the age of 18, that's sixty five million kids are eligible... But only children born after January one, 20, 25, get the thousand dollars from treasury”
Laffont: Anthropic grew from $1B to $3B in revenue within five months
“We've seen Anthropic, we have data in here that I think it took him about a year to get to their first billion in revenue.
It took him three months to get to the next billion, and then it took him two months to get to the one
The next billion after that, right…”
Gerstner: Mag-7 revenue grew 20% compounded while OpEx grew 2%
“If you look at the Mag-seven over the last three or four years, they've grown over 20% compounded, but the number of employees, their OpEx, is growing at two percent.”
Thomas Laffont: AppLovin doubled revenue while cutting headcount over 35%
“At the same time as AI comes in, now the company's lean and mean, innovates, outcompetes companies like Google and Meta, doubles the size of the company as the employee count is down over 35%.”
Gerstner: OpenAI Hit 1 Billion Daily Searches 8 Years Faster Than Google
“Open AI reached you know, four hundred billion annual searches, eight years faster than Google. Right? So they're doing over a billion searches a day now.”
Gerstner: Around zero percent of leading-edge semiconductor chips are fabricated outside Taiwan
“Today, we have about zero percent of leading edge chips fabbed outside of Taiwan, right? Almost nothing in the United States.”
Gurley: China has four deep-pocketed open-source AI models
“Deep Seek, Led to Quinn. Led to Xiaomi has a, has their own model as well. They've all gone open source. And so this will be the fourth deep pocket funded model in China that are all open source.”
Gurley: Manus AI Operates Only on US Models, Not DeepSeek
“So, first of all, they've only operated on U.S. Models. They're a wrapper company. They don't have a foundational model, and they've only operated on top of U.S. Models, so they've actually never operated on top of DeepSeek or any of the other models we were j…”
Gurley: Manus stores all customer data on US hosting servers
“They host all of these on us hosting services. And so they're not actually operating any of their customers workloads in China at all. And the data is all resident on those us hosting companies servers. And so there's no customer data in China either.”
Gerstner: Around 50% of US AI researchers are Chinese
“I saw a stat the other day. I think 50% of researchers, AI researchers in the United States of America are Chinese.”
Gerstner: US Has Effective Embargo on China via 140% Tariffs
“We effectively have an embargo between us and China today, a 140% tariff, export controls, you know, and so we have to renegotiate that deal.”
Gurley: South Korea, China build nuclear power at one-fourth US cost
“Korea and China are delivering fission nuclear at one-fourth the price we are, and I think that Korea example, South Korea, is very important because that is a democracy. So here's one country that's authoritarian, one that's a democracy, and they're both achi…”
Gerstner: China has 100 nuclear power plants under construction
“There are a hundred nuclear power plants under construction in China. Power is not their limit.”
Gerstner: ChatGPT commands an 80% to 90% consumer AI market share
“Clearly chat GPT is a major thrust, a major focus of the business. It's the market leading consumer application probably has 80 to 90% market share.”
Gerstner: 90% of ByteDance's Business Is Outside TikTok US
“About 90% of ByteDance's business is not TikTok US. 90% of the value of the company is things like Doyan, which is the Chinese version of TikTok, and Daobao, which is the Chinese version of ChatGPT, and TikTok around the world, and so there's a huge and profit…”
Gurley: DeepSeek is pricing API inference at one-twentieth of OpenAI
“If you look at the models that are apples to apples on the API right now, that, that DeepSeq's pricing about one-twentieth of OpenAI.”
Gerstner: DeepSeek is diverting most of its compute cluster to inference
“They are massively compute constrained right now. So you've seen some tweets about this, people that are, you know, getting server delay and all this stuff with these guys because of their massive success. So they have a limited cluster to begin with, and they…”
Gerstner: OpenAI Is Already Running Workloads Out of Abilene Facility
“Larry said at the press conference, They have 10 buildings already built in Abilene. They have another 10 under construction, and we know that OpenAI is already running workloads out of Abilene, Texas, which is where this mega project's going to be built, and …”
Haas: OpenAI will have operational control of the Stargate project
“And the operational control will be from OpenAI. So they'll, they'll call the shots relative to all the things relative to the operation.”
Gerstner: Microsoft is on an $80 billion annual CapEx run rate
“In their most recent quarter, they did about twenty billion in capex bill, and you just, you know, he said earlier in the year, which caused, or a couple of days ago, which caused a bit of a kerfuffle that they expected to spend eighty billion in capex this ye…”
Gerstner: Microsoft generates $10 billion in AI inference revenue
“You know, as Satya told us on the pod, they have ten billion in inference revenue, and I think he expects that to grow. Significantly.”
Gerstner: Nvidia inference revenue is upwards of 50% of total revenue
“Jensen made, and maybe we'll insert the clip here on the pod, where he talks about his inference revenue is already 50% of his revenue, so already upwards of 50% of his revenue”
Gerstner: Newsom has not signed Diablo Canyon extension past 2029
“Gavin Newsom still has not signed the extension for Diablo Canyon beyond 2029. It's insane. We have to make that happen this year. It's 10% of California's green, clean power.”
Patel: NVIDIA holds 98% of AI workloads outside Google, 70% overall
“So I would say if you ignored Google, it would be over 98%. But then when you bring Google into the mix, it's actually more like 70 because Google is really that large a percentage of AI workloads especially production workloads.”
Patel: Apple rents Google's TPU silicon, but GCP AI rentals remain GPUs
“While they do have some customers for their internal silicon externally, such as Apple the vast majority of their external rental business for AI in terms of cloud business is still GPUs.”
Patel: Microsoft runs GPT models on AMD hardware for inference
“Microsoft has deployed GPT-style models on On other competitors' hardware, such as AMD they've, and some of their own, but mostly AMD, and so they can wring that out with software because they can spend hundreds of engineers, dozens of engineers' hours hundred…”
Gerstner: Jensen projected $1T in new AI and CPU replacement workloads
“And for the first time, he said, not only are we going to have a trillion dollars of new AI workloads over the course of the next four years he said, but we're also going to have a trillion dollars of CPU replacement, of data center replacement workloads over …”
Patel: AI training has barely tapped vast video data reserves
“We have barely, barely, barely tapped video data, right? So there is a significant amount of data that's not tapped, it's just video data Is so much more information than written data.”
Patel: Cisco was fueled by speculative debt whereas NVIDIA is cash-backed
“Cisco's revenue, a lot of it was funded through private-slash-credit investments into building out telecom infrastructure, right? When we look at NVIDIA's revenue sources, very little of it is private-slash-credit, right? And in some cases, yes, it's private-s…”
Patel: GPT-4 cost hundreds of millions to train, generates billions in revenue
“Hundreds of millions of dollars to train GPT-IV. And it's generating billions of dollars of revenue.”
Patel: Anthropic showed ~70% gross margins in its most recent round
“Or, you know, Anthropic, similarly, in their most recent round, they were showing, like, 70% gross margins.”
Patel: A quarter of Nvidia shipments could provide Llama-7B to humanity
“If you take one quarter of NVIDIA shipments and you said all of them are going to inference Lama-Seven-B, you can give every single person on earth a hundred tokens per minute, right? Or sorry, a hundred tokens per second. You give every single person on earth…”
Patel: SK Hynix memory is NVIDIA's largest COGS item, surpassing TSMC
“When you look at the cost of goods sold of NVIDIA their highest cost of goods sold is not TSMC, which is a Thing that people don't realize. It's actually HBM memory primarily SK Hynix.”
Patel: AWS Trainium2 offers highest HBM capacity and bandwidth per dollar
“Their whole thing at reInvent, if you really talk to them when they announced Trinium II and our whole post about it and our analysis of it is, like, supply chain-wise, this is, looks, you know, you squint your eyes, this looks like an Amazon Basics TPU, right…”
Patel: Google TPU clusters scale up to 8,000 chips today
“Nvidia's talking about GB 200, NVL 72, TPUs go to 8000 today, right?”
Patel: Amazon Trainium 2 costs ~$5,000 per chip versus $30,000+ for NVIDIA
“You're not paying, you know, north of 30,000, you know, 40,000 dollars per chip for the server, you're paying, you know, significantly less, right? 5000 dollars per chip, right?”
Patel: Amazon and Anthropic are building a 400,000-chip Trainium supercomputer
“Amazon and Anthropic have decided to, you know, make a 400,000 tranium server supercomputer, right? 400,000 chips, right?”
Patel: Nvidia Blackwell costs over twice as much to manufacture as Hopper
“The cost to make Blackwell is north of two X that of the cost to make Hopper, right?”
Microsoft missed early deep learning opportunity with Geoffrey Hinton to Google
“In fact, even Hinton worked, like some of the early work in DNNs happened when he was in residency in MSR and then Google hired him. So we missed, I would say, even in the early 20 tens, some of what could have been doubling down at around the same time that G…”
Nadella: OpenAI Briefly Used GCP Before Returning to Microsoft for Transformers
“The first time yeah, actually, Elon and Sam, they were looking for, obviously, Azure credits, and what have you, and we gave them some credits, and that time, they were more into RL, and Dota II, and what have you, and that was interesting and then we stopped …”
Nadella: Shopify, Stripe, and Spotify became Azure users via OpenAI APIs
“Take, you know, Shopify or Stripe or Spotify. These were not customers of Azure. They were all customers of GCP or they were customers of AWS. So suddenly we got access
Do many, many more logos who are all quote unquote digital natives who are using Azure in …”
Nadella: Microsoft Is Power-Constrained, Not Chip Supply Constrained
“I am power, yes. I am not chip supply constrained. We were definitely constrained in 24. What we have told the street is that's why we are optimistic about sort of, ah, the first half of 25, which is the rest of our fiscal year.”
Gerstner: US federal revenues grew at 5.8% CAGR since 2017 tax cuts
“So since the tax cuts in 2009 and 17, we've grown our federal revenues at 5.8%.”
Gerstner: PIF announced shifting more capital into Saudi Arabia than abroad
“FII. Which is doing, you know, like, I think an incredible convening of investors and leaders from around the world. I think they're doing incredible stuff, you know, in the GCC and Saudi in particular, but they said, we're going to deploy more of PIFS dollars…”
Gerstner: Hyperscalers and OpenAI are solely focused on 1-3 gigawatt power sites
“When I talk to CoreWeave, when I talk to Azure, when I talk to OpenAI, when I talk to Amazon, Google, et cetera, not one of them Is talking about anything other than how do we find the facilities that can support a gigawatt or two gigawatts or three gigawatts …”