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 20 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 1 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
Calacanis: Major enterprise customers will abandon OpenAI and Anthropic for open-source
“And I'll make this prediction here, that you're going to see some of the major customers Of Anthropic and major customers of OpenAI. I'm talking about the eight and nine figure customers, people spending fifty million, a hundred million a year. They're leaving…”
Friedberg: Near-term Chinese chip manufacturing breakthroughs will surprise the US
“And just like we were knocked over by DeepSeq, I think we will be knocked over by some semiconductor manufacturing processes coming out of China in the near term.”
Sacks: DeepSeek's compute cluster costs over $1 billion
“You add up the cost of a compute cluster with 50,000 plus hoppers, and it's going to be over a billion dollars. So this idea that you've got this scrappy company that did it for only six million, just not true. They have a substantial compute cluster that they…”
Baker: DeepSeek employees have CCP minders and cannot be near Americans
“If you work for deep sea, you are not allowed to be in the same room as an American while you're in China. And you have kind of a CCP minder with you most of the time.”
Brin admits DeepSeek's release closed the gap with proprietary AI models
“Deep Seek released a really surprisingly powerful model when it was January or so. So that, that definitely closed the gap to proprietary models.”
Pichai: Google benchmarked DeepSeek against Flash and Flash was as efficient
“Interestingly for us internally, we benchmarked it to Flash and, you know, Flash was, ah, as efficient or, you know, you could argue in some ways better.”
Calacanis: Grok 3 has eclipsed Gemini, DeepSeek, Claude, and ChatGPT-4o
“Grok III has now eclipsed Gemini, which is Google's LLM, and DeepSeq from China, Quad, and ChatGPT-IVO.”
Naval: DeepSeek gave China the best open-source AI model
“They sent us back the best open source model when our guys were staying closed source for safety reasons.”
Naval: US lost DeepSeek researchers because we sent foreign students home
“You could even argue deep seek. Part of the reason why we lost that is because a bunch of those kids, they studied in the US, but then we sent them back home.”
Palihapitiya: DeepSeek bypassed Nvidia CUDA software lock-in using PTX
“Everybody is used to building models and compiling through CUDA, which is Nvidia's proprietary language, which I've said for a couple of times is their biggest moat, but it's also the biggest threat factor for lock-in. And these guys worked totally around CUDA…”
Sacks: US AI models' final training runs cost tens of millions
“The final training run cost was more in the tens of millions of dollars about nine or 10 months ago. And so, you know, it's not six million versus a billion, okay?”
David Sacks estimates China's AI model gap narrowed to 3-6 months
“If you had asked most people in the industry a few weeks ago, how far behind is China on AI models, they would say six to 12 months. And now I think they might say something more like three to six months, right?”
Chamath Palihapitiya says DeepSeek's chain-of-thought proves its model was distilled
“So the reason why this is happening is like, you're seeing this chain of thought, you're seeing the several layers, and then it's catching it after the fact. So we know that this is distilled from some other model.”
Chamath: DeepSeek's million-dollar model matched OpenAI's billion-dollar o1 model
“We saw a model that got released by the Chinese under, by the way, a totally M you know, open source MIT license that you can run on a laptop. And it turned out to be as good as open AI's O one model. So a couple of versions old, but the point is that This mod…”
Baker: Divergent Chinese open-source model architectures favor Nvidia GPUs
“If you look at the underlying architectures of the three, what I call big Chinese open source models, and maybe even through a few, or four, You know, if we have Quinn, if we have Kimmy, if we have DeepSeq, and then we have GLM, they're actually evolving in ve…”
Gerstner: DeepSeek, Anthropic, and OpenAI will launch Blackwell-trained models in 4–8 weeks
“And just this year, like we're going to see the first models over the course of the next four to eight weeks out of deep seek, out of anthropic, out of open AI that are trained on blackwell servers, right? You're going to see a next generation of models Far mo…”
Sacks: DeepSeek release erased $1 trillion in market cap in one day
“It's rare that a model release is going to be a global news story or Cause a trillion dollars of market cap decline in, in one day.”
Palihapitiya: Smaller $2M seed rounds force AI startups to innovate
“When the AI company wakes up and rolls out of bed and some VC gives them two hundred million dollars, maybe that's not the right answer for a series A or a seed. And maybe the right answer is two million so that they do these deep seek like innovations.”
Palihapitiya: DeepSeek created GRPO algorithm to slash AI memory requirements
“They invented a totally different algorithm. There was the orthodoxy. Right? This thing called PPO that everybody used, and they were like, no, we're going to use something else called, I think it's called GRPO or something. It uses a lot less computer memory,…”