Jensen Huang on AI & accelerated computing
Huang's founding bet is accelerated computing: pairing parallel GPUs with CPUs, because general-purpose chips stopped scaling and the big gains now come from architecture and algorithms. He sees deep learning as a scalable universal function approximator that turns computing from retrieving files into generating answers, and data centers into factories whose output grows with compute. Energy is the real constraint, and he thinks the next decade is about applying AI to science, robotics and everyday knowledge.
16 points · 7 appearances · 2023–2026
How the view evolved
AlexNet revealed a scalable universal function approximator → Agentic scaling: intelligence will scale with compute
In 2023 he described deep learning as a universal function approximator that scales by making networks deeper on more data; by 2026 he talks about four scaling laws, with agentic scaling feeding back into pre-training, and says intelligence will scale with compute.
Up next
2023–2026 · changed view
“And so the question is the what I'm leading up to is ultimately what is the purpose? What is the purpose of the lawyer? And has the purpose changed? What is the purpose? You know, one of the examples that I gave is that I would give is for example if my car became self-driving will all chauffeers be out of jobs? The answer probably is not because for some per for some chauffeers they for some people who are driving you they could be protectors some people they're part of the experience part of the service so when you get there they you know they could take care of things for you and so for a lot of different reasons not all chauffeers would lose their jobs some chauffeers would lose their jobs and many chauffeers would change their jobs and the type of applications of autonomous vehicles will probably increase you know the usage of the technology within find new homes and so I think you have to go back to what is the purpose of a job you know like for example if AI comes along I actually don't believe I'm going to lose my job because my purpose isn't to I have to look at a lot of documents I study a lot of emails I look at a bunch of diagrams you know the question is what is the job and the purpose of somebody probably hasn't changed a lawyer for example help people that probably hasn't changed studying legal documents generating documents it's part of the job not the job.”
2023–2026 · changed view
“When Nvidia started in 1993, we were trying to create this new computing approach. The question is what's the killer app? And the problem we wanted to the company wanted to create a new type of computing pro a computing architecture a computing a new type of computer that can solve problems that normal computers can't solve. Well, the applications that existed in the industry in 1993 are applications that normal computers can solve because if the normal computers can't solve them, why would the application exist? And so, we had a mission statement for a company that has no chance of success.”
2023–2026 · changed view
“I think I know what consciousness is the sense of experience the ability to know self versus the ability to be able to reflect know our own self the sense of ego I think all of all of those human experiences probably is what consciousness is but why it exists versus the concept of knowledge and intelligence which is what AI is defined by today it has knowledge it has intelligence artificial intelligence we don't call it artificial consciousness artificial intelligence the ability to perceive believe, recognize, understand, plan, perform tasks. Those things are foundations of intelligence to know things, knowledge. I don't, it's clearly different than consciousness.”
2023–2026
“That moment in time for our company gave us so much confidence. And the reason for that is you can succeed in doing something, inventing a future, even if you were not informed about it at all. And it's kind of my attitude about everything now. When somebody tells me about something and I've never heard of it before, or if I've heard of it, don't understand how it works at all, my first thought is always, how hard can it be? And it's probably just a textbook away. You're probably one arXiv paper away from figuring this out. And so I spent a lot of time reading arXiv papers, and it's true.”