Nvidia founder and CEO Jensen Huang said today that the company made real business decisions in 2018 that few people understood would define its future and help define a changing industry. Of course, it's paying off big time, but Huang says it's just the beginning of an AI-powered near future — powered by Nvidia hardware. Is this successful bet luck or wisdom? The answer, it seems, is "yes".He discussed these ideas during a keynote speech at SIGGRAPH in Los Angeles. The watershed moment five years ago, Huang said, was the choice to adopt AI-powered image processing in the form of light hunting and smart enhancement: RTX and DLSS, respectively. (The statements from my notes may not be correct, and some minor corrections may be made after verification of the text.)"We found that rasterization is reaching its limits," he said, referring to the traditional and widely used method for making 3D movies. "2018 is the time of 'betting on the business'. We must reinvent hardware, software, and algorithms. And while we're using AI to adapt CG, we're adapting GPUs for AI. Although ray tracing and DLSS are still being adopted in the diverse and complex world of GPUs and consumer games, their architecture is designed to make them a good partner for the machine development community and - grow up. The large amount of computing needed to train increasingly productive models is best handled not by traditional GPU-powered data centers, but by systems like the H100 designed from the ground up to handle the tasks is important. It would be fair to say that the development of AI is in some ways only limited by access to these resources. Nvidia has successfully sold Beanie Baby as many servers and workstations as it can.
But Huang says it's just the beginning. Not only are new models being purchased, but millions, if not billions, of users are instantly killed. "The future is LLM in front of almost everything: 'Human' is the new programming language," he said. Everything from visual effects to digitizing marketing, industrial design, and heavy industry will get a natural language interface to some extent, Huang has a risk. "All companies will be defined by software and robots, and the cars they build will be built by themselves. Therefore, it is a robot that works with robots," he said. said. Some may not share his opinion, which, while good, is also very good for Nvidia's interests. But although the degree of dependence on the LLM may be unknown, few would argue that it should not be accepted at all, and even the security strategy of the person who will use it and what will require a serious investment in many new computing resources. Investing millions of dollars in next-gen computing equipment like CPU-focused racks is silly when something like the GH200, the recently unveiled AI development hardware for data centers, can do the same thing. wanted.
He happily released a video showing a LEGO-like assembly of several Grace Hopper computing units and blades, then a rack, then a line of GH200s, all connected at a very high speed and they are "one of the largest GPUs in the world.",” which has a full exaflop of unique ML computing power. "It's a perfect size, though," he said, expressing a sense of wonder in the middle of the encounter. "He's probably running Crysis."These will form a key part of the AI-driven digital industry of the future, he suggested. “I don't know who said it, but... the more you buy, the more you save. If I could ask you to remember one thing from my speech today, it would be this," he said, making the audience here at SIGGRAPH laugh. There is no mention of the many challenges of AI, the law, or the whole idea of the AI revolution - as has been the case many times over the past year. It is a rose-tinted view of the world, of course, but when you sell picks and shovels during a gold rush, you can afford to think that way.
“We realized rasterization was reaching its limits,” he said, referring to the traditional, widely used method of rendering a 3D scene. “2018 was a ‘bet the company’ moment. It required that we reinvent the hardware, the software, the algorithms. And while we were reinventing CG with AI, we were reinventing the GPU for AI.”“The future is an LLM at the front of just about everything: “Human” is the new programming language,” he said. Everything from visual effects to a rapidly digitizing manufacturing market, factory design, and heavy industry will adopt in some degree a natural language interface, Huang hazarded. “Entire factories will be software-defined and robotic, and the cars they’ll be building will themselves be robotic. So it’s robotically designed robots building robots,” he said. “This is real size, by the way,” he said, standing for dramatic effect at the center of the visualization. “And it probably even runs Crysis.”
