During the CadenceLIVE Silicon Valley 2024 conference held on April 17th at the Santa Clara Convention Center in Silicon Valley, USA, NVIDIA CEO Jensen Huang engaged in an in-depth conversation with Cadence CEO and President Anirudh Devgan. Cadence, formed from the merger of SDASystems and ECAD in 1988, is the world’s largest provider of electronic design automation, semiconductor technology solutions, and design services.

This cutting-edge dialogue covered multiple topics, including accelerated computing, the future development of Artificial Intelligence (AI), and energy consumption issues. Jensen Huang believes that AI will have a revolutionary impact in three areas: data centers, robotics/autonomous driving, and life sciences, with humanoid robots potentially costing as low as $10,000 to $20,000 in the future.

Regarding the widely discussed issue of AI energy consumption, he stated that although AI consumes a significant amount of computational power, it will fundamentally change the way we address climate change, helping to use less energy and improve energy efficiency, among other benefits.

Without Accelerated Computing, Generative AI Will Be Difficult to Achieve

In a conversation, Jensen Huang emphasized the importance of accelerated computing for the development of artificial intelligence. He used the example of the many benefits that accelerated computing has brought to Cadence's digital twin platform Millennium, and argued that once accelerated computing is adopted, generative artificial intelligence will become a reality. Without the transition to accelerated computing, generative artificial intelligence will be difficult to achieve.

He said that accelerated computing is not the same as general computing. In general computing, you can create a processor that will run all code, which is definitely not the case with accelerated computing. According to him, accelerated computing can bring a 1000x X factor, and there is another 30x X factor on top of that. And if you add generative AI to that, there's another 100,000x factor on top of that. He mentioned that design tools typically only do one processing pass, but designers actually need to do multiple explorations to find the best solution in a multi-dimensional and multi-modal sense. And AI will help us explore and optimize in specific domains.

Huang said he found that a small portion of the code in a program represents the majority of the tool's runtime. For example, CFD (Computational Fluid Dynamics), it may only use 3% of the code, representing 99.9% of the runtime, while the remaining 97% of the code can be rewritten using AI and accelerated computing, resulting in a 100,000x speedup of the application.

As the coiner of the term "accelerated computing," he said: "If we don't turn to accelerated computing, if we don't turn to AI, the computer industry may experience an anti-Moore's Law, for a very clear reason: the amount of work and computing we do is growing, but the rate at which CPUs are scaling is slowing down, so our computing costs will go up, not down."

Humanoid Robots Could Cost as Low as $10,000 to $20,000 in the Future

When asked about which industries NVIDIA is involved in that he is very excited about in the short or medium term, Huang expressed great interest in three industries: data centers/computing, robotics/automation, and life sciences.

Discussing the field of robotics/automation, he said that such systems, whether cars or trucks, pizza delivery robots or humanoid articulated robots, have a lot in common: they all need a lot of sensors, and more importantly, they need functional safety. It is very important that the way computers are designed and verified requires that the operating system is not a normal type of operating system.

He believes that AI is used very widely, and these systems will be connected to the cloud and the data center at any time, so that it can update the experience, report faults and new situations, and then download new models. "You could say that I like the whole field of automation, it's a whole new category."

Huang mentioned in the conversation that the cost of manufacturing humanoid robots may be much lower than people expect. "You can buy a car for $10,000 to $20,000, why can't there be a humanoid robot for $10,000 to $20,000? Robots are very likely to be more agile and functional than humans in an environment designed for humans."

On the topic of life sciences, he said he hopes to turn biology into an engineering field. The process of scientific discovery is very important, but it is fragmented.

He believes that in any case, digital biology will undergo a comprehensive revival, science and engineering are becoming increasingly closer, and this is a very complex field. "Obviously, we don't talk about the Schrodinger equation in chip design, because we change transistors until we can avoid the Schrodinger equation. But in biology, obviously the Schrodinger equation is necessary. So we have a lot of things to innovate, for the first time we have the necessary tools, computing systems and algorithms to help us deal with very large and very complex systems, the integration of data-driven methods with the principled simulation methods you mentioned before may give us an opportunity."

He emphasized in the conversation, "I think the market size of these three industries will be very large, and the market size of humanoid robots alone is large enough."

"Energy Black Hole"? AI Will Completely Change the Way People Deal with Climate Change

On the issue of AI's energy consumption, Jensen Huang admits, "Accelerated computing consumes a lot of energy because there are so many integrated computers."

However, he also said that any optimization of power utilization will directly translate into higher performance, which can be measured, because higher work efficiency will generate more revenue, or directly translate into the cost savings of buying smaller things for the same performance.

"AI can actually help people save energy." He gave the example that a model training investment will benefit millions of engineers like him, and hundreds of millions of people will enjoy the cost savings in the next few decades. The cost savings and energy consumption should be considered vertically from the entire span. He believes that vertically, AI will completely change the way people deal with climate change.