Nvidia’s annual shareholder meeting is this upcoming Wednesday, but the company has already released its annual report:
It’s worth a read even if you’re non-technical. CEO Huang’s letter starts on page 22. Some highlights:
Two transitions are occurring simultaneously—accelerated computing and generative AI—transforming the computer industry and every other industry worldwide.
I think it’s important to understand the differences in these two computing models:
• Generative AI - is the ChatGPT like “Large Language Models” (LLM) that everyone is all excited about.
• “Accelerated Computing” - is a phrase that Nvidia has coined, basically pitching that CPU scaling (aka Moore’s law) has slowed and isn’t keeping pace with the demands of today’s software. This is not AI stuff, but more regular data center computing.
Accelerated computing will modernize the world’s trillion-dollar data center infrastructure. Its increased throughput and cost and energy savings are remarkable. Two NVIDIA HGX computers with Hopper GPUs that together cost $500K can replace 1,000 nodes of CPU servers that cost $10M for AI, scientific computing, or data processing workloads.
Huang goes on to pitch this as ideal both for computing centers on a budget as well as for computing centers looking for the highest throughput.
The letter is several pages long and covers a lot of ground. One thing caught my eye:
The ChatGPT Moment for Robotics Is Coming
NVIDIA has dedicated nearly a decade to robotics AI and is excited to see it all come together. In time, humanoid and task-specific robots will be an industry larger than the auto and consumer electronics industries combined.
Anyway, if you want to know more about what Nvidia does and all the different avenues its pursuing, reading Huang’s letter is a good and deep idea.
I read Huang’s comments, and it is enough to make your head spin. I am not knowledgeable enough to discuss any of it, and while I hate to say it, I’m going to take him at his word (at least until results show otherwise).
One small section I found interesting was his comments on automotive. Of the examples he gave, all of them are Chinese:
DRIVE AGX Orin also powers cars from
Volvo, Polestar, Lucid, Great Wall Motor,
SAIC, ZEEKR, NIO, Xiaomi, and XPENG.
In addition to helping vehicle-makers
develop their AI-defined computers
in their car and build a safe operating
system for autonomous vehicles
Tesla (and I suppose GM and other Detroit makers) are the ones I generally think of when the topic of electric/self driving etc cars comes up. Could it be that China is about to steal their thunder? It wouldn’t surprise me in the least.
It’s happening everywhere, but here, due to tariffs and other barriers. BYD is already the largest EV manufacturer in the world by car volume since the last quarter of 2023.
And in the US, a lot of the innovation that is leverage AI is being lead by companies like Rivian. Their second gen cars (model year being sold now) have a computer 10X more powerful than the gen1 platform (still on sale). Comparing to a home computer, it’s somewhat a computational power of 7 or 8 top-end desktop computers. Autonomous driving (AI driven by definition) will the biggest usage of these on-board computers, followed by ever-growing complex control modules (an entire suspension feel defined by software, a full reality already). And I believe every car manufacturer will follow because it’s their big chance to not fall determinately behind Chinese manufacturers in a few years. Invest heavy now while the US government has NVidia on a leash, keeping their exports of high-end AI chips to China at very low levels. China will eventually circumvent that, but I see them at a technology disadvantage now. But it will not last.