Nvidia’s Jensen Huang breaks down his ‘CEO math’

• Nvidia’s Jensen Huang explains “CEO Math” ahead of Computex in Taiwan.
• By investing in GPUs and CPUs, companies can significantly reduce the time spent on AI tasks, Huang said.
• “The more you buy, the more you save,” he said.

There’s math, and then there’s what Nvidia Jensen Huang He calls it “CEO Math.”

“The more you buy, the more you save,” Huang said He said Before the annual Computex technology fair held in Taiwan. “This is called CEO accounting. It’s not accurate, but it’s true.”

Confused?

Huang explained this concept by describing why companies should invest in both graphics processing units (GPUs) and central processing units (CPUs). The two processors can work independently, reducing the time it takes to execute a task from “100 time units to 1,” he said.

So, the more you buy, the more you save. Sounds like a good sales pitch for a CEO selling processors.

Combining processors is already a common practice in the personal computing industry. “We add a GPU, a $500 GPU, to a $1,000 computer, and the performance increases dramatically,” he said. “We’re doing it in a data center. A billion-dollar data center, we add $500 million worth of GPUs, and all of a sudden, it’s an AI factory.”

Huang then presented a diagram showing that when companies used both, their speed would increase by 100, just 1.5 times the cost.

in March, Nvidia has unveiled the Blackwell B200 graphics processing unit, a $70,000 chip that claims to be “the most powerful AI chip in the world.” It packages the chip into larger designs like the GB200 NVL72, which combines 72 GPUs and 36 CPUs and is intended for “the most compute-intensive workloads” and reduces cost and power consumption by up to 25 times.

Over the past few months, the chipmaker has made headlines as a crucial player in the AI ​​boom. He. She It generated more than $22 billion in the fourth quarter of 2023. Technology executives from Sam Altman to Mark Zuckerberg have come to rely on its chips to advance their AI ambitions.