By now you’re probably aware of the incredible success Nvidia (NASDAQ: NVDA) has had following its introduction of the H100 processor, a chip designed to power computers running resource-intensive AI programs.

The release of the extremely in-demand hardware was perfectly timed with the AI-boom following the introduction of ChatGPT last November. One H100 could sell for as much as $40,000, if you’re lucky enough to find one.

So what do you do after you’ve tripled your stock price and become only the seventh trillion-dollar company ever on the back of the most powerful chip on the market?

You make an even more powerful one, of course.

On Tuesday, Nvidia announced the upcoming release of the Grace Hopper Superchip.

A combination graphics chip and processor, the Superchip (GH200) will feature High-Bandwidth Memory 3 (HBM3e), which can access information at 5 terabytes per second.

Scheduled to go into production around the second quarter of next year, the GH200 will put further pressure on Nvidia’s competitors. Most notably Advanced Micro Devices (NASDAQ: AMD) and Intel (NASDAQ: INTC).

While both chip makers have seen significant boosts from the AI trend, neither company has offered anything as tailor-made for the “fad” as the H100.

This is the main reason Nvidia stands so far above its competitors in its recent growth and has them estimated as controlling 80% of the AI chip market share:

If Nvidia’s first attempt at an AI-specific chip was able to create this kind of separation from its competitors, there will be a lot of work to do in order to keep them from dominating the field with the release of the GH200.

Chipping Away

Are Nvidia’s closest competitors actually competing, or are they just benefiting from the demand that Nvidia’s production limitations are unable to capitalize off of?

Until Tuesday’s announcement, it seemed like AMD may have been ready to step into the ring with The AI Chip Champ.

With two versions of the MI3000 chip set for release later in 2023, AMD was set to offer hardware that would top the H100 in many ways.

Featuring more memory, faster performance, and significantly less power consumption, it looked as though AMD had an opportunity to offer a superior product while demand was white hot.

While the MI3000 chips will beat the GH200 to market by several months, it does appear that Nvidia’s lead on the field is significant and they’re not resting on their laurels.

The GH200 will have a much larger memory capacity, allowing for larger AI models to fit onto one system. Many supercomputers are currently using as many as eight H100s to run AI applications.

Intel, who has fallen even further behind Nvidia, don’t plan to release their first AI-chip until 2025.

While Intel’s “Falcon Shores” chip boasts specs that top both the H100 and the MI3000, it is likely that both Nvidia and AMD will have even further refined their offerings.

If other manufacturers manage to leapfrog Nvidia’s current-model AI-chip temporarily, it still seems as though the current leader remains in the driver’s seat. 

With that, we’d like to hear your thoughts. Do you think Nvidia will be able to maintain its lead in AI hardware? Will other companies step in and provide enough alternatives to balance out the market? Share your thoughts about this, or anything at feedback@technologyprofits.com.