Recommended: Listen to this story (00:24 - 05:49):

The Story: Nvidia, the world’s largest semiconductor chip company, just announced its fastest chip to date.

Earlier this week, the world leader in artificial intelligence computing unveiled the H200, its newest chip, which is designed for training and deploying generative AI models. It's the latest upgrade of Nvidia’s H100, a chip released about a year ago that powers ChatGPT and costs anywhere between $25,000 and $40,000, according to Raymond James.

Nvidia says the H200 will outpace the H100 by nearly twice the speed. Shipments are expected to be available in Q2 of next year.

AI companies will be happy to hear that the H100 and H200 are compatible with each other, meaning that companies using the previous model won't require server or software modifications for the new version.

Although the H200 is a mighty upgrade to computing power potential, Nvidia doesn’t even plan for it to be the company’s best product for much longer. In October, Nvidia shared that it will be releasing its B100 chip sometime next year, a chip that will double the performance of even the H200.

Expert Take: Tony Wang, managing partner at 500 Global, believes the new chip will help to induce a “second wave” of AI.

He says, “The underlying infrastructure of GPUs is now improving much faster than anyone would have expected. The A100s came out in 2020, H100s in 2021, and now we have the H200s. Even though that’s sort of like the new thing, the A100s are still in high demand. It’s almost like having an iPhone 15 announcement, but the iPhone 12s are still flying off the shelves. So we believe that this will enable a second wave that will be a lot more profound than what we’ve even seen already.”

Wang explains that due to the limited and expensive nature of Nvidia’s chips, only the richest and most connected teams get to use them:

*Stay informed about the three biggest stories in venture capital and tech news every weekday morning. 2-min reads only.