Amazon's Jassy touts custom AI chips, NVIDIA stock dips pre-open

NVIDIA's (NASDAQ:NVDA) stock is modestly lower in pre-open trading Thursday, which in part could be related to comments from Amazon.com (NASDAQ:AMZN) CEO Andy Jassy on AI chips in his 2023 annual letter to shareholders today.

At last check, NVIDIA stock was down 0.5% to $866.03.  NVIDIA stock is down 10.6% from its March high but remains up 75.6% year-to-date and 215.6% from last year.

What did Andy Jassy say about NVIDIA?

In the letter, Jassy said that while they continue to offer Nvidia chips, he touted Amazon's custom AI chips amid supply issues and price performance.

"To date, virtually all the leading FMs have been trained on Nvidia chips, and we continue to offer the broadest collection of Nvidia instances of any provider," Jassy commented in the letter. "That said, supply has been scarce and cost remains an issue as customers scale their models and applications. Customers have asked us to push the envelope on price-performance for AI chips, just as we have with Graviton for generalized CPU chips. As a result, we’ve built custom AI training chips (named Trainium) and inference chips (named Inferentia). In 2023, we announced second versions of our Trainium and Inferentia chips, which are both meaningfully more price-performant than their first versions and other alternatives. This past fall, leading FM-maker, Anthropic, announced it would use Trainium and Inferentia to build, train, and deploy its future FMs. We already have several customers using our AI chips, including Anthropic, Airbnb, Hugging Face, Qualtrics, Ricoh, and Snap."

What is Amazon doing in AI?

Jassy highlighted that while the early public focus has primarily been on GenAI applications, particularly ChatGPT, there are three significant layers in the GenAI stack, each of considerable size. He said Amazon is heavily investing in all of them.

At the base of the GenAI stack, developers and companies focus on building foundation models (FMs), supported by compute power and software tools. Nvidia chips dominate but efforts to improve price-performance include custom AI chips like Trainium and Inferentia. Amazon SageMaker simplifies model development. In the middle layer, Amazon Bedrock provides a managed service for leveraging FMs and cloud features. At the topmost layer, Amazon leads in creating diverse GenAI applications, such as Rufus and Amazon Q. AWS aims to democratize AI, prioritizing security.