The cloud-computing division of Amazon is making a strategic move that may redefine the landscape of artificial intelligence. The company announced Tuesday that it would use Nvidia Corp.’s breakthrough chip-linking technology in its next-generation AI processors, in one of the most significant partnerships yet between two technology giants vying to control a fast-growing market for AI.
AWS plans to include Nvidia’s “NVLink Fusion” technology in its next-generation Trainium4 chip, although the company has not announced a release date for the new processor. The technology is thought to be one of Nvidia’s most valuable inventions, tying together different kinds of chips at ultra-fast speeds.
The capability is becoming increasingly important as companies race to develop more powerful AI systems.
The announcement came during AWS’s annual re:Invent conference in Las Vegas, which attracts approximately 60,000 attendees each year. The event serves as a prime platform for Amazon to showcase its latest cloud computing and AI innovations before customers and partners.
NVLink technology addresses one of the biggest challenges in modern AI development: getting thousands of machines to work together seamlessly.
Amazon Adopts Nvidia’s NVLink for High-Speed AI Infrastructure
When training large AI models, such as those powering ChatGPT or other advanced systems, a company needs to connect massive numbers of chips that can communicate with each other at lightning speed. Nvidia’s technology makes this possible, and Amazon’s adoption signals just how critical these high-speed connections have become.
Nvidia has been aggressively courting other chipmakers to take up NVLink, and the strategy seems to be paying dividends. Intel and Qualcomm have already signed on, and now Amazon joins the growing ranks. As part of the deal, AWS customers will have access to what the company refers to as “AI Factories”-that is, dedicated AI infrastructure hosted in their own data centers for greater speed and performance.
“Together, Nvidia and AWS are creating the compute fabric for the AI industrial revolution – bringing advanced AI to every company, in every country, and accelerating the world’s path to intelligence,” said Nvidia chief executive Jensen Huang in a statement announcing the collaboration.
In addition to the future Trainium4 chip, Amazon is also introducing new servers based on its current generation Trainium3 processor. The new servers, which start shipping Tuesday, offer 144 chips apiece and are more than four times as powerful as AWS’s previous generation while using 40% less energy.
Dave Brown, vice president of AWS compute and machine learning services, said Amazon’s competitive strategy relies on offering better price-performance ratios than competitors, including Nvidia itself.
“We’ve got to prove to them that we have a product that gives them the performance that they need and get a right price point so they get that price-performance benefit,” Brown said. “That means they can say, ‘Hey, yeah, that’s the chip I want to go and use’.”
Amazon Ups Its AI Game with Nova 2, Sonic, and Nova Forge; AWS Sales Surge 20%
Amazon also leveraged the conference to announce its updated Nova AI models. Nova 2 promises faster and more responsive performance and includes a versatile version that can accept text, images, speech, or video as input and respond accordingly. Another variant dubbed Sonic can engage in speech-to-speech conversations that AWS CEO Matt Garman, during his keynote presentation, explained is “human-like.”
While Amazon’s Nova models haven’t received the same level of attention as competitive offerings like OpenAI’s ChatGPT, Anthropic’s Claude, or Google’s Gemini, its overall cloud business remains healthy. AWS saw sales increase by 20% in its most recent quarter, with the performance driven mainly by strong demand for cloud computing and AI infrastructure services.
The company also introduced Nova Forge, a new service for helping businesses build custom AI models from their own data. “That allows you to produce a model that deeply understands your information, all without forgetting the core information that the thing’s been trained on,” Garman told the conference audience.
These announcements illustrate Amazon’s larger strategy of positioning itself as an end-to-end AI provider, offering not only the infrastructure on which to run AI systems but also the tools for building and tailoring applications. As the competition in AI heats up across the technology industry, partnerships such as the one inked with Nvidia may play a decisive role in determining which companies emerge as leaders in this transformative technology.
Amazon shares rose in response, up 0.9% at $235.98 midday Tuesday, as investors saw these moves as enhancing its competitive position in the fast-growing AI market.
