Amazon’s cloud computing unit, AWS, announced Tuesday that it will integrate key Nvidia technology into its next-generation artificial intelligence computing chips, signaling a major push to attract enterprise AI customers. The company plans to adopt Nvidia’s NVLink Fusion technology in a future chip named Trainium4, though it did not provide a release date. NVLink is designed to create high-speed connections between chips, allowing them to communicate more efficiently—a critical factor in training large AI models that require thousands of interconnected machines.

The announcement came during AWS’s annual cloud computing conference in Las Vegas, which draws roughly 60,000 attendees. Nvidia has been expanding the adoption of NVLink among other chip makers, with Intel and Qualcomm already on board, and AWS now joining the list.

“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 CEO Jensen Huang in a statement.

As part of the partnership, AWS customers will gain access to “AI Factories,” exclusive AI infrastructure housed in their own data centers for faster performance and readiness.

In addition to future plans, Amazon unveiled new servers powered by its current Trainium3 chips. Each server contains 144 chips, offering more than four times the computing power of AWS’s previous AI generation while using 40% less energy, according to Dave Brown, AWS vice president of compute and machine learning services. Although exact performance metrics were not disclosed, Brown emphasized that AWS aims to compete on price and performance against rivals, including Nvidia.

“We’ve got to prove to them that we have a product that gives them the performance they need at the right price point,” Brown said. “That means they can say, ‘Yes, that’s the chip I want to go and use.’”

AWS also introduced upgraded versions of its AI models, branded Nova. Nova 2 promises faster, more responsive outputs and includes a version capable of handling text, images, speech, or video prompts. Another model, Sonic, generates speech outputs in response to speech inputs, which AWS CEO Matt Garman described as “human-like” during his keynote.

Despite these advancements, Amazon faces strong competition from OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini. However, AWS posted a 20% sales increase in its most recent quarter, driven largely by cloud computing and AI infrastructure services.

At the Las Vegas event, Amazon also introduced Nova Forge, a service that allows businesses to create AI models tailored to their proprietary data while retaining the core knowledge of the underlying model. “This allows you to produce a model that deeply understands your information, all without forgetting the core information that the thing has been trained on,” Garman said.

Following the announcements, Amazon shares rose 0.9% to $235.98 by midday Tuesday.