NVIDIA CEO Jensen Huang used his opening speech at Computex 2024 in Taiwan to introduce the next-generation technology and the roadmap the corporate will follow to grow to be the leading AI hardware platform.
The theme of his speech was “Accelerating Everything,” and Huang unveiled latest chips, faster AI networks, and an ambitious development plan. From AI PCs to massive AI factories, NVIDIA goals to be the trailblazer for the longer term of AI.
“The way forward for computing is accelerated,” said Huang. “With our innovations in AI and accelerated computing, we’re pushing the boundaries of what is feasible and driving the following wave of technological advancement.”
Here’s a fast take a look at a few of the announcements:
New chips
The Blackwell GPUs are still fresh off the production line, but Huang says we will expect an improved Blackwell Ultra GPU later this 12 months. He said NVIDIA's development plan is to release a brand new family of chips yearly.
“Our company operates on an annual basis. Our basic philosophy could be very easy: we construct your entire data center, dismantle it and sell you parts of it on an annual basis, and push all the things to the boundaries of technology,” Huang explained.
Huang announced that its Ruby platform will replace Blackwell's GPU family in 2026.
AI factories
Huang named several computer firms that can construct “AI factories” that can deploy cloud, on-premises, embedded and edge AI systems using NVIDIA GPUs and networks.
Huang said: “The next industrial revolution has begun. Companies and countries are working with NVIDIA to remodel traditional trillion-dollar data centers to accelerated computing and construct a brand new sort of data center – AI factories – to provide a brand new product: artificial intelligence.”
NVIDIA MGX (Modular Server Reference Architecture) is a blueprint for constructing data center servers that focuses on accelerated computing. It uses a standardized approach to constructing servers with NVIDIA products, making it much easier to construct these AI factories.
NVIDIA Spectrum X
NVIDIA uses NVLink to maneuver data between GPUs, but moving data between GPU pods inside or between data centers presents an extra challenge.
NVIDIA Spectrum-X is the world's first Ethernet networking platform specifically designed to spice up the performance of AI workloads. Of course, NVIDIA Spectrum-X is optimized for NVIDIA hardware.
NVIDIA says Spectrum-X accelerates the performance of generative AI networks by 1.6x in comparison with traditional Ethernet structures.
Last week we reported that major technology computing firms are working to develop an open, accelerated AI networking standard without NVIDIA's involvement.
Nevertheless, Huang said Spectrum-X is already getting used by firms equivalent to Dell Technologies, Hewlett Packard Enterprise and Lenovo.
NIM
NVIDIA NIM (NVIDIA Inference Microservices) is a set of tools designed to simplify and speed up the deployment of generative AI models.
Before you should utilize an AI model, developers must deploy it. Instead of getting to cope with the main points of this process, NVIDIA NIMs simplify deployment by mechanically packaging AI models into containers and optimizing them for NVIDIA hardware.
It's just like the difference between buying, chopping and assembling all of the ingredients for a meal and putting a ready-made meal within the oven.
Huang announced that the Llama 3 NIM is now available as a free download for developers to check out. Huang also unveiled Nvidia Ace NIMs, which include a digital human model able to generating AI human models with lifelike skin.
You can watch the total keynote address to learn more details about AI RTX-powered PCs and laptops, Nividia's Isaac ROS 3.0 robotics platform, and the way factories are using NVIDIA's technology to create digital twins of their factories to coach robots.
