[Computex 2024] NVIDIA’s Vision For AI Future Is “AI Factories”

Kicking off the Computex 2024 week, NVIDIA founder and CEO Jensen Huang announced several key advancements in AI and datacenter segments.

NVIDIA Datacenter Announcements

The keyword here is “AI Factories” – NVIDIA is teaming up with global server OEMs unveiled systems featuring its latest Blackwell GPUs and ARM-based Grace CPUs to build AI factories and data centers for the “next industrial revolution.” Companies like ASRock Rack, ASUS, GIGABYTE, Supermicro, and others will deliver AI systems on cloud, on-premise, embedded and edge environments using NVIDIA GPUs and networking.

The company also introduced the Spectrum-X Ethernet networking platform – which the chipmaker calls “the world’s first Ethernet fabric built for AI.” CoreWeave, GMO Internet Group, Lambda, and others are among the first to adopt the new networking system to massively boost networking performance for their AI infrastructures; the company has also announced that more companies are launching new platforms featuring Spectrum-based products.

Finally, the chipmaker launched NIM (NVIDIA Inference Microservice), which allows the developers around the world to deploy generative AI applications quickly and efficiently via APIs. These can be deployed on cloud, datacenters and workstation environments, which offers an easy way to build generative AI applications such as chatbots, assistants, and more, with significant developmental time savings.

Related Article

$57,000 COVID Relief Fund Spent On Pokemon — Man Hunted Down By Government!

This is Why Kingdom Hearts 3 Took So Long to Develop

GIGABYTE Z590 AORUS PRO AX Review — great on paper, but not so much in practice…

Pokdepinion: One takeaway I had from the keynote is “tokens” – and AI factories are designed to products lots of tokens (which is what GenAI uses).