Before COMPUTEX 2024 officially starts on June 4, NVIDIA held a keynote on June 2 to discuss several key advancements and innovations. NVIDIA CEO and Founder Jensen Huang emhpasized a lot about AI, and revealed in his 2-hour keynote all of the advancements they have made. Here’s a quick recap of all the things te talked about on stage:
AI Factories will soon be a thing. NVIDIA is arguably the leader when it comes to designing key architecture for generative AI systems, and Huang announced that it is partnering with several key computer brands like ASUS, Gigabyte, ASRock, and more to build AI factories and data centers for the “next industrial revolution” using NVIDIA’s Blackwell-powered systems that feature Grace CPUs. Part of this more include the use of NVIDIA’s MGX platform, which allow for building systems of different sizes and configurations.
NVIDIA has the tools to help developers become “Generative AI Developers”. Huang announced the availability of NVIDIA NIM to more developers to help them roll out Generative AI applications more efficiently–NVIDIA claims that its NIM platform can help deploy Generative AI applications in just minutes compared to weeks on more conventional platforms. Aside from being part of its AI enterprise platform, NVIDIA announced that members of its Developer Program will get free access to NIM starting next month for research, development, and testing.
NVIDIA also has tools to improve AI network performance. Aside from its Blackwell infrastucture, NVIDIA also revealed Spectrum-X, which is “the world’s first Ethernet fabric built for AI”. This new hardware aims to improve overall generative AI network performance by as much as 1.6x to help generative AI clouds make their workflows more efficient compared to traditional Ethernet fabrics.
More RTX AI Laptops are coming. As we are in the age of AI PCs, NVIDIA confirmed that it will be expanding its AI expertise to laptops that have its RTX GPUs. The initial batch of AI-ready RTX laptops include the TUF A14/A16, Zephyrus G16, and ProArt PX13/P16 from ASUS and Stealth A16 AI from MSI, along with more coming from other brands.
Guidelines for SFF GPUs and systems revealed. NVIDIA revealed its guidelines for GPUs that fit under the small form factor (SFF) system, accommodating compact builds to be equipped with NVIDIA’s powerful RTX 40 series GPUs. Among the requirements include a 2.5 slot size (50mm), a length of 304mm, a width of 151mm, and at least an RTX4070 or higher GPU. ASUS is one of the first brands to have SFF-ready RTX GPUs, and expect more brands to have their own offerings soon.
