michelangelus - Fotolia
Nvidia partners launch Nvidia-Certified Systems for AI
According to analysts, Nvidia-Certified Systems' launch is a shrewd move for Nvidia, as the vendor looks to cement its lead in the AI hardware marketplace.
Nvidia is teaming up with systems makers to launch the first wave of Nvidia-Certified Systems, servers bearing Nvidia technology and designed to handle AI and data analytics workloads.
Shipping on Tuesday are servers from Dell, Gigabyte, Inspur, Supermicro and Hewlett Packard Enterprise (HPE). According to Nvidia, at least 11 systems makers, with nearly 70 servers, are in the program.
To carry an Nvidia-Certified Systems badge, servers must pass a set of certification tests that tackle a variety of use cases. Currently certified systems contain the latest Mellanox cables, switches and network cards, and up to eight Nvidia A100 GPUs.
A 'shrewd move'
The launch of the systems comes as Nvidia, a leader in AI hardware, faces competition from other big-name vendors, including AMD and Intel, as well as a number of new startups.
"This is a shrewd move by Nvidia to further cement their lead in AI, while other chip vendors such as Intel scramble," said Forrester analyst Mike Gualtieri.
While systems vendors, including Dell and HPE, were already building systems that included Nvidia AI chips, the Nvidia-Certified Systems program "formalizes the trend and will allow it to expand by setting standards for AI systems based on Nvidia," he said.
Gualtieri said the systems will likely draw three types of buyers: academic and research groups that use Nvidia hardware but want to expand; enterprises that want to experiment with AI but don't want to move data into the cloud; and enterprises that currently use the cloud for AI projects but find it too expensive.
According to Nvidia, Nvidia-Certified Systems are tested explicitly on workloads that include deep learning training, AI inference, data science algorithms, security and intelligent video analytics.
Mike GualtieriAnalyst, Forrester
"These types of solutions are a great step for the market, especially for an increasing number of organizations that want more options to build and run machine learning solutions and operations on premises and in their IT environments," said Nick McQuire, head of enterprise and AI research at CCS Insight.
He added that organizations may choose one of these systems to meet its performance, manageability, governance, risk and compliance requirement. This is especially important for "global organizations that have to navigate a host of data residency and processing challenges for [machine learning]," McQuire said.
Pricing for the Nvidia-Certified Systems varies by configuration. Nvidia said it plans to announce more partners and systems in the future.
Nvidia launches integrated AI platform for VMware vSphere 7
Nvidia releases new AI software, hardware
Nvidia partners with Google Cloud to boost AI for 5G app dev