Macquarie analysts said Tuesday they believe Microsoft (NASDAQ:MSFT) is uniquely positioned “to monetize both Nvidia's (NASDAQ:NVDA) new hardware and the software-layer value” tapped by the company’s latest releases announced at the GTC conference.
“Microsoft, which is one of the first adopting NVIDIA's GB200, will use NVIDIA's Triton Inference Server to help power Microsoft Copilot, and is launching a new Azure SKU offering with access to cloud NVIDIA H100 GPUs,” analysts wrote.
Triton Inference Server is Nvidia’s open-source project designed to simplify the deployment of AI models at scale in production environments.
Macquarie’s remarks come after Nvidia announced a series of major products and services during the AI-focused GTC conference.
Among those is also NIM, a microservices-based software that simplifies AI app development and deployment on NVIDIA hardware, ensuring portability across CUDA-enabled devices and standardizing model access through APIs, all while optimizing inference engine performance.
“NIM is a significant announcement for the generative AI app ecosystem as containerized AI accessible through standardized APIs provides a level of consistency that previously did not exist within any ecosystem,” analysts said.
“We think this portability could be a game changer for locally deployed generative AI apps,” they added.
The chipmaker also included CUDA-X microservices to simplify the development of LLM-based applications, like custom AI copilots.