JHVEPhoto
On Wednesday, AMD (NASDAQ:AMD) will host its long-awaited event focusing on artificial intelligence and formally launch its MI300 data center GPU, placing it deeper in competition with Nvidia (NASDAQ:NVDA). Wedbush Securities is expecting the event to focus on several areas, including performance.
Analyst Matt Bryson, who has an outperform rating and $130 price target on AMD (AMD), said he expects the Dr. Lisa Su-led company to get “more granular” around performance and highlight advantages over Nvidia’s (NVDA) H100 offering, including greater memory and bandwidth and how it impacts real world performance.
“We’d note this latter point would very much fit one of the tenets of our positive thesis on DRAM as not only do higher memory specifications lift overall demand, but newer product ramps (DDR5 and HBM3) built on advanced nodes require increased capital (process steps) pulling sidelined capacity naturally back on-line while not significantly impacting bit output (in part because ECC is integrated into newer technologies),” Bryson wrote in an investor note.
Bryson said AMD (AMD) may also give initial details for its EPYC architecture and how it integrates with its AI roadmap.
Also expected to be a focus of the event are partnerships, with Microsoft (MSFT) likely to have a presence. Microsoft has already said it will integrate MI300 into its Azure cloud.
Bryson also said he expects AMD (AMD) to give some information on how AI will be incorporated into edge devices, particularly PCs, where AMD has a significant presence via its CPUs.
In October, AMD (AMD) said it expects the MI300 to generate more than $2B in sales in 2024.
AMD (AMD) shares fell 0.5% in premarket trading on Tuesday.
