Intel’s Upcoming “Low-Power” AI GPU Development with Jaguar Shores Set to Launch Next Year

Intel appears to be diversifying its AI chip offerings, as a recent report indicates the company is developing a new GPU designed to operate with low power requirements.

Intel’s ‘Low-Power’ AI Chip: Targeting Inference Workloads with a Strategy Like Qualcomm

Historically, Intel has struggled to gain a foothold in the competitive AI landscape, facing challenges due to a relatively limited portfolio compared to its rivals. The company has depended heavily on its Gaudi series over recent years; however, reports suggest that market adoption for these products has not gained significant momentum. A recent CRN report reveals that Intel is planning to introduce an AI chip characterized by its low power consumption alongside the upcoming Jaguar Shores GPU, which is aimed primarily at the inference market.

Outside of Jaguar Shores, the company has an additional, unannounced GPU design with a lower power requirement for servers on its road map that could arrive next year at some point, according to a source familiar with the company’s plans.

Details regarding Intel’s new GPU remain largely under wraps, with the most notable point being that this unannounced design will launch concurrently with Jaguar Shores. Given Jaguar Shores’ positioning as a high-end product primarily for training applications—which may also target rack-scale solutions—it’s reasonable to assume that the low-power GPU will be engineered for inference workloads.

One plausible speculation is that this “unannounced”design could be based on Intel’s Battlemage architecture, targeting Edge AI applications. Intel has previously indicated plans to develop such a model. Notably, the company is reportedly working on a ‘BMG-G31’ silicon variant that may include up to 24 GB of VRAM. While this could potentially be the direction of the new chip, the expected timeline for its release raises questions about the future relevance of the Battlemage design. Thus, it may evolve into a different series altogether, but for now, these insights remain speculative.

As Intel continues to explore low-power AI solutions, it may carve out a market segment similar to Qualcomm’s Cloud AI 100 inference cards, which are designed for seamless deployment in consumer systems.

Source & Images