Intel Remains Committed to Desktop GPUs: New Battlemage Models Appear in Shipping Records

Intel remains committed to making significant strides in the discrete GPU arena, as newly uncovered shipping manifests hint at the ongoing development of a new Battlemage model and a cryptic SKU.

Intel has resumed development on the Battlemage “BMG-G31” variant, which seems to be tailored for AI and professional applications. Last year, Intel made waves by introducing two new Battlemage GPUs, the Arc B580 and B570. Interest skyrocketed back then because NVIDIA and AMD’s next-generation solutions weren’t yet available. With the second quarter nearly over and Intel silent about its desktop GPU strategy, some speculated that the company had lost its momentum. However, recent shipping manifests tell a different story.

A discovery in the manifests points to the “BMG-G31,” which boasts more advanced silicon than current Battlemage models. Reports suggest this variant would have around 24-32 Xe2 cores, a 256-bit memory bus, and 16 GB of GDDR6 memory. Rumors circulated online about the BMG-G31’s potential cancellation, implying Intel might not release this model for consumers. The manifests, however, list the BMG-G31 under “R&D purposes,” suggesting its design is geared toward AI or professional workloads.

Intriguingly, another manifest references a peculiar variant, the BMG “C32.” Its mixed-up naming stirs confusion since it combines Battlemage and Celestial designations. Nonetheless, it’s said to belong to the Battlemage series, possibly as a “reworked” version of the “BMG-G31,” according to @mikdt. Clearly, Intel is working on multiple Battlemage variants, but it’s uncertain whether they are intended for the desktop sector.

It’ll be fascinating to observe Intel’s strategy for its discrete GPUs, especially since the company’s working on the Xe3 “Celestial” architecture for the upcoming Panther Lake SoCs. With Intel’s new CEO at the helm and an apparent pivot towards AI, it seems likely that the company may prioritize GPUs targeting AI-heavy workloads.