Intel continues to push forward in the discrete GPU space, undeterred by the competition. There are new developments brewing with their Battlemage series, and a new mysterious SKU has surfaced.
Intel appeared to shake things up last year with the introduction of two fresh Battlemage GPUs, the Arc B580 and B570. These models grabbed considerable attention, mostly because NVIDIA and AMD hadn’t yet launched their next-generation offerings. Fast forward to the current scenario—Q2 is almost over, and Intel has been noticeably silent about their plans for desktop GPUs. This has led some to believe that Intel’s ambitions might be waning. But fresh insights from shipping documents suggest otherwise.
Recently uncovered shipping manifests reveal the presence of a potentially more advanced silicon, named “BMG-G31,” compared to current Battlemage models. This variant is speculated to be equipped with 24 to 32 Xe2 cores, a 256-bit memory bus, and 16 GB of GDDR6 memory. Despite this, rumors have circulated online hinting at the possible cancellation of the BMG-G31, suggesting that Intel might not aim for the consumer market with this product. The “R&D purpose” label in the shipment records further indicates that this model might be geared toward AI or professional workloads.
Adding intrigue, there’s talk of another variant, tagged BMG “C32.” The naming is puzzling, merging the initials of Battlemage and Celestial. However, it’s said to belong to the Battlemage range and may be a “reworked” iteration of the BMG-G31, as noted by @mikdt. This suggests Intel is developing multiple Battlemage versions, though their ultimate destination—whether for desktops or other purposes—remains anyone’s guess.
With Intel now gearing up for its Xe3 “Celestial” architecture, destined for Panther Lake SoCs, the company’s trajectory in the GPU segment is keenly watched. The new leadership at Intel is prioritizing AI, so it’s plausible that their next wave of GPUs will be tuned for AI-focused tasks. It will be fascinating to see how Intel carves out its path in the evolving landscape of discrete GPUs.
