NVIDIA's upcoming graphics architecture has surfaced in Linux kernel patches under the unexpected name "Blackwell-Next." The latest code submissions include support for Compute Express Link (CXL), suggesting the company is actively developing this next-generation GPU platform. These patches reveal technical groundwork ahead of any public launch.

The naming creates confusion since NVIDIA has already confirmed its next-gen architecture will be called "Rubin," not "Blackwell-Next." We've seen "Blackwell Ultra" appear in the B300 series, but the mainline successor carries a different public codename. Both consumer GeForce RTX cards and server AI accelerators are slated to use the Rubin architecture, making this patch naming inconsistency notable.

NVIDIA reportedly plans to pair its RTX Spark SoC with a "Vera" CPU and "Rubin" GPU for future products. The Blackwell-Next reference appears isolated, potentially indicating internal development terminology or an engineering placeholder. Either way, these kernel updates signal that Rubin GPU development is advancing toward eventual release.

Linux kernel integration typically precedes hardware launches by months, suggesting Rubin-based products could arrive sooner than expected. The CXL support hints at improved connectivity for AI and compute workloads.