The collaboration will provide close technical collaboration and rapid iteration with the aim being to combine ZeroPoint's memory optimisation IP with Rebellions AI accelerator technology.
Critically, given the explosive demand for generative AI workloads, the companies said that they will move rapidly to unveil new products in 2026 with the intention of delivering unprecedented tokens-per-second-per-watt performance, setting the blueprint for more sustainable AI data centres in the future.
Specifically, as part of this collaboration, the two companies will focus on:
- Increasing effective memory bandwidth and capacity for foundational model inference workflows
- Enhancing tokens-per-second-per-watt without sacrificing accuracy, using lossless model compression to reduce model size and energy required to move model components
Commenting on the announcement, Sunghyun Park, CEO of Rebellions, said, "Our partnership with ZeroPoint enables us to redefine what's possible in inference performance per watt - delivering smarter, leaner, and more sustainable AI infrastructure for the generative AI era."
"The partnership with Rebellions is a powerful validation of the transformative impact of ZeroPoint's hardware-accelerated memory optimisation technologies," added Klas Moreau, CEO of ZeroPoint. "We are convinced that memory acceleration will rapidly evolve from a competitive edge to an indispensable component of every advanced Inference accelerator solution, and we believe that Rebellions shares our commitment to making AI datacentres far more efficient."
