Kalray and Bull partner on high speed AI data networks
TL;DR
On July 28, 2026, French chipmaker Kalray and Atos subsidiary Bull announced a strategic partnership to co develop next generation high speed networking technologies for AI and high performance computing infrastructures. The collaboration aims to improve data movement and interconnects for large scale AI training and inference systems.
About this summary
This article aggregates reporting from 2 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Read next
Race to AGI Analysis
This partnership is about plumbing, not flashy demos, but plumbing is where a lot of AI advantage hides. Training and serving frontier‑class models is increasingly constrained by memory bandwidth and interconnect performance, not just FLOPs. Kalray already plays in data‑centric processing, and Bull brings deep experience with BXI and other European interconnects for exascale systems. Combining those capabilities to build AI‑optimized networks is a logical next step.
For the AGI race, better interconnects effectively enlarge the practical compute envelope for everyone who can afford this hardware. That can shorten training times, make larger context windows viable and enable more complex agent simulations. The strategic angle is that this is a Europe‑centric effort: a French chip company and a European HPC integrator trying to carve out space in a segment dominated by Nvidia’s NVLink and InfiniBand ecosystem. If they deliver a competitive alternative, it slightly reduces Europe’s dependence on US networking stacks for sovereign AI compute.
