d-Matrix plugs AI inference chips into Nvidia’s ecosystem
d-Matrix will use Nvidia's NVLink Fusion technology to integrate its inference processors into Nvidia-based AI systems.
AI infrastructure is becoming more modular as specialised accelerators look for ways to coexist with Nvidia rather than replace it entirely.
What happened
d-Matrix will adopt Nvidia’s NVLink Fusion technology so its Raptor inference processors can connect directly into Nvidia-based data-centre systems. Integrated racks are expected in 2027.
Why it matters
Specialised inference chips can target workloads where cost and efficiency matter more than the flexibility of general-purpose GPUs. Compatibility with Nvidia infrastructure lowers the barrier to adoption.
The bigger picture
Nvidia’s ecosystem may become more valuable even when some compute moves onto competing accelerators. The company can remain central by owning the interconnect and system architecture around a broader set of chips.