Marvell Demonstrates Photonic Module Placing HBM 50 Meters From GPU
Memory and compute can now scale on separate axes, which could reshape how AI datacentres provision capacity and ease HBM supply bottlenecks. Infrastructure engineers, chip architects and cloud buyers should track Marvell's progress against rivals like NVIDIA's NVLink Fusion and CXL-based memory pooling as they plan cluster topologies.
Marvell's Dave Lazovsky demonstrated a small photonic module that moves high-bandwidth memory (HBM and DDR) up to 50 meters away from a GPU, compared with the usual ~1.5 mm on-package placement. The link keeps latency in the hundreds of nanoseconds, close enough for practical memory use. This decouples memory from the accelerator package for the first time, letting AI clusters share memory capacity across nodes rather than locking it to each GPU.
Marvell’s Dave Lazovsky is holding a tiny module that uses photonics to move high-bandwidth memory (HBM + DDR) as far as 50 meters away from the GPU.
this can sit up to 50 meters from a GPU instead of the usual 1.5 mm.
For the first time, memory no longer has to live right next to compute, letting AI clusters share massive capacity without being locked to the accelerator package.
Latency stays in the hundreds of nanoseconds, close enough for practical memory use. Basically, Memory and compute now scale on separate axes instead of being locked together. https://t.co/jl7wyH9gfd