AMD expects deployment of its latest graphics processing unit (GPU)-based high-performance computing platform, Helios, to begin in the second half of this year across several countries, including India. The company aims to capitalise on demand for artificial intelligence (AI) and the proposed data centre capacity build-out by global technology giants.
According to AMD, a single Helios rack integrates 72 interconnected MI455X GPU accelerators, delivering up to 2.9 exaflops of FP4 compute performance per rack. FP4 refers to a number format commonly used in AI workloads. Initial deployments of Helios systems, including in India, are expected to begin in the second half of 2026.
Demand for GPUs has surged sharply on the back of widespread AI adoption globally, with US chip company Nvidia dominating the market with over 80 per cent share. The Helios platform will compete with Nvidia’s Vera Rubin POD, which claims to deliver a compute speed of 3.6 exaflops, though at a comparatively higher cost.
