NewsNTech
The competitive unit in large-scale AI infrastructure has moved up the stack.
Hyperscalers sourcing capacity for training and inference want a validated rack, components co-designed and tested together, ready to deploy at scale. Nvidia built a durable market position there.
AMD is now competing at that same level with Helios, its first complete AI rack system, and has added Microsoft as a new customer alongside Meta, OpenAI, and Oracle.
The integration layer AMD is entering Selling individual GPU accelerators puts a chip vendor at the mercy of whoever assembles the full system. A rack platform changes that dynamic.
Keep reading