How the Hyperscalers Are Moving Beyond NVIDIA
Broadcom’s recently announced plans to create a $10 billion custom AI chip for OpenAI highlights how the hyperscalers and even AI companies are attempting to reduce their reliance on NVIDIA. And their efforts could change the AI marketplace.
For the past several years, Amazon, Microsoft, Google, and others have struggled with NVIDIA’s corner on the market for AI training and inference. This isn’t just about vendor lock-in; the hyperscalers also are increasingly concerned about NVIDIA’s supply chain issues and the threat of not having the chips they require.
AMD is increasingly catching up to NVIDIA in some situations. Oracle, for instance, has a strong partnership with AMD as well as with NVIDIA. And AMD’s Instinct GPUs have become major components in the networks of Meta and Microsoft. At least one neocloud, TensorWave, has built its network entirely on AMD chips.
The hyperscalers have also poured enormous resources into creating their own chips to improve their AI networks and services. While NVIDIA remains a mainstay in most hyperscaler footprints, efforts to increase self-sufficiency and reduce reliance on a single vendor are ongoing. Let’s take a look at what they’re doing.
Hyperscaler Proprietary AI Chip Efforts
Alibaba
To access the rest of this article, you need a Futuriom CLOUD TRACKER PRO subscription — see below.
Access CLOUD TRACKER PRO
|
CLOUD TRACKER PRO Subscribers — Sign In |
