AWS NVIDIA shortages demonstrate need for custom silicon
Share This Post
Our research shows that even Amazon Web Services can’t get access to the latest NVIDIA technology. For a week, TechInsights probed AWS to see what accelerators were available in what regions. In general, accelerator-based instances from AWS are widely accessible from a wide range of regions. However, end users seeking access to newer NVIDIA H100 instances, ideal for large language models (LLMs), are likely to face frustration due to ongoing supply chain issues. Such supply shortages and reliance on NVIDIA help explain why cloud providers are making their own chips.
