Amazon has announced the development of a significant supercomputer and new server powered by its proprietary AI chips.
The chip cluster, designated Project Rainier, is designed for Anthropic, an AI startup in which Amazon has invested $4 billion.
Expected to be operational in 2025, it will rank among the largest clusters for AI model training.
At its annual re:Invent conference, AWS also introduced the Ultraserver, which integrates 64 interconnected chips for efficient AI processing.
Amazon is positioning its custom silicon, particularly Trainium chips, as alternatives to Nvidia’s graphics processing units (GPUs), which currently dominate the AI semiconductor market.
Semiconductor market
AWS executives noted the need for alternatives to Nvidia. The global AI semiconductor market, valued at $117.5 billion in 2024, is projected to reach $193.3 billion by the end of 2027.
Annapurna Labs, Amazon’s Austin-based chip design facility, is at the forefront of these efforts. The facility emphasizes a flexible, rapid development approach without traditional specialization.
Amazon’s strategy involves offering lower-cost solutions without compromising performance. The recent announcements highlight its aim to enhance control over its supply chain while reducing reliance on Nvidia.
