#custom-silicon

#custom-silicon

Amazon is introducing tools to enhance generative AI capabilities while prioritizing affordability and reliability for its customers.

AWS's Graviton CPU chips are now used by over 90% of its largest data center customers, highlighting their success since launch in 2018.

Google's C4A virtual machines deliver up to 10% better price-performance for general workloads compared to competitors, enhancing its cloud service offerings.

Google announced its first Arm-based CPU, Axion, designed using Arm's Neoverse V2 CPU.

Axion offers 30% better performance than current Arm-based instances and 50% better than x86-based instances.

Google has become the third-largest designer of datacenter processors, with TechInsights estimating 2 million shipments of its custom TPUs in 2021.

The growth of Google's chip business, particularly with TPU v4 and large language models, has significantly increased its market share and benefited its main ASIC partner, Broadcom.

Google announces Google Axion Processors, their first custom Arm-based CPUs for data centers, promising industry-leading performance, energy efficiency, and advanced capabilities.

Google unveiled its custom Arm CPU, Axion, based on Neoverse V2 design.

Axion aims for superior performance compared to existing Arm instances and Intel/AMD processors.

Nvidia's largest customers are developing AI chips that compete with Nvidia, but the company is willing to help them do so.

Nvidia is forming a group to help clients build their own solutions using Nvidia IP, potentially becoming an AI licensing giant.

Nvidia is creating a business unit to sell its intellectual property and design services to cloud providers.

Cloud providers like AWS, Microsoft, and Meta have been developing their own custom silicon alternatives to Nvidia's GPUs for AI workloads.