Amazon Prioritizes Nvidia and In-House Chips for AI Cloud Infrastructure

```html

Amazon's AI Strategy: Focus on Nvidia and In-House Developments

Amazon Web Services (AWS) is not currently utilizing AMD's new AI chips in its cloud infrastructure. Gadi Hutt, Senior Director for Customer and Product Development at Annapurna Labs, Amazon's chip division, explained that customer demand for AMD's MI300 series is currently insufficient to justify their deployment in the AWS cloud.

Nvidia and Trainium: The Spearhead of the AI Strategy

Instead, Amazon is relying on proven technologies and internal developments. At re:Invent, Amazon's annual conference, the integration of the next generation of Nvidia's Blackwell graphics units into the EC2 P6 instances was announced. These are expected to offer 2.5 times higher performance and are specifically optimized for generative AI. In parallel, Amazon presented EC2 instances and UltraServers equipped with Trainium2 chips from Annapurna Labs. The development of the successor, Trainium3, is already underway. This is expected to deliver four times the performance while simultaneously improving energy efficiency compared to the current generation.

AMD in the AI Market: Challenges and Opportunities

AMD's launch of series production for the Instinct MI300 accelerators last year was met with high expectations. The Instinct MI300X was positioned as a cost-effective alternative to Nvidia's H100 chip. Despite rising sales of AI chips in the current year, AMD still lags significantly behind industry leader Nvidia. Nvidia dominates the market for AI accelerators with a market share of approximately 80 percent.

AMD CPUs: Still an Integral Part of the AWS Infrastructure

Despite the current decision against using AMD's AI accelerators, Hutt does not rule out their future integration into the AWS cloud. The collaboration between Amazon and AMD remains close in other areas. AWS continues to rely on AMD's EPYC processors for its servers. Compared to Intel's Xeon CPUs, the EPYC processors offer more cores and a faster memory subsystem, which gives them advantages in compute- and memory-intensive applications.

The AI Chip Market: Dynamics and Competition

The market for AI chips is highly dynamic and competitive. The demand for high-performance hardware for AI applications is rising rapidly. Companies like Amazon, Google, and Microsoft are investing heavily in the development of their own AI chips to reduce their dependence on external suppliers and optimize their cloud infrastructure. At the same time, new players are entering the market with innovative solutions and intensifying competition. The future of the AI chip market remains exciting and will be significantly influenced by technological advancements and the strategic decisions of the major tech companies.