Amazon is currently putting a new server design to the test, utilizing its own custom-built AI chips. This move signals a direct challenge to industry giant NVIDIA, currently dominating the market with its powerful AI GPUs. The new server design is being tested by a select group of engineers and is considered a “closely guarded” secret.
The news comes directly from Amazon executive Rami Sinno, who revealed details about the server during a recent visit to the Amazon AI chip lab. Amazon’s ambitious goal is to minimize its reliance on NVIDIA’s high-priced GPUs by developing its own AI processors. The company actively uses AI across its Amazon Web Services (AWS) platform and plans to invest a whopping $100 billion in data centers specifically designed for AI in the future.
David Brown, Vice President of Compute and Networking at AWS, emphasizes the potential cost savings and performance gains with Amazon’s new chips. He claims that the new chips could offer up to 40% to 50% improvement in price and performance, essentially making it half as expensive to run AI models compared to using NVIDIA’s offerings.
Amazon aims to use its in-house AI chips to help its customers handle complex computations and process massive amounts of data more efficiently and cost-effectively. Rami Sinno, the director of engineering for Amazon’s Annapurna Labs (part of its cloud business AWS), highlights that customers are increasingly demanding more affordable alternatives to NVIDIA’s offerings.
During the company’s recent Prime Day 2024 sales, Amazon deployed a significant number of its new AI chips. The company utilized 250,000 Graviton AI chips and 80,000 of its custom AI chips to manage the surge in activity across its platforms. This large-scale deployment further underlines Amazon’s commitment to its new AI chip strategy and its ambition to become a major player in the AI hardware market.
