Amazon is spending big money on AI chips to reduce its reliance on Nvidia

Amazon will roll out its second-generation Trainium 2 AI chip next month with a promise of up to four times better performance than the first generation.

As reported by The Financial Times, Annapurna Labs, an Israeli microelectronic startup, designed the chip. Amazon acquired the company in 2015. Their job is to create chips that rival Nvidia's offerings and become a healthy alternative for their reliance on Nvidia. 

Speaking with Financial Times, Dave Brown, vice-president of compute and networking services at AWS, said they want "to be absolutely the best place to run Nvidia, but at the same time we think its healthy to have an alternative." 

Nvidia has a commanding 80% hold of the AI chip market, with an astounding $26.3 billion in revenue for chip sales in the second fiscal quarter of 2024. Amazon expects to spend almost $75 billion in capital through 2024, the majority allocated to technology and infrastructure. The company says it will use up even more in 2025. 

Amazon expects to spend almost $75 billion in capital through 2024, the majority allocated to technology and infrastructure. The company says it will use up even more in 2025. While Amazon does not submit its chips for independent performance benchmarks, it said that "Inferentia", its second-line of specialist chips, was already 40% cheaper to run for generating AI responses. 

"It's not [just] about the chip, it's about the full system," according to Remi Sinno, ex-Softbank and Intel, and currently the director of engineering at Annapurna. 

Sinno said, "it's really hard to do what we do at scale. Not too many companies can."