Amazon is nearing the natural conclusion to a partnership with NVIDIA in competitive tech market that has the company looking to produce its own AI chips rather than depending on NVIDIA’s GPUs in the future.
According to a report from the Financial Times, Amazon is ramping up development of its internal custom chip making efforts with the intent to gain more control over its costs as well as an off ramp to the increasingly competitive bidding for NVIDIA’s tightening chip supply.
Amazon joins the likes of Microsoft and Google who are both working on their own custom AI-powered processors for servers and cloud computing development. Amazon’s Trainium and Inferentia chips have been in development for some time with Trainium2 being released late last year and notching close to 50 percent performance gains than the NVIDIA GPUs currently being used by the company.
While NVIDIA’s stock price rose astronomically during the early months of the AI hype cycle, it was only a matter of time before the company would be a victim of its own success.
With every company under the sun looking to NVIDIA for GPUs to power the AI promises levied at investors and customers alike, demand and constrained chips supply is forcing companies to explore independent and cost-saving measures in the form of in-house development that could hurt NVDIA’s current marketplace domination.
As for Amazon, balancing out its dependencies on AI-powered chips was something the company had a bit of foresight into with its 2015 acquisition of Israeli microelectronics provider, Annapurna Labs. Yes, Amazon rushed to NVIDIA’s door like every other tech company in 2021 to get chips and GPUs to help power its AI led initiatives, but like others, the company is realizing that perhaps the best path forward is to bring as many of its efforts in-house.
