Amazon is stepping up its push into artificial intelligence, leaning on two homegrown technologies: its Nova family of AI models and its Trainium chips, according to a report from Chosunbiz surfaced via Google News.
The details available are limited, but the framing is clear. Rather than relying solely on outside suppliers, Amazon is pairing its own AI models (Nova) with its own custom silicon (Trainium) to accelerate its AI efforts. Trainium is Amazon's purpose-built chip line aimed at the heavy computing work that training and running AI systems require.
Why pair the two? Controlling both the model and the chip it runs on is a strategy that can lower costs and reduce dependence on third-party hardware. In an industry where access to specialized AI chips has been a major bottleneck, owning that stack is a meaningful lever.
It is worth being precise about what this report does and does not establish. Chosunbiz's piece signals that Amazon is intensifying this combined model-and-chip approach, but the source provided here does not include specific figures, timelines, or performance claims. Readers should treat it as a directional signal rather than a detailed roadmap.
Why it matters: if Amazon can make its own models run efficiently on its own chips, it strengthens its position against rivals in the increasingly competitive — and expensive — race to supply AI computing power.