IT之家 · ITHome
Amazon’s Jassy says broader enterprise AI adoption needs inference costs cut to a tenth
Amazon CEO Andy Jassy says AI inference costs must fall to roughly one-tenth of current levels for broad adoption in enterprise software. He described everyday business software as an underdeveloped market, with costs limiting adoption. Amazon hopes its Trainium chips can provide a lower-cost alternative, and Jassy said its chip business is helping reduce inference expenses.

What to know
- 01
The report says Trainium offers about 30%–40% better price-performance than comparable processors.
- 02
Amazon previously disclosed more than $225 billion in future revenue commitments for Trainium, including agreements with Anthropic and OpenAI.
Why it matters
Jassy’s comments highlight inference costs as a constraint on enterprise software adoption of AI and a problem Amazon is seeking to address through its own chips.
AI-assisted brief. Read the source for full details. Editorial & sourcing policy

