Large language models are helping food manufacturers save billions annually with predictive analytics for growing and buying.

In tests, the algorithm reduced energy consumption by 95 percent for tensor multiplications and by 80 percent for dot products, both of which are common tasks in AI computations.