The generative AI revolution has a hidden, multi-trillion-dollar problem. While the world fixates on the rapid emergence of AI tools and the race to build ever-larger models, a quieter but more consequential shift is unfolding inside data centres. The cost of running these models and executing inference eclipses the cost of training them, turning deployment economics into the defining challenge of the AI era.
“We are entering the age of inference,” says Sid Sheth, President and CEO of d-Matrix, a semiconductor company building purpose-designed…
