AI chip Computational Infrastructure
noun phrase
Definition: A semiconductor chip designed or optimized to execute artificial-intelligence workloads, especially machine-learning training and inference, more efficiently than general-purpose processors by accelerating operations such as matrix multiplication, parallel computation, and data movement [IBM].
Example in context: “Graphcore’s IPU-GC200, launched in 2020, is an AI chip measuring around 800 mm2 and containing 23.6 billion transistors on a 7-nm node.” [Ozkan et al. 2025]
Synonyms: AI processor; AI accelerator chip
Related terms: AI accelerator; GPU; TPU; NPU; inference chip; AI chipset