AI GlossaryㅋInfrastructure and chips
Compute
The umbrella term for the computational power used to build and run AI. The "race for compute" is really a race for GPUs, electricity, and data centers.
In plain words
Compute is the catch-all term for the computational power needed to build and run AI. It covers everything from GPU counts to data center scale to electricity supply, and in this industry it's treated almost like a currency.
When people say "we're short on compute," they mean they couldn't get enough GPUs. The "race for compute" is a fight over chips, power, and land, and securing compute is a core item in national AI strategies. Along with data and talent, it's one of the three key resources for AI competitiveness—and the one most directly convertible into dollars.
See also
Stories using this term
- Google Discloses Server-Side AI Memory Design for Private AI ComputeAI · 2026.09.25
- Perplexity Computer Shifts Only Sensitive Tasks to Local Mac ProcessingAI · 2026.09.02
- Mistral and Saudi Arabia's HUMAIN to Co-Develop Arabic AI ModelsBusiness · 2026.08.25
- Mistral strikes hundreds-of-millions-euro partnership with Saudi Arabia's HUMAINBusiness · 2026.08.25
- Cerebras CS-4 boosts inference speed up to 30x over GPUsAI · 2026.08.19
- Claude Beats Industry Average Success Rate in Designing Drug-Binding ProteinsAI · 2026.08.19