DoubleAI Launches Maxwell to Cut Compute Costs
The new code-execution model optimises live workloads across GPUs and CPUs, with pricing tied to the savings it produces.
DoubleAI has launched Maxwell, a code-execution model designed to optimise software while it is running.
What happened
Unlike a coding assistant that suggests changes to a developer, Maxwell sits between software and hardware and automatically looks for ways to use GPUs and CPUs more efficiently. DoubleAI plans to charge customers a share of the compute savings it generates.
Early customers reported lower costs and faster inference, but the performance claims still need broader independent validation across different workloads.
Why it matters
Compute is one of the largest costs for AI companies. Improving hardware utilisation could increase margins without requiring new chips or lower model demand.
The bigger picture
AI infrastructure is gaining a new optimisation layer between code and silicon. If the approach works reliably, software that makes existing hardware more productive could become as valuable as adding more capacity.