Microsoft has launched the Maia 200, the successor to its Maia 100 AI chip, optimized for inference—the process where trained models generate responses.
Key specs include:
- 30% higher performance per dollar compared to Maia 100.
- Up to 10 petaflops (10 quadrillion operations per second).
- Reduced energy consumption for efficient AI workloads.
The chip is already deployed in Azure’s US Central data center, with plans to expand to other regions soon. This boosts Azure’s AI inference capabilities amid growing demand for high-performance computing.
Source: Tom’s Hardware.
