New Delhi, March 11 -- As artificial intelligence adoption accelerates across enterprises, one constraint is becoming increasingly visible in India's technology ecosystem: access to GPUs. High costs, limited supply chains, and dependence on imports have made large-scale AI infrastructure expensive and difficult to scale. For many startups, enterprises, and public-sector institutions, this hardware bottleneck is slowing the pace of AI deployment.

A Bengaluru- and Palo Alto-based startup, Ziroh Labs, believes the solution may lie in rethinking how AI systems are built rather than simply scaling compute resources. The company recently launched Kompact AI, a platform designed to run large language model inference on standard CPUs instead of ...