L&T to Build 10,000-GPU Nvidia AI Factory in Chennai for Together AI

Larsen & Toubro has landed one of the largest AI infrastructure orders awarded in India to date, with its data centre arm Vyoma.AI set to build a 10,000-GPU Nvidia B300 AI factory in Chennai for US-based AI cloud company Together AI, in a deal valued at an estimated Rs 10,000-15,000 crore.

The Deal

Under the arrangement, Vyoma.AI will develop a large-scale AI computing cluster housing 10,000 Nvidia B300 graphics processing units at its Chennai data centre campus, built out with high-performance networking, advanced storage systems and low-latency interconnects designed to support demanding AI training and inference workloads. Once operational, the facility is being positioned as one of India's largest single-cluster AI computing installations, reflecting the scale of investment now flowing into the country's data centre and AI infrastructure ecosystem.

Together AI, the US-based company commissioning the facility, specialises in providing cloud infrastructure and compute capacity for training and running large AI models, serving startups, enterprises and research labs that need access to high-end GPU clusters without building their own data centres from scratch. The Chennai facility will give the company a significant compute footprint in India, a market that has seen surging demand for AI infrastructure over the past year.

Part of a Broader Push

The order builds on an earlier partnership announced between L&T and Nvidia earlier this year to develop gigawatt-scale AI data centre capacity in India, underscoring how quickly the country's AI infrastructure ambitions have scaled up through 2026. The Chennai project also fits into a wider national push: the government's IndiaAI Mission, backed by an outlay of roughly Rs 10,372 crore, has already helped expand the country's shared AI computing capacity to more than 45,000 GPUs as of mid-2026, with over 200 projects having accessed subsidised computing capacity through the mission by August.

For L&T, the deal represents a major expansion of its footprint in the data centre and digital infrastructure space, an area the engineering conglomerate has been building out as it diversifies beyond its traditional construction and engineering businesses. Large-scale AI infrastructure orders of this size have historically been rare in India, making Thursday's announcement notable both for its scale and for what it signals about investor and corporate confidence in India's ability to host frontier AI compute capacity.

Why It Matters

India has increasingly positioned itself as a destination for AI compute investment, driven by a combination of relatively lower power costs, growing domestic demand for AI services, and government incentives aimed at building indigenous AI capabilities. Large clusters like the one planned in Chennai could help address a persistent shortage of high-end GPU capacity that has constrained AI development for Indian startups and enterprises, many of which have previously had to rely on overseas cloud providers for large-scale training runs.

As more global AI companies look to diversify their compute footprint beyond the US and China, deals like the L&T-Together AI partnership suggest India is emerging as a serious contender for large-scale AI infrastructure investment, with implications for the broader semiconductor, power and construction ecosystems that support such projects.

This article is an original editorial summary based on publicly reported information. It has been independently written for publication and does not reproduce content from any single source.