Indian engineering conglomerate Larsen & Toubro (L&T) will build the country’s largest single-cluster AI data center for Together AI, according to a press release the company published on August 13. The deployment, which L&T classifies internally as a “Mega” order, will house 10,000 Nvidia B300 GPUs at a campus in Chennai.
Deal terms
L&T did not disclose an exact dollar figure, but Indian media reports the order falls in the company’s “Mega” tier, worth roughly ₹10,000–15,000 crore — about $1 billion to $1.6 billion at August 2026 exchange rates. The GPUs will run at a data center operated by Vyoma.AI, L&T’s AI infrastructure arm, through its subsidiary LTN Compute. Phase 1 of the site is designed for 250 megawatts of IT capacity, backed by 150 megavolt-amperes of power infrastructure already in place.
Who’s behind it
Together AI is a US-based AI cloud provider that hosts, fine-tunes and serves open-weight models for enterprise customers; it raised an $800 million Series C round in July at an $8.3 billion valuation. Co-founder and CEO Vipul Ved Prakash said the partnership brings “the scale, resilience and engineering excellence this moment demands” to the company’s push into India, without saying when the cluster is expected to go live.
For L&T, the order marks the first major deployment under what chairman S. N. Subrahmanyan calls the company’s “Gigawatt AI Infrastructure Mission” — a push to build hyperscale AI data centers, sovereign cloud platforms and GPU-as-a-service offerings across India. L&T says the goal is to make the country “a global hub for next-generation AI infrastructure,” entering a market where Nvidia hardware already anchors multibillion-dollar buildouts from Japan to South Korea.
The Chennai project adds India to a fast-growing list of countries racing to secure GPU capacity, following deals such as Riot Platforms’ $9.1 billion data center agreement with Anthropic and Nvidia-backed Firmus’ $2 billion AI factory expansion. It underscores how demand for inference and fine-tuning capacity is pushing AI infrastructure investment well beyond the US and China.