L&T Wins Up to Rs 15,000 Crore Order to Build India's Largest AI Factory
On 13 August 2026, Larsen and Toubro announced it had secured an order in the range of Rs 10,000 to Rs 15,000 crore to build India's largest single-cluster NVIDIA B300 AI factory at the Chennai data centre campus of Vyoma.AI, an L&T company. The project is being executed by LTN Compute, an AI infrastructure subsidiary under Vyoma.AI's business, and will deploy 10,000 NVIDIA B300 graphics processing units in a single integrated cluster for Together AI, a US-based AI cloud provider. At the upper end of the order value range, this is the largest AI infrastructure build order announced by any Indian company as of August 2026, and the project's GPU count makes it one of the largest single-cluster AI compute deployments in Asia.
The NVIDIA B300 and What 10,000 GPUs Represent at Scale
The NVIDIA B300 is a Blackwell-architecture GPU designed for high-throughput AI inferencing, model fine-tuning, and large-scale training workloads. A cluster of 10,000 B300 units constitutes the kind of compute infrastructure that enables hyperscale AI inference — simultaneously serving large numbers of API requests against large language models — as well as the fine-tuning of frontier-scale models on domain-specific or proprietary data. For context, cloud infrastructure at this scale is typically associated with hyperscaler facilities operated by AWS, Microsoft Azure, and Google Cloud. L&T's order represents an Indian enterprise building this tier of compute infrastructure domestically, under a commercial contract with a US-based AI cloud customer. The single-cluster configuration is significant: rather than distributing 10,000 GPUs across disconnected zones, the facility is designed as one integrated fabric using high-performance networking, ultra-low-latency interconnects, and high-throughput parallel storage — characteristics that enable the coordinated training and fine-tuning workloads that distributed clusters cannot efficiently support.
Vyoma.AI and LTN Compute: L&T's AI Infrastructure Play
Vyoma.AI is L&T's AI infrastructure company, and LTN Compute is the AI infrastructure subsidiary through which the Together AI project will be executed. The Chennai campus at which the factory is being built is designed as a gigawatt-scale AI infrastructure site — a facility planned at a total capacity that matches the compute footprint of major global hyperscalers. Phase 1 of the Chennai campus is designed for 250 megawatts of operational capacity, with power infrastructure readiness of 150 MVA provisioned at the outset. The 250 MW Phase 1 design gives the campus room to absorb not only the 10,000-GPU Together AI cluster but additional capacity as demand from other enterprise AI customers materialises. The infrastructure platform combines hyperscale data centre design, accelerated computing, high-performance networking, ultra-low-latency interconnects, high-throughput parallel storage, and AI infrastructure operations — the full stack required to support simultaneous inference, fine-tuning, and training workloads at commercial scale.
Together AI's Role and Strategy in India
Together AI, founded in 2022 and headquartered in San Francisco, operates a cloud platform specialising in AI inference and fine-tuning for enterprise and AI-lab customers. The company's decision to anchor a major GPU cluster in India, through a build partnership with L&T, reflects a broader shift in enterprise AI infrastructure strategy: rather than routing all inference through US-based hyperscaler regions, AI cloud providers are establishing regional compute presence in markets with large enterprise AI demand and regulatory expectations for data localisation. India's combination of a large developer and enterprise software market, a government AI mission backed by significant GPU procurement, and data localisation expectations across regulated industries makes it a compelling location for an AI cloud infrastructure investment of this scale. Together AI's Chennai cluster will make low-latency inference and data-residency-compatible AI processing available to Indian enterprises using its platform, without those enterprises bearing the capital cost of GPU infrastructure ownership themselves.
What This Means for India's AI Infrastructure Landscape
L&T winning a Rs 10,000-15,000 crore AI factory order demonstrates that India's AI infrastructure build-out is no longer confined to hyperscaler data centre expansion by AWS, Google, or Microsoft. A domestic infrastructure conglomerate constructing a 10,000-GPU AI factory for a US AI cloud company is a qualitatively different signal from the government's IndiaAI Mission GPU procurement: it confirms that commercial enterprise AI demand for India-based compute is real enough to attract private investment at hyperscale economics. For Indian software firms building AI-powered products for BFSI, healthcare, or public sector clients where inference must occur on India-domiciled infrastructure, a 10,000-B300 cluster in Chennai operating as an AI-native cloud platform expands the supply of sovereign-adjacent compute available via commercial contracts — without requiring those firms to stand up their own GPU infrastructure. For Indian engineering teams with specialisation in data centre software, AI infrastructure management, distributed computing, and HPC networking, L&T's AI factory investment signals growing domestic demand for the skills that have historically required working with overseas hyperscaler clients.
The Bottom Line
On 13 August 2026, L&T announced an order valued at Rs 10,000 to Rs 15,000 crore from Together AI to build India's largest single-cluster AI factory at Vyoma.AI's Chennai campus. The facility will house 10,000 NVIDIA B300 GPUs in a single integrated cluster, executed by LTN Compute, L&T's AI infrastructure subsidiary. Phase 1 of the Chennai campus is designed for 250 MW of operational capacity with 150 MVA power readiness. The cluster will power Together AI's AI-native cloud platform for large-scale inference, model fine-tuning, and training workloads. For Indian software teams building AI products that require India-domiciled compute, the Together AI cluster expands the supply of commercially available sovereign-adjacent GPU infrastructure in Chennai. For Indian engineers and infrastructure specialists, L&T's AI factory investment confirms that hyperscale AI operations is now a domestic growth market, not only a requirement for overseas cloud engagements.
Frequently Asked Questions
What order did L&T announce on 13 August 2026 and what does it involve?+
On 13 August 2026, Larsen and Toubro announced it had secured an order in the range of Rs 10,000 to Rs 15,000 crore to build India's largest single-cluster NVIDIA B300 AI factory at the Chennai data centre campus of Vyoma.AI, an L&T company. The project is being executed by LTN Compute, an AI infrastructure subsidiary under Vyoma.AI's business. The factory will deploy 10,000 NVIDIA B300 GPUs — Nvidia's Blackwell-architecture GPU designed for high-throughput AI inferencing, fine-tuning, and training — in a single integrated cluster for Together AI, a US-based AI cloud provider. The cluster will power Together AI's AI-native cloud platform for large-scale inference, model fine-tuning, and training workloads.
What is the infrastructure specification of the Chennai AI factory site?+
Vyoma.AI's Chennai data centre campus is designed as a gigawatt-scale AI infrastructure site. Phase 1 is designed for 250 megawatts of operational capacity, with power infrastructure readiness of 150 MVA provisioned at the outset. The platform is built to combine hyperscale data centre infrastructure, accelerated computing, high-performance networking, ultra-low-latency interconnects, high-throughput parallel storage, and AI infrastructure operations. The single-cluster design — rather than distributing GPUs across disconnected zones — is specifically chosen to support the coordinated compute required for simultaneous inference, fine-tuning, and training workloads that distributed multi-zone clusters cannot efficiently deliver.
Who is Together AI and why is it building a cluster in India?+
Together AI is a San Francisco-based AI cloud provider founded in 2022, specialising in AI inference and model fine-tuning for enterprise and AI-lab customers. Its decision to anchor a major GPU cluster in India reflects growing commercial demand from Indian enterprises for AI inference that is processed on India-domiciled infrastructure — driven by data localisation expectations in regulated sectors including banking, financial services, insurance, and government. Rather than routing all inference through US-based hyperscaler regions, Together AI is establishing regional compute presence in India. The Chennai cluster, once operational, will make low-latency, data-residency-compatible AI processing available to Indian enterprises using Together AI's cloud platform without those enterprises purchasing and operating their own GPU infrastructure.
What does L&T's AI factory order mean for Indian software and infrastructure teams?+
For Indian software teams building AI products for regulated-sector clients — banking, healthcare, government — where inference must occur on India-domiciled infrastructure, Together AI's 10,000-GPU Chennai cluster expands the supply of commercially available sovereign-adjacent compute, reducing dependence on overseas cloud regions. For Indian engineering and infrastructure teams with expertise in data centre software, AI infrastructure management, HPC networking, and distributed computing, L&T's Rs 10,000-15,000 crore AI factory investment signals that hyperscale AI infrastructure operations is now a growing domestic market — not only a skill required for overseas hyperscaler engagements. The scale of the order also validates the commercial demand for India-based AI compute from international AI cloud providers, suggesting that further infrastructure investment by global AI companies in India is likely.
Written by
TechPillow Team
Sharing insights on technology, product development, and the Indian tech ecosystem.
