A Rather Unusual Line Item for L&T's Order Book

L&T has spent decades building refineries, metro corridors, defence platforms and power plants, so an order for ten thousand graphics processing units reads a little strange next to that legacy. Yet that is exactly what landed in the company's order book this week. Vyoma.AI, L&T's data centre arm, has won a contract to build a Nvidia B300 AI factory at its Chennai campus for Together AI, a US-based AI cloud company. The order is valued between Rs. 10,000 crore and Rs. 15,000 crore, which puts it among the largest AI infrastructure deployments announced in India to date.

It is worth sitting with that number for a second. This is not a small pilot project or a proof-of-concept facility. It is a full-scale AI factory, and the fact that an engineering and construction major like L&T is the one building it tells you something about where large-scale AI infrastructure spending in India is actually headed.

What L&T Actually Won

The contract covers a 10,000-GPU Nvidia B300 cluster to be built at Vyoma.AI's Chennai data centre campus. The B300 is Nvidia's latest-generation chip built for the heaviest AI workloads, so this is not a facility designed for light experimentation. It will support AI training, fine-tuning and inference at scale, backed by high-performance networking, storage and low-latency interconnects, the kind of specification list you would expect from a facility meant to serve serious commercial AI demand rather than internal research.

Phase 1 of the Chennai campus has been designed to support 250 MW of capacity, with 150 MVA of power infrastructure readiness already built in. Power, more than chips, tends to be the real bottleneck in AI infrastructure projects globally, so having that readiness baked into phase 1 says the campus was planned with expansion in mind rather than being scaled up reactively later.

Who Is Together AI, and Why Does a US Company Need Capacity in Chennai

Together AI is a US-based AI cloud platform that provides training, fine-tuning and inference infrastructure to companies building AI applications. For a company like this, GPU capacity is effectively inventory, and demand for that inventory has been outpacing supply globally for a while now. Setting up dedicated capacity in India gives Together AI a geographically diversified compute base and access to India's cost and talent advantages, while giving L&T a marquee international anchor client for its data centre ambitions rather than just domestic enterprise demand.

This Is Part of a Much Bigger L&T Push

This order did not appear out of nowhere. It comes close on the heels of L&T transferring its own data centre and cloud services business into Vyoma.AI for Rs. 1,400 crore, effectively consolidating its AI infrastructure ambitions under one dedicated unit before handing that unit a headline order many times the size of the transfer itself. L&T has also separately announced a broader partnership with Nvidia to build gigawatt-scale AI infrastructure in India, with plans to scale GPU clusters at the Chennai campus and add a new 40 MW facility in Mumbai. L&T chairman and managing director S N Subrahmanyan has described the ambition in terms of building secure, scalable and sovereign infrastructure for sectors ranging from manufacturing and energy to healthcare and financial services, which gives you a sense of how central this bet has become to the company's broader strategy rather than being a side project.

L&T's Recent Data Centre Moves, Side by Side

Move	When	Scale
Data centre business transferred to Vyoma.AI	August 2026	Rs. 1,400 crore
Nvidia B300 AI factory for Together AI	August 2026	Rs. 10,000 to 15,000 crore, 10,000 GPUs
Broader Nvidia gigawatt-scale partnership	2026, ongoing	Chennai scaling up, new 40 MW Mumbai facility

Why This Deal Is Bigger Than It Looks on Paper

The numbers alone explain why this matters to L&T's shareholders, but the sequencing matters just as much. Setting up Vyoma.AI as a dedicated entity first, then landing a marquee order of this size soon after, suggests the group had this pipeline lined up before the corporate restructuring rather than the restructuring being a reaction to unexpected demand. That is generally a healthier pattern for investors to see than the reverse.

How India's Largest AI Deployment Compares Globally

Context helps here, because ten thousand GPUs sounds abstract without something to measure it against. Saudi Arabia's Humain project, backed by a broader USD 10 billion infrastructure programme, involves an initial deployment of 18,000 Nvidia Grace Blackwell GB300 GPUs for a 500 MW facility. L&T's Chennai project is smaller in raw GPU count but is still described as the largest such deployment within India, which says less about India's ceiling and more about how early the country still is in this particular race.

GPU Deployment Scale: India vs a Global Comparable

What It Means for L&T Stock and the Broader Market

L&T shares rose as much as 1.2% on the announcement, a fairly modest move for a company of L&T's size, but the more interesting story is what this signals about where the market is willing to reward infrastructure names for AI-adjacent exposure. This comes at a time when global capital has been rotating between AI-linked trades, something we have tracked closely, including how a similar rotation played out in why Nifty IT recently became FPIs' favourite trade again. The AI infrastructure theme is not confined to IT services companies chasing AI-linked revenue. It is increasingly showing up in construction, power and data centre plays as well, and L&T sits at an interesting intersection of all three.

It also fits into a broader global pattern investors are watching closely, where the scale of AI capital expenditure is becoming a market-moving variable in its own right. We saw versions of this play out globally in how Amazon and Apple's diverging AI spending stirred fears among Magnificent Seven investors, and again in how Microsoft and Meta's AI trades split on the same night. L&T's Chennai order is a reminder that this capex race has real, physical infrastructure counterparts being built in India right now, not just headline numbers from US earnings calls.

The Bigger Picture for India's AI Infrastructure Story

This order lands alongside a broader wave of India-linked AI infrastructure activity we have been covering, including how Tata's semiconductor investments are positioning India within the chip supply chain, and the global backdrop of China's own massive semiconductor investment push through CXMT. Whether India can meaningfully compete for this kind of large-scale AI infrastructure demand over the next few years is still an open question, but deals like this one are the closest thing to a data point investors currently have.

For readers tracking which listed names actually stand to benefit from this theme rather than just the narrative around it, our roundup of five Indian stocks riding the AI data centre wave is worth revisiting alongside this update, and if you are trying to separate genuine earnings-driven IT sector strength from flow-driven rallies, our piece on whether IT stocks are actually on the rise in 2026 covers that distinction in more depth. Investors interested in how this fits within the Nifty IT basket specifically can also refer to our explainer on how the Nifty IT index is constructed to understand which companies actually move that index.

As with any single order announcement, the real test will be execution over the coming quarters rather than the headline number itself. Rs. 15,000 crore worth of construction and infrastructure work does not get delivered overnight, and GPU availability, power tie-ups and commissioning timelines can all shift a project's actual revenue recognition well beyond the initial announcement date. This article is for informational purposes only and is not investment advice, and any investment decision should factor in your own risk profile rather than a single order win.

Frequently Asked Questions (FAQ)

1. What did L&T actually win the order for?

L&T's data centre arm Vyoma.AI won an order worth Rs. 10,000 to 15,000 crore to build a 10,000-GPU Nvidia B300 AI factory at its Chennai campus for US-based Together AI.

2. Who is Together AI?

Together AI is a US-based AI cloud platform that provides training, fine-tuning and inference infrastructure to companies building AI applications, and is the anchor client for this Chennai facility.

3. How large is this AI factory compared to other global projects?

At 10,000 Nvidia B300 GPUs, it is described as India's largest such deployment so far, though smaller than some global projects like Saudi Arabia's Humain initiative, which involves an initial 18,000 GPU deployment.

4. How did L&T's stock react to the announcement?

L&T shares rose as much as 1.2% following the announcement, a modest but positive reaction reflecting the market's growing interest in AI infrastructure exposure.

5. Is this L&T's only AI infrastructure project?

No. This order follows L&T transferring its data centre and cloud services business into Vyoma.AI for Rs. 1,400 crore, and sits within a broader Nvidia partnership to build gigawatt-scale AI infrastructure, including expansion at Chennai and a new facility in Mumbai.