Quick summary: L&T's ₹10,000–15,000 Cr order from Together AI is not construction-only EPC — L&T builds and operates, so the spillover lands with equipment suppliers, not civil contractors. Evidence-backed candidates: ABB, Hitachi Energy and CG Power (transformers/switchgear), Blue Star and Thermax (cooling/MEP), Cummins and Kirloskar (gensets), and JSW Steel (CRGO steel).
If you have been watching India's data-centre buildout, the L&T–Together AI headline crossed your feed. The obvious read: L&T won a construction job, so look for civil-work contractors. That read is wrong — and it points at the wrong set of listed companies.
L&T's AI-infrastructure arm LTN Compute, part of its Vyoma.AI venture, is not merely building the Chennai facility. It will own and operate it — hosting 10,000 NVIDIA B300 chips for US cloud platform Together AI in what L&T calls "India's largest single-cluster AI infrastructure" and its "foray into the AI factory business." The order is worth ₹10,000–15,000 crore ($1.05–1.57 billion).
Because L&T is builder, owner and operator in one, it is the party doing the procurement. The spillover therefore sits with the equipment suppliers — electricals, cooling, gensets, fibre. This piece maps the listed names with the strongest evidence, and the honest gaps in that evidence.
Together AI is a tenant buying AI cloud capacity — inference, fine-tuning, training — from L&T's facility. That makes L&T the procurer of the electrical, cooling, power-backup, cabling and connectivity equipment that goes into the build. Its data-centre business is service-led: the Vyoma/Cloudfiniti arm had roughly 14 MW live at Sriperumbudur plus Panvel, 18 MW under development, about 30 MW total, and the Chennai AI campus is Phase 1 of a 250 MW campus with 150 MVA of power readiness.
The data-centre value chain runs: general construction/EPC → electrical infrastructure (transformers, switchgear, switchboards/PDU, UPS, generators) → mechanical/cooling (precision cooling, chillers, fans) → racks and cabling → fibre/connectivity → compute (GPUs, servers) → ownership/operations. Two structural points matter for who benefits:
The L&T–Together AI order is one project in that much larger pipeline. Its electrical slice is modest against the wave — which is why "benefit" here mostly means: these companies' data-centre order books are the market's existing proxy for the whole pipeline, and this order increments it.
Every L&T order announcement in the filings checked shows L&T as the winner (thermal, process equipment, HVDC, substations) — none shows a listed supplier announcing L&T as its customer. So the practical linkage is not "X already supplies L&T"; it is "X already has hyperscale data-centre order books and delivery slots."
The genuinely discriminating variable is capacity, not demand. Global transformer lead times average around 30 weeks versus a peak above 100 weeks in 2023, and up to four years in constrained pockets. ABB India itself flags possible supply constraints by 2027–28, and CG Power has said demand outstrips all planned domestic capacity through FY29-30. In that regime, spillover lands with whoever has delivery slots — which is exactly why the mid-caps adding capacity sit where they do in the rankings below.
Market caps are the screener snapshot as of 2026-08-15; order books are the latest disclosed (mostly Q1 FY27 end).
| Company | Mcap ₹ Cr | Product | Latest order book | Data-centre evidence | Spillover likelihood |
|---|---|---|---|---|---|
| ABB India | 162,082 | Transformers, switchgear, breakers, switchboards/PDU | ₹11,000 Cr backlog (Q1 FY27); DC = 12–13% ≈ ₹1,320–1,430 Cr (derived) | Hyperscaler contracts signed with delivery schedules through 2028; US$75 Mn capacity expansion including a Nelamangala DC factory; pipeline "very strong"; launches 'ArTu Formula' LV switchgear for critical infrastructure | HIGH — largest disclosed DC share; caveat: ABB itself flags possible supply constraints by 2027–28 |
| Hitachi Energy India | 159,222 | Large power transformers, GIS, HVDC | ₹32,222 Cr backlog (Q1 FY27) | DCs a "major contributor" to order book alongside rail/metro; ₹2,000 Cr Karjan large-power-transformer factory explicitly targeting AI data centres (FY28); 15 GW national DC target; "strong visibility" | HIGH — the one maker adding dedicated large-power-transformer capacity for AI DCs; hyperscale LPTs are exactly the bottleneck item |
| CG Power | 140,199 | Power transformers, switchgear (incl. EHV/GIS) | ₹17,333 Cr standalone backlog, +45% YoY (Q1 FY27) | ₹900 Cr US hyperscale-DC transformer order (Jan-2026, largest-ever single order), 12–20 month delivery; transformer capacity 22,000→75,000 MVA in four quarters; ₹748 Cr switchgear expansion; EHV breaker capacity +80%; management cites DCs as a "contributing factor" to +84% pipeline growth; demand outstrips all planned domestic capacity through FY29-30 | HIGH — proven hyperscale wins on both products; binding constraint is capacity/slots, not demand |
| GE Vernova T&D India | 111,382 | Power transformers, GIS, breakers, reactors, HVDC | FY26 intake ₹14,776 Cr (+37%); Q1 FY27 bookings ₹1,140 Cr | ~₹140–150 Cr US export order for the DC market plus multiple DC grid-automation orders (Q1 FY27); export DC pipeline "very strong"; India DC ~1.4–1.5 GW now, scaling toward 400/765 kV | MEDIUM-HIGH — confirmed DC orders but small in quantum so far; strong export positioning |
| Siemens | 140,264 | Switchgear, breakers, building technology | ₹46,670 Cr backlog (+9.6% YoY) | Smart Infrastructure orders +83.6% vs the prior comparable period, DCs named a key driver; India DC capacity "more than double by 2030" | MEDIUM-HIGH — DC named, share undisclosed; DC is one of many drivers |
| Voltamp | 10,260 | Distribution/power transformers | ₹1,200 Cr order book (entry FY27) | DCs a served sector, no share disclosed; new EHV power-transformer plant operational from Jul-2026; copper-price spike slowing order closure | MEDIUM — exposure unquantified; hyperscale-scale LPTs skew to Hitachi/CG/ABB; copper a margin headwind |
| TARIL / Transformers & Rectifiers | 8,947 | Power/distribution transformers | ₹6,630 Cr backlog (Q1 FY27); inflow ₹2,114 Cr, +218% YoY | DC expansion cited only as "emerging demand" — no disclosed DC win; ₹600 Cr capex plus backward integration including in-house CRGO processing to cover 80–85% of raw-material needs | MEDIUM — strongest order momentum of the mid-caps, but no DC proof yet; fresh capacity = potential slot availability |
| Shilchar | 4,466 | Distribution transformers | ~₹500 Cr backlog; ~₹800 Cr pipeline | No DC disclosure found; growth cited as the global transformer shortage, grid, renewables; +6,500 MVA plant by Apr-2027 | LOW — indirect via the global shortage |
| JSW Steel | 309,880 | CRGO electrical steel | — | Sole listed domestic CRGO producer: 50 ktpa now → 350 ktpa by FY28-29 (JFE JV; Nashik 50→250 ktpa at ₹4,300 Cr; Vijayanagar 100 ktpa), framed as reducing India's import dependence, serving "power infrastructure, renewables and data centres" | INDIRECT / STRUCTURAL — CRGO is an input to every transformer build, not this order; the DC wave raises transformer output → CRGO demand, and JSW JFE is the domestic supplier |
Also checked: Havells — no data-centre exposure surfaced in its results calls or annual report (switchgear is a consumer/LV focus) → LOW. Techno Electric — a data-centre operator (Chennai 24→35–40 MW via Techno Infra Developers, ₹550 Cr of QIP proceeds) and T&D EPC contractor (₹11,000 Cr order book), not an equipment maker — an indirect beneficiary of the DC wave as an operator, not a supplier into this build.
Coverage = latest disclosed order book ÷ TTM revenue (screener, as of 2026-08-15) — how many years of current revenue the backlog covers.
| Company | Order book (₹ Cr, as-of/scope) | TTM revenue (₹ Cr) | Coverage (×) | Read |
|---|---|---|---|---|
| Hitachi Energy India | 32,222.1 (Q1 FY27 backlog, standalone) | 9,163 (standalone) | 3.52× | Booked out — ~3.5 years of revenue on the books; matches its "visibility strong, capacity is the constraint" stance |
| TARIL | 6,630 (Q1 FY27, standalone) | 2,552 (consol) | 2.60× | Booked out — plus inflow of ₹2,114 Cr in Q1, +218% YoY |
| GE Vernova T&D | 14,776 (FY26 intake — a flow, not a stock) | 8,000 (standalone) | 1.85× (flow proxy) | Strong flow; no stock backlog figure disclosed — treat as book-to-bill, not coverage |
| CG Power | 17,333 (Q1 FY27 backlog, standalone) | 12,821 (consol) | 1.35× (1.48× on the consolidated backlog of 18,965) | Mid — ratio understated by standalone-vs-consol mismatch |
| ABB India | 11,000 (Q1 FY27 backlog) | 13,611 (consol) | 0.81× | Headroom — needs fresh orders to sustain revenue; DC is 12–13% of the book |
| Shilchar | ~500 (Q1 FY27) | 628 | 0.80× (1.27× with the ~800 Cr pipeline) | Headroom |
| Voltamp | 1,200 (entry FY27, 10,270 MVA) | 2,274 | 0.53× | Most headroom — half a year of revenue booked |
How to read this honestly: these are company-wide coverage ratios (ABB's revenue includes motion/robotics), order books are as of slightly different dates (mostly Q1 FY27 end; Voltamp entry-FY27), and coverage is not capacity utilisation — backlogs execute over multiple years. But the shape is consistent with the structural finding: the binding constraint in this market is delivery slots, not demand. Hitachi and TARIL are effectively booked out; ABB, Voltamp and Shilchar are the ones with capacity headroom to absorb incremental orders — which is exactly why the mid-caps are adding capacity (TARIL ₹600 Cr capex plus in-house CRGO, Voltamp's EHV plant, Shilchar's +6,500 MVA).
| Company | Order book (₹ Cr, as-of/scope) | TTM revenue (₹ Cr) | Coverage (×) | Read |
|---|---|---|---|---|
| Thermax | 14,045 — consolidated closing order balance, Q1 FY27 | 10,839 | 1.30× | Loaded but building — balance +23% YoY while Q1 booking was only +2% YoY (₹2,809 Cr); ~₹300 Cr of shipments slipped out of Q1, which inflated the backlog. Data centres named a "high-potential growth sector", no ₹ quantum disclosed |
| Blue Star | 7,764 — Segment I (Electro-Mechanical Projects & Commercial AC) carry-forward, Jun-30-2026 | 12,798 | 0.61× | Headroom — but DC MEP is the inflow engine: ₹1,500 Cr of DC MEP inflow in Q1 FY27 alone, FY27 guidance ₹3,000 Cr inflow / ~₹1,400 Cr revenue. DC MEP carry-forward was ~₹1,500 Cr at Mar-31-2026 |
| Cummins India | Not disclosed — no order-book ₹ value in Q1 FY27 results | 12,662 | — | Qualitative only: "robust order board" across US and Indian DC players; DC genset orders placed 6–12 months before site readiness. Cannot quantify headroom |
| Kirloskar Oil Engines | Not disclosed for Q1 FY27 — nearest ₹ figure is ₹798 Cr (NPCIL + marine order, as of Q4 FY26) | 7,937 | — | The 192 MW HyperNext DC order is disclosed in capacity, not value (96 × 2,500 kVA, 800 VDC); genset revenue hits FY27, O&M runs 5–6 years. No book-to-revenue math possible |
The table splits exactly on disclosure style. Every name on the electrical side publishes an order-book stock, so coverage was computable for all of them. Here, only Thermax and Blue Star disclose order-book positions; both genset makers report orders qualitatively (capacity/MW, named contracts) but no aggregate ₹ backlog — so the "booked out vs headroom" question is answerable only for the cooling/MEP pair.
On the computable pair: Thermax is the loaded one (1.30×, and rising — inflows keep outpacing execution), Blue Star the one with headroom (0.61×) but with by far the hottest DC inflow — ₹1,500 Cr of DC MEP bookings in a single quarter against ~₹1,400 Cr of guided FY27 DC revenue. As a rough derived cross-check of two disclosed numbers (dates not identical): DC MEP book of ~₹1,500 Cr at Mar-26 vs ~₹1,400 Cr guided FY27 DC revenue ≈ ~1.1× — an estimate, not a reported figure. Scope caveats: Blue Star's ₹7,764 Cr is Segment I only (excludes room-AC/unitary product revenue), so its 0.61× is understated for the projects business; Thermax's 14,045 is consolidated and matches consolidated revenue — the cleanest pair in this table. Both genset names also make non-DC gensets, so even a disclosed backlog would not isolate the DC exposure — their DC books show up in the next three-to-six quarters of revenue, not in a disclosed stock.
Every power and distribution transformer uses grain-oriented electrical steel (CRGO) in its magnetic core, and India imports roughly 90% of its needs — FY23-24 demand was ~400,000 MT against domestic capacity of ~50,000 MT, with 239,200 MT imported. Voltamp's FY26 annual report puts electrical steel + copper + oil + insulation at ~70% of total transformer inputs, so this is a big cost line for the very companies in the shortlist above.
JSW Steel, via its JSW JFE joint venture, is India's only domestic CRGO producer, scaling 50 ktpa → 350 ktpa by FY28-29 (Nashik 50→250 ktpa at ₹4,300 Cr, Vijayanagar 100 ktpa) and explicitly serving "power infrastructure, renewables and data centres". But note the two-sided read: a CRGO squeeze is a cost headwind for ABB, Hitachi, CG and GE, partially offset by their pricing power in a shortage market. That is why TARIL's move to process CRGO in-house (covering 80–85% of raw-material needs) is a margin-protection play, not a demand play.
| Milestone | Detail |
|---|---|
| Complaint | JSW JFE Electrical Steel Nashik (per ET/GTRI/BusinessWorld); CNBC-TV18 names JSW Steel, JSW Vijaynagar Metallics and Jindal Steel Odisha — the two reports differ on the petitioner list |
| Initiation | June 22, 2026 |
| Product scope | CRGO electrical steel and amorphous metal |
| Countries | China, Japan, South Korea, Russia |
| Period of investigation | Apr 1, 2025 – Mar 31, 2026; injury analysis FY2022-23 – FY2024-25 |
| DGTR stance at initiation | Prima facie evidence of dumping found; product allegedly exported "significantly below normal value" |
| Status today (mid-Aug 2026) | Investigation ongoing, no duty imposed; TRIL management "waiting and seeing" |
The standard sequence that remains — no dates were disclosed, so none should be read into it: exporter questionnaires → domestic-industry verification → oral hearing → preliminary findings (where the DGTR may recommend a provisional duty) → final recommendation to the Finance Ministry → Ministry notification makes any duty effective.
What a duty would do to the transformer makers' input costs:
Did any listed company get an order from L&T for the Together AI data centre?
No filing or announcement retrieved names any supplier for this specific build. L&T does not disclose sub-vendor awards, and news coverage of the order names no suppliers. The linkage is indirect: the companies in this piece hold hyperscale data-centre order books, which is the pool that could capture spillover.
Who makes the transformers and switchgear for Indian data centres?
ABB India, Hitachi Energy India and CG Power have the strongest disclosed data-centre exposure — ABB at 12–13% of its order book, CG with a ₹900 Cr US hyperscale transformer order, Hitachi building a dedicated large-power-transformer plant for AI data centres. GE Vernova T&D India and Siemens have confirmed DC orders with smaller disclosed quantum.
Which listed companies have the most data-centre order exposure?
On the electrical side, ABB, Hitachi Energy and CG Power lead; in cooling/MEP, Blue Star booked ₹1,500 Cr of DC MEP inflow in Q1 FY27 alone and Thermax calls data centres a high-potential sector; Cummins and Kirloskar Oil Engines supply gensets (KOEL's 192 MW HyperNext order); Sterlite and HFCL supply fibre connectivity.
What is CRGO steel and why does it matter?
Grain-oriented electrical steel is the core material of every power and distribution transformer. India imports roughly 90% of its needs. JSW Steel, via its JSW JFE joint venture, is the only listed domestic producer, scaling from 50 ktpa to 350 ktpa by FY28.
Is there an anti-dumping duty on CRGO imports?
Not yet. The DGTR initiated an investigation on June 22, 2026 into CRGO and amorphous steel imports from China, Japan, South Korea and Russia. As of mid-August 2026 the investigation is ongoing and no duty has been imposed; any duty would follow the DGTR's final recommendation and a Finance Ministry notification.
Are the transformer companies sold out?
Measured as order book ÷ TTM revenue, Hitachi Energy (3.52×) and TARIL (2.60×) are effectively booked out — about 3.5 and 2.6 years of revenue on the books. ABB (0.81×), Shilchar (0.80×) and Voltamp (0.53×) have headroom to absorb incremental orders.
How big is the L&T–Together AI order?
₹10,000–15,000 crore ($1.05–1.57 billion), covering a Chennai AI campus with 10,000 NVIDIA B300 chips — L&T calls it India's largest single-cluster AI infrastructure. It is one project in a larger pipeline: India's committed data-centre capex was about $126 billion at end-2025, projected near $180 billion in 2026.
Figures are drawn from company quarterly filings, concall transcripts, annual reports and DGTR/news coverage retrieved on 2026-08-15/16 via Cofacto's research platform. Order books are as of slightly different dates (mostly Q1 FY27 end; Voltamp entry-FY27; Blue Star Jun-30-2026). Coverage ratios (order book ÷ TTM revenue) are computed from those disclosed figures; Blue Star's DC-MEP coverage of ~1.1× is an estimate derived from two disclosed numbers (Mar-26 carry-forward vs guided FY27 DC revenue), not a reported figure, and ABB's DC order-book slice (₹1,320–1,430 Cr) is derived from its disclosed 12–13% share. No retrieved document ties any listed supplier to the specific L&T–Together AI order; all linkages are inferred from each company's own data-centre order books and commentary.
Disclaimer: This is historical/descriptive analysis of publicly available information, for educational purposes only. It is not investment advice and was not prepared by a SEBI-registered Research Analyst. Nothing here is a recommendation to buy, sell or hold any security.