The number that changes the Lightstorm view is not India's next gigawatt forecast. It is Rs. 78 crore of billed monthly recurring revenue. CARE Ratings says Lightstorm Telecom Connectivity's billed MRR reached about Rs. 78 crore as of December 31, 2025, up from Rs. 49 crore a year earlier, while residual total contract value rose 43 per cent to Rs. 2,870 crore after new data-centre interconnect contracts with Microsoft, Meta and Amazon. The same March 2026 note says total operating income reached Rs. 589 crore in 9MFY26, profit before interest, lease rentals, depreciation and tax margin reached 56.8 per cent, and the weighted average contract expiry was about six years: https://www.careratings.com/upload/CompanyFiles/PR/202603140317_Lightstorm_Telecom_Connectivity_Private_Limited.pdf. Those numbers move the company out of the realm of fibre-route claims and into the economics of contracted control: large buyers are already paying recurring money for the ability to move data between Indian sites, clouds and data-centre campuses.

That is the purchase order behind the AI story. A bank in Mumbai wants to run a fraud model across card transactions without sending its most sensitive data through a public internet path. An insurer in Gurgaon wants claims documents read by an AI system while the original files remain inside a controlled Indian environment. A streaming platform wants recommendation systems and cache refreshes to move between a coastal data-centre campus and a second site without waiting for a cross-connect order that takes weeks. A retailer wants its customer data lake close to public cloud tools, but not so close that every transfer becomes an uncontrolled bill. In each case the glamorous object is the model, the GPU cluster or the cloud service. The purchase decision underneath is more prosaic: can the enterprise get powered racks in the right Indian metro and fibre paths that behave like an operating instrument rather than a construction project?

This is where Lightstorm matters. Lightstorm is not best understood as another company joining India's data-centre publicity boom. The more interesting economic role is narrower and more powerful. Lightstorm sits in the control layer between AI-ready racks and the physical limits of the cities where those racks can be powered, cooled, connected and contracted. Its public material says it connects clouds, data centers, GPUs and edge locations globally in minutes, and lists 30,000-plus kilometres of terrestrial fibre, 21,000-plus kilometres of subsea fibre, more than 120 connected data centres and 3,000-plus Tbps of deployed capacity on its home page: https://www.lightstorm.net/. Its product language is not about renting a generic server cabinet. It is about data-centre interconnect, cloud connect, internet-exchange access, CDN access, virtual network edge and optical capacity. The question is whether that fibre-control position can keep earning a premium as powered racks become more valuable and hyperscalers become more demanding.

The entity name in focus, Lightstorm Data Centers Private Limited, should be handled with discipline because the public evidence is strong but split across the group. A Lightstorm environmental and social management document says Lightstorm Data Centers Private Limited was established in 2019, headquartered in Gurgaon, Haryana, and 100 per cent owned by I Squared Capital. It describes the business as operation and management services in the data-centre field, including data-centre equipment, colocation, managed hosting, IT infrastructure, remote help desk, remote technical assistance centre, remote network monitoring and related services: https://www.lightstorm.net/assets/downloads/ESMS_LDC_India_Final.pdf. The Government of India DPIIT active-foreign-subsidiaries list also shows Lightstorm Data Centers Private Limited as active and a subsidiary of a foreign company: https://www.dpiit.gov.in/static/uploads/2025/07/973f25fda7999f8cfef4c1f1de024ea7.pdf. A public Wint Wealth bond listing for Lightstorm Data Centers Private Limited shows a private issue with face value of Rs. 100, coupon of 9.13 per cent, issue size of Rs. 40 crore and ISIN INE0I3A08019, while not displaying a public credit rating: https://www.wintwealth.com/bonds/lightstorm-data-centers-private-limited/ine0i3a08019/. That is a direct financing clue for the named company, not enough for a standalone valuation.

The customer-facing Lightstorm brand and the richest operating evidence still run through Lightstorm Telecom Connectivity Private Limited, SmartNet and Polarin. The site footer uses Lightstorm Telecom Connectivity Pvt Ltd. The rating reports are for Lightstorm Telecom Connectivity. The network records include Lightstorm Data Centers on one autonomous system and Lightstorm Telecom Connectivity on another. A fair reading is that Lightstorm Data Centers is part of a wider Lightstorm infrastructure platform whose economics cannot be read only from rack inventory. That distinction matters because India's AI data-centre story is already crowded with operators that own or develop large powered campuses: STT GDC India, NTT, Sify, CtrlS, Nxtra, Yotta, AdaniConneX, Colt DCS, Digital Connexion, Equinix and others. Lightstorm's more defensible position is not to out-build all of them in megawatts. It is to make those megawatts useful to customers whose applications, disaster-recovery arrangements, cloud exits and AI workloads span more than one building.

The product evidence matches that reading. The home page calls SmartNet the foundation layer and Polarin the control layer, and describes a low-jitter, loss-optimised transport fabric for data centres, cloud hubs and distributed AI zones: https://www.lightstorm.net/. The data-centre interconnect product page claims more than 120 connected data centres, six countries, 10G, 100G and 400G bandwidth availability, 99.99 per cent metro uptime and lower latency: https://www.lightstorm.net/offerings/data-center-interconnect-wave. The public product catalogue describes Polarin DCI Wave as an on-demand Layer 1 wave service across more than 60 data centres in India, with SmartNet DCI supporting ultra-low-latency private networks connecting data centres across India and Indonesia: https://www.lightstorm.net/products. Lightstorm's own platform page says the network is built to connect clouds, data centers, GPUs and edge locations; CARE's customer list names Amazon, Meta, Microsoft and Google among hyperscaler relationships and Apple, Bloomberg, Flipkart, HDFC, ICICI and PhonePe among enterprise relationships. That combination is the company-specific evidence that matters: cloud buyers, banks, payment firms and software-heavy enterprises need both powered sites and controlled paths between them.

The opening enterprise workload therefore creates two linked scarcity problems. First, the enterprise needs powered data-centre capacity in the metros where customers, cloud regions, landing stations and regulated data needs concentrate. JLL's India Data Centre Market Dynamics report for H1 2025 put total inventory at 1,123 MW of IT load capacity, net take-up at 97.9 MW in the first half and vacancy at only 4.3 per cent: https://www.jll.com/en-in/insights/market-dynamics/india-data-centers. CBRE's 2025 India report said operational stock had reached about 1,530 MW, or 23 million square feet, by the first nine months of 2025, with nearly 90 per cent of existing capacity anchored in Mumbai, Chennai, Delhi-NCR and Bengaluru: https://www.cbre.co.in/insights/reports/india-s-data-centre-market-in-a-new-era. Savills' H1 2025 market watch put operator-based stock at 1,332 MW IT, supply additions at 162 MW IT and absorption at 212 MW IT, led by Mumbai and Chennai: https://pdf.savills.asia/asia-pacific-research/india-research/data-centre-market-watch--h1-2025.pdf. Those numbers matter here only because a scarce powered megawatt is less useful if it cannot be connected quickly, privately and resiliently to the rest of the workload.

Second, the enterprise needs the right network path between that powered capacity and the cloud or second site. A GPU cluster is not a self-contained island if the training data, vector stores, user requests, billing systems, disaster-recovery copy and model outputs move between premises, colocations and public-cloud regions. The cost is not only bandwidth. It is time-to-connect, jitter, tail latency, maintenance coordination, route diversity, visibility and the ability to scale for a temporary workload without buying a permanently overbuilt circuit. Lightstorm's Polarin DCI Wave page says dedicated optical connectivity can be provisioned in ten minutes and claims 99.99 per cent uptime: https://polarin.lightstorm.net/polarin-dci-wave-service. The Polarin services page says the platform supports cloud-to-cloud, data-centre-to-cloud and internet-exchange connectivity, with dedicated connections from 10 Mbps to 100 Gbps: https://polarin.lightstorm.net/services. That is the commercial promise: turn a slow engineering and procurement cycle into a consumable network service.

The same CARE note explains why the infrastructure economics are different from a conventional fibre trenching story. It says Lightstorm developed much of its utility-grade fibre network through long-term lease arrangements with power transmission utilities, gas-transmission operators and other infrastructure bodies for their right-of-way corridors, involving upfront lease payments and lower ongoing operating costs. It also says much of the optical fibre runs overhead on power transmission lines and is less prone to cuts than terrestrial networks: https://www.careratings.com/upload/CompanyFiles/PR/202603140317_Lightstorm_Telecom_Connectivity_Private_Limited.pdf. TelecomTV reported the same basic model in 2022, describing a fibre network across major Indian economic hubs, connected to multi-tenant data centres and running overground along power-grid infrastructure rather than through conventional digging: https://www.telecomtv.com/content/access-evolution/lightstorm-set-to-disrupt-india-s-data-transport-sector-44439/. Ciena's customer story said Lightstorm introduced a 12,000 km SmartNet network across five major Indian cities for hyperscalers and digital enterprises, using a utility-grade fibre design to support high availability and low latency: https://www.ciena.com/about/customer-stories/lightstorm-gets-smartnet-right-the-first-time-with-ciena-services.

This route strategy is the first key economic advantage. In Indian metros, physical execution is not a detail. Road cuts, permissions, rights of way, crowded ducts, last-mile coordination and repair windows can eat the margin that a wholesale transport provider thought it had. If a company can lease existing infrastructure corridors, light capacity as demand rises and connect a growing set of data centres, it may avoid some of the capital and disruption that burdens a conventional build. The advantage is not free. Upfront lease payments still consume capital, operational coordination depends on lessors, and CARE notes that repair and maintenance for leased network is undertaken by the respective lessors, including state power transmission utilities and public sector undertakings. But the model changes the payback calculation. Lightstorm is not necessarily betting that every future customer will rent a rack from its own data-centre company. It is betting that many customers will need to move data between powered racks, clouds and sites, and that the owner of a controlled fibre fabric can participate in that spend.

India's data-centre boom makes that bet more plausible, but not automatically more valuable. The Council on Energy, Environment and Water's February 2026 white paper said India's domestic data-centre capacity had risen from about 520 MW in 2020 to almost 1.5 GW by mid-2025 and could reach 4.5 GW to 6.5 GW by 2030, with committed investments from 2019 to 2025 around USD 95 billion and expected to exceed USD 100 billion by 2027: https://www.ceew.in/sites/default/files/ceew-data-centre-study-web-ready-final.pdf. Cushman & Wakefield's India H1 2025 update put operational supply at about 1.3 GW and upcoming supply by 2030 at about 2.9 GW, while noting that Mumbai is a leading APAC data-centre hub and that Chennai, Delhi NCR and Hyderabad are also emerging: https://assets.cushmanwakefield.com/-/media/cw/apac/india/insights/indiadatacentreupdateh12025v4.pdf?rev=4a5dd53b2388421ea3935afb03302def. The numbers point toward growth, but the value of Lightstorm's fabric depends on where that growth lands and whether the new facilities need neutral interconnection rather than captive carrier arrangements.

AI changes the shape of the network bill. Traditional enterprise colocation often meant a primary site, a disaster-recovery site and enough public internet or MPLS capacity to support users. AI workloads push more east-west movement: data lake to GPU cluster, inference endpoint to application tier, training run to storage, safety review to archive, model serving to edge cache. The data-centre interconnect product page explicitly names distributed AI training and inference workloads, large-scale data replication, backup and disaster recovery, test and development environments across regions, high-capacity long-haul interconnects and multipath design: https://www.lightstorm.net/offerings/data-center-interconnect-wave. This is not accidental positioning. For an enterprise, an AI workload can make an old network contract look like a bottleneck. For Lightstorm, AI is useful because it turns connectivity from an afterthought into a board-visible dependency.

The powered-rack constraint is equally important. CEEW says power and water availability, climate risk, connectivity, utilities and scalability are central to site selection, and that stable uninterrupted 24/7 power is fundamental to data-centre operations: https://www.ceew.in/sites/default/files/ceew-data-centre-study-web-ready-final.pdf. The same report says electricity costs and reliability dominate operational considerations, often outweighing land, fiscal incentives or labour availability, and that energy costs typically represent 60 to 70 per cent of data-centre operating cost. It also warns that tight uptime and latency guarantees drive conservative power-system design, that diesel remains hard to eliminate, that storage will become more important, and that grid congestion in major clusters constrains renewable procurement. These are not abstract sustainability notes. They are pricing inputs. A powered rack in Navi Mumbai or Chennai is a scarce asset because it sits at the intersection of grid connection, land, cooling, fibre and permits.

Lightstorm benefits from that scarcity only if fibre is scarce in the same decision. If every new data-centre campus came with abundant, neutral, low-latency connectivity to all clouds, all exchanges and all peer sites, the fibre-control premium would compress. India's reality is messier. CBRE told ETTelecom in 2025 that policy interventions to ensure land, water and energy availability would decide the proliferation of data centres, and that the bigger challenge than land was getting the power required to run a data centre: https://telecom.economictimes.indiatimes.com/news/internet/data-centers-future-hinges-on-land-and-power-policies-insights-from-cbres-anshuman-magazine/123298011. CEEW says connectivity-related factors such as access to optical fibre networks, proximity to subsea landing stations and latency requirements are critical to high-speed, low-latency operations. Savills says H1 2025 absorption was concentrated in Mumbai and Chennai, with hyperscalers driving much of the demand: https://pdf.savills.asia/asia-pacific-research/india-research/data-centre-market-watch--h1-2025.pdf. Concentration creates load and competition; it also creates the need for alternatives, route diversity and fast provisioning.

The second key advantage is investor backing. Lightstorm is I Squared Capital's kind of infrastructure platform: difficult to build, tied to long contracts, useful to large counterparties and capable of expansion beyond one country. I Squared's own digital portfolio page describes Lightstorm as a provider of high-capacity fibre infrastructure interconnecting data centres throughout Asia, the Middle East and the Pacific: https://isquaredcapital.com/txnm_sector/digital/. Mint reported in 2019 that I Squared planned to invest around USD 300 million in Lightstorm Telecom Ventures to set up a carrier-neutral telecom infrastructure platform in India: https://www.livemint.com/industry/telecom/i-squared-cap-to-invest-300-million-in-telecom-infra-platform-lightstorm-1561656887496.html. CARE says I Squared has cumulatively infused about Rs. 1,776 crore in Lightstorm Telecom Connectivity through debt and equity as of December 31, 2025, and that the promoter Asia Connectivity Private Limited has given undertakings around debt support and I Squared retaining control: https://www.careratings.com/upload/CompanyFiles/PR/202603140317_Lightstorm_Telecom_Connectivity_Private_Limited.pdf. That backing reduces financing risk, but it also creates return expectations.

The pressure from those expectations is visible in market signals. Moneycontrol reported in March 2026, citing multiple unnamed industry sources, that I Squared was evaluating a minority stake sale of around 20 per cent in Lightstorm at a valuation of up to USD 1.5 billion, with a possible public listing in 18 to 24 months targeting a higher valuation. The report said Lightstorm and I Squared declined to comment: https://www.moneycontrol.com/news/business/companies/i-squared-capital-weighing-part-stake-sale-in-digital-infra-platform-lightstorm-at-up-to-1-5-bn-valuation-13850766.html. That report should not be treated as completed transaction evidence. It is a useful market signal. Infrastructure investors do not want fibre platforms merely to grow route kilometres; they want visible contracted revenue, utilisation, margins, leverage control and a credible exit multiple. Lightstorm's CARE report provides some of that proof. It also shows the test: the business must keep adding long-tenure contracts and avoid debt-funded expansion that weakens leverage.

NIIF Infrastructure Finance's 2025 funding added a domestic institutional stamp. Lightstorm announced Rs. 700 crore of funding from NIIF Infrastructure Finance Limited to support India growth, saying it had built a 30,000-plus km fibre network across India with around 700 Tbps of capacity and about 100-plus customers: https://www.lightstorm.net/resources/news-pr/ctg4km92onuttjjbwivprogq. The same announcement said funds would be used to expand network footprint, enhance operational capabilities and meet the needs of large enterprises. This is important because India's data-centre story is sometimes presented as a foreign-hyperscaler story. Lightstorm's financing mix shows a more local infrastructure-bank logic: Indian institutional capital is willing to finance the connectivity layer that lets cloud and AI demand become usable domestic infrastructure.

Hyperscalers are both the best proof and the main risk. CARE's customer list includes Microsoft, Meta, Amazon and Google, and says new contracts with Microsoft, Meta and Amazon drove the 43 per cent rise in residual total contract value. That is strong evidence of relevance. It also creates bargaining exposure. Hyperscalers buy in volume, compare vendors aggressively, demand resilience, negotiate service credits, and may shift traffic to their own network or another carrier when economics change. They also have their own cloud on-ramp and subsea strategies. A fibre platform can earn attractive margins when it controls a difficult local path, but it cannot assume hyperscaler loyalty. The CARE report identifies contract renewal risk because the term loan tenure is longer than the weighted average contract expiry of about six years: https://www.careratings.com/upload/CompanyFiles/PR/202603140317_Lightstorm_Telecom_Connectivity_Private_Limited.pdf. The credit question and the strategic question are the same: what happens when today's large contracts come up for renewal in a more competitive market?

Enterprise demand is the counterweight to hyperscaler bargaining. Lightstorm's public materials show customer or trust signals involving PhonePe, AirPay, Cisco, Equitas and others on the home page: https://www.lightstorm.net/. CARE names Apple, Bloomberg, Flipkart, HDFC, ICICI and PhonePe among large enterprise clients. Polarin's cloud-connect positioning says customers can connect with major public cloud providers including AWS, Azure, Google Cloud and Oracle Cloud: https://polarin.lightstorm.net/services. Lightstorm's IX and CDN page says its service gives access to internet exchanges and content-delivery networks to reduce latency, improve end-user experience and reduce transit costs: https://www.lightstorm.net/offerings/IX-CDN. This enterprise surface matters because a network platform with only hyperscaler demand becomes a wholesale supplier under strong buyer power. A platform that also serves banks, fintechs, retailers, media, insurers, gaming companies and software firms can price more around complexity, speed and assurance.

The operating proof around service delivery is not only in press material. The TL 9000 public profile for Lightstorm Telecom Connectivity lists a certification valid until March 31, 2027, with scope covering provisioning, delivery and customer support services including network operation centre functions for transmission and IP services of Lightstorm Data Centers Private Limited. It lists profile locations in Navi Mumbai and Noida: https://portal.questforum.org/tl9000/public_profile.jsf?tlid=7485. Lightstorm's careers page shows roles across Polarin, sales, network operations and NOC, with locations including Mumbai, Delhi, Gurugram, Hyderabad, Chennai, Noida and international cities; a NOC business role describes business development around NOC services, while an IP operations role describes proactive network monitoring and uptime maintenance: https://resources.lightstorm.net/careers/. Job pages are not revenue proof, but they are useful operating signals. A company claiming to automate network control still needs support labour, network operations, enterprise sales and compliance capacity.

Network-resource evidence is mixed and therefore interesting. PeeringDB lists AS142505 as LIGHTSTORM DATA CENTERS with organisation LIGHTSTORM and a company website override that appears as http://www.lighstorm.net, while showing zero IPv4 and zero IPv6 prefixes: https://www.peeringdb.com/net/28671. BGP.Tools shows AS142505 as LIGHTSTORM DATA CENTERS PVT LTD, active and allocated under APNIC, also with zero originated prefixes: https://bgp.tools/as/142505. That is evidence of a public resource record, not evidence of a live production backbone. By contrast, PeeringDB lists AS135709 for Lightstorm Telecom Connectivity Private Limited with IRR set AS135709:AS-LTCIN and prefix counts, while BGP.Tools shows AS135709 active with originated IPv4 and IPv6 prefixes, multiple peers and upstreams: https://www.peeringdb.com/net/34243 and https://bgp.tools/as/135709. IPinfo also shows ranges associated with AS135709 and Lightstorm Telecom Connectivity Private Limited: https://ipinfo.io/AS135709. For the article's thesis, the operating network evidence sits more heavily with Lightstorm Telecom Connectivity and the Lightstorm group fabric than with AS142505 alone.

The regional expansion evidence also matters because Indian AI workloads are not purely domestic. Lightstorm announced in January 2024 that it had signed asset purchase agreements to acquire three submarine cable assets, including JGA-North and significant portions of JGA-South and SEA-US, with about 21,000 km of total length and about 64 Tbps of allocated capacity connecting the US, Guam, Japan and Australia: https://www.lightstorm.net/resources/news-pr/lxb3jxqz2n6wjduucfao17lr. In May 2025, Lightstorm said it had completed integration of RTI Cables' assets, adding more than 75 Tbps of capacity, 25-plus data-centre connections and coverage in Tokyo, Sydney, Brisbane, Guam, Hawaii, Los Angeles and Singapore: https://www.prnewswire.com/news-releases/lightstorm-completes-integration-of-rti-cables-assets-to-strengthen-data-center-connectivity-in-asia-australia-and-the-us-302444930.html. The Lightstorm Asia AS152144 records on PeeringDB and BGP.Tools show a Singapore-registered network surface with peers, upstreams and prefixes: https://www.peeringdb.com/asn/152144 and https://bgp.tools/as/152144. For customers, this turns Lightstorm from a domestic DCI provider into a possible Asia-Pacific interconnection platform.

That Asia-Pacific story is attractive, but it raises execution risk. Integrating distressed subsea cable assets is not the same as lighting a metro wave between two Mumbai facilities. Cross-border operations involve permits, landing-station rules, maintenance windows, currency exposure, partner networks and geopolitical route preferences. The advantage is that global AI and cloud buyers increasingly care about secure, low-latency corridors that avoid single points of regional dependence. The risk is that global expansion can consume management attention and capital before the Indian demand curve fully matures. CARE says total debt was Rs. 2,102 crore as of December 31, 2025, including Rs. 905 crore of unhedged foreign-currency promoter external commercial borrowings, or 43 per cent of total debt; it also says liquidity was supported by Rs. 368 crore of cash and bank balance, a Rs. 20 crore debt-service reserve account and expected external annual debt repayments of Rs. 80-110 crore during the projected period: https://www.careratings.com/upload/CompanyFiles/PR/202603140317_Lightstorm_Telecom_Connectivity_Private_Limited.pdf. That is the debt-service version of the same unit-economics test. A company selling Indian fibre services but carrying a meaningful foreign-currency funding exposure has to make sure each route, data-centre connection and service layer converts into cash fast enough to absorb currency and refinancing shocks.

Competition is not a footnote. CARE names Tata Communications and Bharti Airtel as large established players with significant terrestrial and undersea cable capacities: https://www.careratings.com/upload/CompanyFiles/PR/202603140317_Lightstorm_Telecom_Connectivity_Private_Limited.pdf. Sify, Jio, Airtel Nxtra, Tata Communications, NTT and others can all bundle network, cloud, data-centre or enterprise services in different ways. Data-centre operators can develop richer interconnection ecosystems inside campuses. Internet exchanges can reduce transit costs and improve local performance. Global cloud providers can sell direct-connect services and partner fabrics. Lightstorm's defence is neutrality, route design, automation and speed. Its vulnerability is that every large customer wants those same qualities at a lower price once the market matures.

The unit economics turn on utilisation of routes attached to scarce powered capacity. The CARE report's margin improvement is explicitly linked to operating leverage and better capacity utilisation. That is the characteristic infrastructure-fund bargain: spend or commit capital early, win long contracts, then let incremental traffic carry high contribution margins. But utilisation is not a single number. A fibre pair on a route with no customer is underused. A route with one hyperscaler is exposed. A route connecting several powered data-centre sites, cloud on-ramps, internet exchanges, backup locations, gaming traffic, banks and payment companies is a portfolio. Lightstorm does not need to own every megawatt for this to work; it needs enough control around the network paths between megawatts that customers pay for reliability, speed of change and route diversity. Its economic quality improves as the same physical fabric supports more products and more counterparties: DCI, cloud connect, IX/CDN access, virtual network edge, NOC services, global DCI and subsea corridors.

This is why the company keeps emphasising automation. Manual cross-connects and old provisioning cycles create labour cost, error risk and customer frustration. Lightstorm's August 2024 Polarin DCI Wave announcement said enterprises and hyperscalers could self-provision optical layer connectivity in under ten minutes across more than 60 top data centres in India, using a partnership with FiberSmart and robotic optical switching: https://resources.lightstorm.net/lightstorm-announces-polarin-dci-wave-the-worlds-first-l0-l1-naas/. The announcement said the product replaces day-to-day manual cross-connection with remote robotics and automates the physical network. The claim should be tested in customer operations, not accepted as magic. But if it works, the economic effect is real: lower provisioning friction, faster revenue activation, fewer manual errors and stronger customer control over short-term network needs.

The enterprise AI use case makes that automation more than a convenience. AI workloads are lumpy. A bank may need extra bandwidth during a model refresh, a retailer during a festival-sale demand surge, a media company during a tournament, and a software company during a migration window. Traditional leased-line economics make the customer choose between overbuying permanent capacity and waiting through slow upgrade cycles. Polarin's public services page says users can scale dedicated connections from 10 Mbps to 100 Gbps and manage hybrid and multi-cloud architectures dynamically: https://polarin.lightstorm.net/services. The commercial promise is pay for control rather than idle headroom. For customers, that can turn a capital-heavy network plan into a more flexible operating decision. For Lightstorm, it can turn the same fibre asset into more granular, higher-yield services.

Power and land constraints make this even more important. If India's best powered sites are scarce, customers will not always get the perfect primary and secondary facility combination. They may need one site in Mumbai, another in Chennai, a cloud region elsewhere, a disaster-recovery option in Delhi-NCR or Hyderabad, and an overseas path for global services. CEEW says Mumbai leads the market by number of data centres, followed by Chennai, Hyderabad and Bengaluru, and notes that hyperscaler facilities in India are often located in coastal regions such as Mumbai and Chennai because of landing stations, seawater cooling and undersea connectivity: https://www.ceew.in/sites/default/files/ceew-data-centre-study-web-ready-final.pdf. Savills says Mumbai contributed 61 per cent of H1 2025 absorption and Chennai 26 per cent: https://pdf.savills.asia/asia-pacific-research/india-research/data-centre-market-watch--h1-2025.pdf. This concentration does not remove the need for other cities. It creates the need to connect them with predictable performance.

One should be careful, though, not to overstate Lightstorm as a pure winner from every data-centre constraint. If power shortages delay new data-centre supply, there are fewer new racks to connect. If hyperscalers pre-commit entire campuses and build their own private network arrangements, neutral DCI opportunities may be narrower. If state incentives pull new capacity into locations where Lightstorm's network is less dense, it must invest ahead of demand. If data-centre operators bundle interconnect inside their own campuses and prefer incumbent carriers, Lightstorm has to win access and trust site by site. The company benefits from India's rack boom only when the boom produces multi-site, multi-cloud, multi-customer network demand that its fabric can serve.

The legal and ownership layer is also not fully transparent from public material. The Lightstorm Data Centers environmental document states 100 per cent I Squared ownership of that company. CARE describes Lightstorm Telecom Connectivity as wholly owned by private-equity investors through I Squared via Asia Connectivity Private Limited. The public site describes a group with India operations, APAC leadership and global assets. Tracxn and IndiaFilings pages add corporate details for Lightstorm Data Centers Private Limited, including incorporation and directors, but those are secondary directories rather than statutory filings. IndiaFilings lists incorporation on March 25, 2019, Delhi registrar status and an active status: https://www.indiafilings.com/search/lightstorm-data-centers-private-limited-cin-U72900DL2019FTC347596. The market analysis does not need to resolve every legal shell. It should treat Lightstorm Data Centers as a real company in the Lightstorm platform and make clear that the strongest revenue and contract evidence currently comes from Lightstorm Telecom Connectivity.

There is a valuation implication in that split. If investors value Lightstorm as a pan-Asia digital infrastructure platform, they will look at the combined platform: Indian DCI contracts, fibre footprint, subsea capacity, global DCI, customer quality, growth in recurring revenue and expansion into countries such as Indonesia. If a buyer is evaluating only Lightstorm Data Centers Private Limited, the standalone question is narrower: what assets, contracts, service obligations and revenues sit inside that legal entity? The public evidence does not disclose enough to price that standalone company precisely. The economic essay therefore has to be about the Lightstorm platform around Lightstorm Data Centers, with the caveat that legal-entity cash flows require further documentation.

The strongest bullish argument is that Lightstorm has inserted itself into the exact scarcity India is creating. Data-centre demand is rising because cloud, AI, mobile data, fintech, e-commerce, digital public infrastructure and enterprise modernisation all point toward more domestic compute. Powered racks are scarce because land, grid connections, cooling and approvals concentrate in a few metros. Hyperscalers and large enterprises need private, low-latency, resilient paths between sites. Lightstorm has a fibre model designed around power corridors, a product layer designed around fast provisioning, a customer base that includes the biggest cloud buyers and Indian enterprises, and investor backing from a specialist infrastructure owner. If utilisation keeps rising, margins can expand faster than revenue because the physical fabric carries more services.

The strongest bearish argument is that Lightstorm's success invites buyer pressure and capital ambition. Hyperscalers are sophisticated counterparties, not passive tenants. They can renegotiate, dual-source, build, partner or shift demand. The CARE rating itself flags renewal risk and competition. The business still depends on leased infrastructure and repair arrangements with outside lessors. It carries foreign-currency exposure. It needs to prove that automation claims produce customer retention and not only conference-stage excitement. It may have to spend ahead of demand in new markets. And if I Squared does seek a stake sale or listing path, the company may face pressure to keep reporting expansion metrics even when the better economic move is to wait for utilisation.

The facts that would change the view are specific. First, a public breakdown of residual contract value by customer type would show whether Lightstorm is balanced between hyperscalers and enterprises or concentrated in a handful of cloud buyers. Second, renewal outcomes on the Microsoft, Meta, Amazon and other large contracts would prove whether the six-year weighted average contract expiry is a bridge to durable cash flow or a cliff. Third, utilisation by route and product would show whether the same fibre supports a diversified service portfolio. Fourth, legal-entity financials for Lightstorm Data Centers Private Limited would clarify whether the data-centre company is a meaningful revenue centre or a service/holding layer inside the group. Fifth, proof of completed Polarin DCI Wave customer deployments at scale would validate the automation premium. Sixth, evidence of power-grid or right-of-way disruption would test the leased-corridor model. Seventh, a completed minority stake sale at the rumoured valuation, or a failed process, would reveal how infrastructure investors price the platform after the NIIF funding and CARE upgrade.

There is also a public-policy dimension. India wants data sovereignty, AI capability and cloud infrastructure, but the CEEW paper warns that today's siting, power sourcing and cooling decisions will lock in land, energy and water impacts for decades: https://www.ceew.in/sites/default/files/ceew-data-centre-study-web-ready-final.pdf. A fibre company is not exempt from that policy environment because it makes data-centre capacity more usable. If India tightens reporting, energy-performance expectations or grid-connection rules for large digital infrastructure, the companies that connect powered sites will become part of the same resilience conversation. Lightstorm's advantage may grow if it helps customers use multiple sites efficiently rather than overbuilding in one constrained cluster. It may shrink if policy delays slow the creation of the new sites it expects to interconnect.

For an Indian enterprise buyer, the practical comparison is not fibre versus no fibre. It is Lightstorm's controlled fabric versus a bundle of incumbent carrier circuits, cloud-provider private links, data-centre cross-connects and internal network labour. The buyer has to ask how long a new connection takes, how predictable the route is, what happens when capacity is needed for a temporary AI training run, whether cloud egress and private-connect charges are understood, how quickly a second site can be added, and whether the provider can show operational visibility rather than only monthly invoices. The answer may not always favour Lightstorm. A single-site workload with modest bandwidth and little compliance pressure may accept a cheaper conventional link. A workload that touches a GPU cluster, a bank data lake, a public-cloud region and a disaster-recovery copy is different. For that buyer, delay and opacity have costs.

The same comparison matters to lenders and future equity buyers. A plain fibre company is often valued on route kilometres, contracted revenue, utilisation, debt and renewal risk. A data-centre interconnect platform can claim a richer multiple if it looks like a control plane for cloud and AI demand rather than a passive route owner. CARE's report gives the platform side of the argument: recurring monthly revenue increased, residual total contract value expanded, margins improved, customer quality is strong, and capex intensity is expected to fall: https://www.careratings.com/upload/CompanyFiles/PR/202603140317_Lightstorm_Telecom_Connectivity_Private_Limited.pdf. But buyers will still discount the business if growth depends on a small set of hyperscaler contracts or if the promised automation does not translate into renewal power. The difference between a high-quality infrastructure platform and a crowded telecom services company is visible when contracts renew.

This is why the Polarin layer is more than brand architecture. Fibre routes can be copied slowly by well-funded rivals; a dense set of data-centre relationships can be challenged by an incumbent carrier; a subsea asset can be repriced if international capacity becomes abundant. What is harder to copy is a habit. If enterprise network teams learn to treat Lightstorm as the place where they create, scale, observe and modify private data-centre and cloud connections, switching costs rise. The product then becomes part of daily workflow, not just an invoice for wavelengths. The public pages say users can get real-time visibility, scale bandwidth, choose flexible terms and avoid months of waiting: https://www.lightstorm.net/ and https://polarin.lightstorm.net/polarin-dci-wave-service. The economic test is whether those claims become repeat customer behaviour at enough scale to offset the bargaining power of cloud giants.

India's geography reinforces the same logic. Mumbai has cable landings, financial-sector demand and a huge concentration of capacity. Chennai has coastal access, cloud demand and a southern enterprise base. Hyderabad has policy momentum, real-estate availability and a deep technology workforce. Delhi-NCR has government, enterprise and northern-market demand. Bengaluru has software concentration but faces real-estate and power pressures. A single national AI workload may touch several of these markets even if the company headquarters sits in one. Cushman & Wakefield's H1 2025 update describes Mumbai, Chennai, Delhi NCR and Hyderabad as major or emerging hubs, while Savills' H1 2025 figures show the absorption weight of Mumbai and Chennai: https://assets.cushmanwakefield.com/-/media/cw/apac/india/insights/indiadatacentreupdateh12025v4.pdf?rev=4a5dd53b2388421ea3935afb03302def and https://pdf.savills.asia/asia-pacific-research/india-research/data-centre-market-watch--h1-2025.pdf. Lightstorm's addressable market is the movement between those hubs, not merely their individual growth.

There is a further operating nuance. AI demand can make the value of low latency visible, but it can also make network planning unstable. A customer may not know whether its next workload needs a few extra 10G links, a 100G connection, a short-term migration burst or a more permanent multi-site design. If the provider forces every change into a custom project, the customer's AI programme inherits telecom delay. If the provider makes capacity appear too easy, it may create expectations that the physical network cannot always satisfy. Lightstorm's automation story has to walk that line. The company must make connectivity feel fast while preserving the engineering discipline that keeps routes resilient and service levels credible.

The most attractive version of the business is therefore not a pure growth story. It is a discipline story. Lightstorm should benefit from India's data-centre growth, but only if it chooses routes and products where it can earn repeatable control premiums. It should benefit from AI, but only if AI workloads create sustained high-capacity movement rather than short bursts followed by idle links. It should benefit from hyperscaler contracts, but only if enterprise and cloud-native demand reduce counterparty concentration over time. It should benefit from I Squared backing, but only if financial discipline remains stronger than the temptation to chase every new geography. Those conditions make the company worth watching precisely because the outcome is not guaranteed.

The image of an Indian enterprise AI workload needing both metro fibre and powered racks is therefore not a metaphor. It is the purchase order behind the next phase of Indian digital infrastructure. A model that cannot reach its data is a stranded asset. A rack that cannot get reliable power is a liability. A data centre that cannot connect privately to cloud, backup, exchanges and second sites is a warehouse of expensive heat. Lightstorm's opportunity is to make those pieces behave like one controlled system. Its risk is that the largest buyers will treat that control as a commodity once enough capacity exists.

Another way to read the company is through the build-versus-buy decision inside a large Indian technology team. An enterprise can try to assemble its own network relationships, negotiate separately with each data-centre operator, order private cloud links, manage cross-connects, watch latency, co-ordinate maintenance and keep enough engineers available to make changes at short notice. That may be rational for a very large cloud-native firm with deep network talent. It is less rational for a bank, insurer, retailer, payments company, broadcaster or industrial group whose strategic problem is the application and the data, not the craft of operating optical connectivity. Lightstorm's product is attractive when it removes the need for the customer to become a miniature carrier.

The value proposition becomes sharper when the workload is not static. Enterprise AI programmes rarely arrive as a clean capacity forecast. They begin with experiments, then a regulated deployment, then a second model, then a higher traffic period, then a security review, then a demand from the business to connect a new site because the first site is full or the risk team wants redundancy. In that environment, procurement delay is not just frustrating. It changes which projects are attempted. A network service that can make controlled connectivity more predictable may enlarge the set of AI and data projects that Indian enterprises are willing to run domestically.

This also explains why Lightstorm's best customers are not only the companies with the largest bandwidth needs. They are the customers whose internal politics make predictable connectivity valuable. A chief information officer may be able to approve a cloud proof of concept quickly, but a production AI workload that touches financial data, customer identity, regulated records or critical operations brings in security, compliance, finance and business continuity teams. Those teams do not want a heroic one-off network build. They want repeatable controls, renewal clarity, route visibility and a counterparty that can explain what is happening. If Lightstorm can make network change feel governed rather than improvised, it earns more than a transport margin.

That is the strategic ceiling of the business. It can be a neutral fibre provider with good contracts, or it can become part of the operating discipline of Indian AI infrastructure. The first version is valuable. The second version is harder to displace. The difference will show up in contract breadth, renewal rates, enterprise customer count, use of multiple products per customer and the share of revenue that comes from automated, repeatable services rather than bespoke circuits. Those figures are not fully public today, which is why the conclusion should remain conditional. But they are the figures that would tell whether Lightstorm is simply riding India's data-centre boom or helping shape how that boom is consumed.

On the available evidence, Lightstorm is one of the more economically legible private infrastructure stories in India's AI and data-centre market. It has real group ownership, visible financing, public product surfaces, credible customer names, rating-agency financial evidence, a distinctive route model, Asia-Pacific expansion and network-resource records. It is not a simple data-centre landlord, and the exact Lightstorm Data Centers Private Limited cash flow remains less visible than the Lightstorm Telecom Connectivity evidence. But that ambiguity does not weaken the main point. India's AI infrastructure bottleneck is not only a shortage of chips or buildings. It is the need to coordinate powered capacity, low-latency fibre, cloud access, automation, service assurance and capital discipline. Lightstorm's value will rise if it remains the trusted control layer for that coordination. It will fall if fibre becomes abundant, hyperscalers squeeze renewals, power scarcity slows rack growth, or investors discover that growth metrics are outrunning durable contracted cash flow.

Evidence register