Sterlite Technologies has launched a new line of plenum-rated fiber trunk assemblies built specifically for AI data centers, the Indian connectivity manufacturer announced on September 10, 2026. The move lands as hyperscalers and colocation operators race to wire GPU-dense racks fast enough to keep pace with AI infrastructure buildouts, and it puts Sterlite Technologies (known commercially as STL) in more direct competition with established cabling giants Corning, CommScope, and Amphenol.
The announcement, reported by outlets including TradingView (via Reuters) and STL’s own press release page, covers a certified portfolio spanning 48F to 576F fiber counts built with Intermittently Bonded Ribbon (IBR) construction. The assemblies carry OFNP certification in the United States and comply with NFPA 262 fire-safety standards for indoor air-handling spaces, which is the specific detail that makes this launch relevant to anyone specifying cabling for a plenum-rated data hall right now.
What Sterlite Tech Actually Announced
STL’s press release, titled “STL launches Plenum-rated Fiber Trunk Assemblies for AI Data Center Connectivity,” describes factory-built cable assemblies with connectors pre-installed, polished, and tested before they ever leave the plant. That pre-termination detail matters more than it sounds. Field-terminated fiber runs require technicians to splice and polish connectors on-site, a process that eats installation time and introduces variability in signal loss. Pre-terminated trunks skip that step, which is exactly the kind of time savings a hyperscaler racing to bring a new AI cluster online wants to hear about.
The fiber count range, 48F up to 576F, positions the product for both mid-size enterprise rooms and the ultra-dense trunk runs that hyperscale AI clusters now require. STL frames the assemblies as built for “next-generation AI data center overhead and underfloor pathways,” according to coverage from MarketResearchReports.com, which tracks the broader data center wire and cable market. The plenum rating is the part installers care about most: any cabling that runs through an air-handling space used for heating, ventilation, or air conditioning has to meet NFPA 262 fire and smoke standards, and OFNP certification is the US designation that confirms compliance.
This isn’t STL’s first move into pre-terminated AI data center connectivity this year. Back in May 2026, the company’s US subsidiary, STL Optical Connectivity NA, announced it had received US Conec certification to offer pre-terminated fiber trunks using MMC connectors for AI data center deployments, according to a release distributed via PRNewswire. The September announcement builds directly on that certification push, extending the plenum-rated, fire-code-compliant version of the same pre-terminated philosophy.
Why Plenum Rating Matters for AI Infrastructure
Plenum spaces are the return-air pathways above drop ceilings or below raised floors that HVAC systems use to circulate air through a building. Any cable running through those spaces has to meet stricter fire and smoke standards than cable run through conduit or open trays, because a fire in a plenum space can spread smoke through an entire structure via the same ducts. NFPA 262, the test standard referenced in STL’s certification, measures flame spread and smoke density under conditions that simulate a plenum fire.
AI data centers complicate this picture because they pack far more compute density per rack than a traditional enterprise server room. A GPU-dense rack running Nvidia’s latest accelerators can draw ten times the power of a conventional 1U server rack, and the cooling infrastructure needed to manage that heat load often relies on extensive overhead or underfloor air pathways. That means more of the fiber plant in a modern AI data center now has to run through plenum-rated spaces, not less. A cabling vendor that can’t certify to NFPA 262 effectively locks itself out of a growing share of new hyperscale builds.
Intermittently Bonded Ribbon construction, the technology underpinning STL’s new assemblies, also plays into the density story. IBR fiber uses small bonding points between ribbon fibers rather than continuous adhesive, which lets the ribbon flex for mass fusion splicing while keeping the overall cable diameter smaller. Smaller diameter cable means more fiber pairs fit through the same conduit or cable tray, a real constraint in AI data centers where duct and tray capacity is often the limiting factor on how fast a new cluster can be wired and brought online.
The AI Data Center Cabling Market STL Is Chasing
The global data center wire and cable market was estimated at $18.17 billion in 2025 and is projected to reach roughly $27.42 billion by 2031, according to figures cited by MarketResearchReports.com. That growth trajectory, roughly 7% compound annual growth, is being pulled almost entirely by AI infrastructure spending rather than traditional enterprise IT refresh cycles. Hyperscalers building out GPU clusters need dramatically more fiber per rack than a conventional cloud data center, both for east-west traffic between accelerators and for north-south links back to storage and networking spines.
STL is not alone in chasing that spend. Corning’s Optical Communications segment has spent much of 2025 and 2026 expanding fiber production capacity specifically to serve AI data center customers, while CommScope and Amphenol both operate large structured cabling and connectivity divisions that compete directly for hyperscale contracts. Panduit also plays in this space with its own pre-terminated trunk cable lines. What distinguishes STL’s pitch is a combination of Indian manufacturing scale, US certification, and a specific bet on the plenum-rated segment of the market, which is a narrower but fast-growing slice of overall data center cabling demand.
STL’s own financials show why the company is leaning hard into this segment. The company reported consolidated revenue of ₹4,745 crore for fiscal year 2026, up 19% year over year, with EBITDA climbing 39% to ₹628 crore and net profit recovering to ₹56 crore after a loss the prior year, according to disclosures reviewed by ScanX Trade. The momentum carried into the first quarter of fiscal 2027, when STL reported total income of ₹1,922 crore, up 87% year over year, a jump the company and market commentators have tied directly to AI-driven data center demand.
Sterlite Tech vs Corning, CommScope, Amphenol: How the Players Compare
| Company | Primary AI Data Center Product | Key Differentiator | Market Position |
|---|---|---|---|
| Sterlite Technologies (STL) | 48F–576F IBR plenum-rated trunk assemblies | NFPA 262 / OFNP certified, US Conec MMC connector certification | Challenger scaling US manufacturing and certification footprint |
| Corning | Optical fiber cable and connectivity systems | Deep fiber manufacturing scale, long hyperscaler relationships | Market leader in optical fiber for data centers |
| CommScope | Structured cabling and fiber connectivity | Broad enterprise and hyperscale connectivity portfolio | Established incumbent, major structured cabling supplier |
| Amphenol | High-speed interconnect and cable assemblies | Diversified connector and cable assembly manufacturing | Major player in copper and optical interconnect for AI racks |
| Panduit | Pre-terminated fiber trunk cable systems | Long-standing structured cabling brand for enterprise data centers | Established mid-tier competitor in trunk cabling |
The table above reflects publicly disclosed product categories and market positioning as reported across the sources cited in this article rather than a head-to-head lab benchmark, since none of the companies involved have published directly comparable third-party test data for their AI data center trunk assembly lines as of this writing.
STL’s Financial Turnaround, By the Numbers
| Metric | FY26 (Full Year) | FY25 (Full Year) | Q1 FY27 |
|---|---|---|---|
| Revenue | ₹4,745 crore | ₹3,996 crore | ₹1,922 crore |
| EBITDA | ₹628 crore | ~₹452 crore (implied) | Not separately disclosed |
| Net Profit | ₹56 crore | -₹123 crore (loss) | ₹197 crore |
| Revenue Growth (YoY) | 19% | — | 87% |
Figures are drawn from STL’s disclosed financial results as reported by ScanX Trade and Angel One. The swing from a full-year loss in FY25 to profitability in FY26, followed by a sharp acceleration in Q1 FY27 revenue, lines up closely with the timeline of STL’s AI data center connectivity push, including the May 2026 MMC certification and the September 2026 plenum-rated trunk assembly launch.
Historical Context: How Data Center Cabling Got Here
Structured fiber cabling in data centers has gone through several distinct phases over the past two decades. Early enterprise data centers relied almost entirely on field-terminated fiber, with technicians splicing connectors on-site rack by rack. Pre-terminated trunk cables started gaining traction in the 2010s as hyperscale cloud providers like Amazon Web Services, Microsoft Azure, and Google Cloud needed to stand up data halls faster than field termination allowed, and vendors like Corning and CommScope built out large pre-terminated product lines to serve that demand.
The generative AI boom that started accelerating in 2023 changed the math again. GPU clusters need far denser fiber interconnects than traditional cloud racks because accelerators talk to each other constantly during distributed training and inference, not just to storage and the outside network. That has pushed data center operators toward higher fiber counts per trunk, tighter cable diameters to preserve duct capacity, and faster installation timelines to match the pace at which chipmakers are shipping new accelerator generations. STL’s 48F-to-576F range and IBR construction are a direct response to that shift, and so is the parallel push by Corning and Amphenol to expand their own high-density optical and copper interconnect lines for AI racks.
Market Impact: What This Means for Data Center Operators
For hyperscalers and colocation operators currently planning new AI data center builds, STL’s entry adds a certified alternative supplier in a segment where lead times and manufacturing capacity have been under strain. As GPU cluster construction has accelerated across the US, India, and Southeast Asia through 2026, cabling and connectivity components have periodically shown up on procurement teams’ lists of constrained inputs, alongside better-known bottlenecks like transformers, cooling equipment, and the GPUs themselves. A new qualified vendor with NFPA 262 and OFNP certification gives buyers another sourcing option rather than depending solely on Corning, CommScope, or Amphenol capacity.
For STL specifically, the announcement reinforces a strategic pivot the company has been signaling through its financial disclosures over the past year: a shift away from lower-margin telecom fiber supply toward higher-margin, higher-complexity data center connectivity products. That shift tracks with the EBITDA margin expansion STL has reported across FY26 and into FY27, and with the company’s own guidance raising its FY27 EBITDA margin target, according to ScanX Trade’s coverage of STL’s Q1 FY27 results.
There’s a competitive angle here too. Corning, CommScope, and Amphenol have spent 2025 and 2026 investing heavily in US manufacturing capacity to serve AI data center demand domestically, partly in response to supply chain and tariff pressures. STL’s approach, building out a US-certified, US-subsidiary-marketed product line while manufacturing in India, represents a different bet: that certification and product performance can offset a non-domestic manufacturing base, at least for now. Whether that holds up as US trade policy around electronics and telecom components continues to shift through 2026 and 2027 is one of the bigger open questions for the company.
Predictions: Where AI Data Center Cabling Goes From Here
- Fiber counts per trunk keep climbing. If STL’s 576F ceiling becomes a competitive benchmark, expect Corning, CommScope, and Amphenol to publicize comparable or higher fiber-count pre-terminated products within the next two to three quarters, pushing the industry standard higher across the board.
- Plenum certification becomes table stakes, not a differentiator. As more AI data center floor plans route cabling through air-handling spaces to manage GPU rack cooling, NFPA 262 and OFNP certification will likely shift from a selling point to a baseline requirement that every serious vendor in this category needs to hold.
- Pre-termination keeps eating into field-termination revenue. Installers and integrators that built businesses around field-terminated fiber splicing will likely see continued pressure as hyperscalers standardize on factory pre-terminated trunks to compress build timelines.
- STL’s US manufacturing footprint likely grows. Given the company’s May 2026 US Conec certification and its US subsidiary structure, a domestic manufacturing or assembly announcement in the next 12 months would be a logical next step, especially if trade policy makes India-manufactured components less attractive for US hyperscale contracts.
- Consolidation risk rises for smaller cabling vendors. As Corning, CommScope, Amphenol, Panduit, and now STL all compete for a data center cable market projected to grow from $18.17 billion in 2025 toward $27.42 billion by 2031, smaller regional cabling manufacturers without AI-specific certifications may become acquisition targets rather than standalone competitors.
What Sets STL’s Approach Apart From Competitors
Beyond the certification checklist, STL’s messaging leans specifically on speed. Factory pre-termination, polishing, and testing before a cable ever ships means installers spend less time on-site with test equipment confirming insertion loss and return loss on every connector. For a hyperscale build where every day of delayed rack activation represents lost compute revenue, that time compression is the actual product being sold, not just the fiber count or the fire rating on its own.
It’s also worth noting what STL has not disclosed. The company’s press materials, as reported by outlets including Yahoo Finance and MarketScreener, describe the certification and fiber count range in detail but do not name specific hyperscale customers already deploying the new plenum-rated assemblies. That’s typical for a component-level supplier serving hyperscalers, most of which don’t publicize their vendor lists for competitive reasons, but it does mean the actual customer adoption of this specific product line remains unverified beyond STL’s own claims at this stage.
The Broader AI Infrastructure Supply Chain Picture
STL’s cabling announcement is one piece of a much larger AI infrastructure buildout that has strained supply chains across chips, power, cooling, and now connectivity. GPU shortages, transformer lead times measured in years rather than months, and now periodic tightness in high-density fiber and cable assembly capacity have all become recurring themes in hyperscaler earnings calls through 2026. A cabling vendor bringing meaningful new certified capacity online, even in a relatively narrow product category like plenum-rated trunk assemblies, is a small but real data point in that broader supply chain story.
It also reflects how AI infrastructure spending has started pulling in suppliers well outside the traditional chip and server ecosystem. Companies whose core business has historically been telecom fiber, like STL, are increasingly repositioning around data center connectivity because that’s where growth and margin both point right now. Expect more of STL’s India-based and internationally focused peers in the fiber optics space to make similar pivots over the coming year as AI data center capital expenditure continues outpacing traditional telecom infrastructure spend globally.
Frequently Asked Questions
What did Sterlite Technologies actually announce?
On September 10, 2026, Sterlite Technologies (STL) announced a new line of plenum-rated fiber trunk assemblies for AI data center connectivity, spanning 48F to 576F fiber counts, built with Intermittently Bonded Ribbon construction, and certified to US OFNP and NFPA 262 standards.
What does “plenum-rated” mean for fiber cabling?
Plenum-rated cable meets fire and smoke safety standards, specifically NFPA 262, required for cabling that runs through air-handling spaces like the areas above drop ceilings or below raised floors used for HVAC circulation. This matters more in AI data centers because dense GPU racks often require extensive overhead or underfloor cooling pathways.
Who are Sterlite Tech’s main competitors in AI data center cabling?
The main competitors in this space are Corning, CommScope, Amphenol, and Panduit, all of which offer pre-terminated fiber trunk and structured cabling products for hyperscale and enterprise data centers.
How big is the data center cabling market?
The global data center wire and cable market was estimated at $18.17 billion in 2025 and is projected to grow to approximately $27.42 billion by 2031, according to figures cited by MarketResearchReports.com, driven largely by AI infrastructure buildouts.
What is Intermittently Bonded Ribbon (IBR) fiber?
IBR is a fiber ribbon construction that uses intermittent bonding points between fibers rather than continuous adhesive, allowing the ribbon to flex for mass fusion splicing while keeping the cable diameter smaller, which helps fit more fiber capacity into limited conduit or duct space.
How did STL’s finances look heading into this launch?
STL reported FY26 consolidated revenue of ₹4,745 crore, up 19% year over year, returning to profitability with a ₹56 crore net profit after a prior-year loss. Q1 FY27 revenue jumped 87% year over year to ₹1,922 crore, a swing the company has tied to AI data center demand.
Is this STL’s first AI data center connectivity certification?
No. In May 2026, STL’s US subsidiary announced it had received US Conec certification for pre-terminated fiber trunks using MMC connectors for AI data center deployments. The September 2026 plenum-rated launch extends that same pre-terminated product strategy with fire-code certification added.
Does this announcement name specific hyperscaler customers?
No. STL’s public materials describe the product specifications and certifications in detail but do not disclose which hyperscale or colocation customers are deploying the new assemblies, which is typical for component-level suppliers in this market.