Sterlite Technologies Wins $288-Million Hyperscaler Fibre Contract as AI Data-Centre Demand Accelerates

Indian optical-connectivity company Sterlite Technologies Limited has secured a long-term supply agreement valued at approximately $288 million with a leading international hyperscaler, strengthening its position in the rapidly expanding global market for fibre infrastructure supporting artificial intelligence, cloud computing and hyperscale data centres.

The agreement was disclosed on August 29, 2026, with a subsequent clarification on August 31 confirming that the contract had been signed through a wholly owned subsidiary of Sterlite Technologies. Under the arrangement, STL will supply high-density optical fibre cable products manufactured according to the hyperscaler’s specifications during calendar years 2027, 2028 and 2029. Purchase orders will be issued periodically over the duration of the contract rather than the entire value being booked as an immediate order for delivery.

The three-year agreement can also be extended for a further two years by mutual consent, potentially creating a longer relationship extending beyond 2029. The identity of the hyperscaler has not been disclosed publicly, and STL has not identified the precise data-centre locations at which the cables will eventually be deployed.

High-Density Fibre for the Hyperscale Computing Era

The products covered by the agreement are high-density optical fibre cables, which are becoming increasingly important as hyperscale data centres accommodate much larger concentrations of computing hardware.

Traditional cloud data centres already require extensive fibre connectivity between servers, switches, storage equipment and external networks. Artificial-intelligence infrastructure dramatically increases this requirement because large AI models are trained and operated across clusters containing thousands or even tens of thousands of graphics processing units and other accelerators.

These processors must continuously exchange enormous quantities of information at very high speeds and with extremely low latency. The network connecting the computing equipment therefore becomes almost as important as the processors themselves.

High-density cable designs allow substantially larger numbers of individual optical fibres to be installed within constrained ducts, trays and data-centre pathways. This enables operators to increase bandwidth without proportionately increasing the physical space consumed by cabling infrastructure.

For hyperscale data centres, where floor area, rack space and cable-routing capacity are valuable resources, increasing fibre density can therefore become an important part of network design.

STL Has Developed Cables Containing Up to 6,912 Fibres

Sterlite Technologies has been positioning its product portfolio specifically for this emerging requirement.

The company’s Celesta Intermittently Bonded Ribbon, or IBR, cable family can accommodate as many as 6,912 individual optical fibres within an ultra-high-density cable structure. STL says the product uses 200-micron bend-insensitive single-mode fibre and is designed for applications where operators need extremely large fibre counts without consuming excessive duct space.

Its 6,912-fibre Celesta cable can place thousands of optical pathways into a cable approximately 32 millimetres in diameter, according to company material. Intermittently bonded ribbon technology allows groups of fibres to remain flexible while also permitting mass-fusion splicing, which can reduce installation time when large numbers of fibres must be connected.

These characteristics have become particularly relevant for AI data centres because GPU-intensive computing architectures generate significantly greater east-west network traffic between servers than conventional enterprise computing systems.

Instead of information primarily travelling between users and individual servers, AI workloads require constant communication between large numbers of processors operating together on the same computational task.

That shift is forcing data-centre operators to build much denser internal optical networks.

Neuralis Developed Specifically for AI Data Centres

STL has responded to this transition by developing Neuralis, a dedicated portfolio of optical-connectivity products intended for AI-era data centres.

The platform addresses two major areas of data-centre connectivity. The first is AI whitespace, covering high-density fibre infrastructure inside the data centre itself, while the second is high-speed Data Centre Interconnect, which connects buildings, campuses and external network infrastructure.

The Neuralis portfolio includes pre-terminated fibre trunks, high-density fibre arrays, Celesta IBR cables and specialised enclosures and panels. STL has designed these products for emerging network speeds including 800G and 1.6T environments, where individual links can carry exceptionally large amounts of data.

Pre-terminated systems are particularly useful in large data-centre projects because connections can be assembled and tested in the factory before being delivered to the site. This reduces the amount of fibre termination required during construction and can help shorten deployment schedules.

Speed of installation has become increasingly important as hyperscalers race to bring new AI computing capacity online.

AI Is Producing a Fibre Explosion Inside Data Centres

The growth of artificial intelligence is changing the physical architecture of computing facilities in ways that are creating unusually strong demand for optical connectivity.

STL has cited industry estimates indicating that GPU-dense racks can require around 36 times more fibre than conventional CPU-based racks, while AI data centres can require substantially higher fibre density than traditional facilities. The precise requirements vary considerably according to architecture, but the underlying trend towards dramatically greater optical connectivity is well established.

Modern AI clusters operate by distributing computing workloads across enormous numbers of processors. These processors must function almost as a single system, which means delays in the communications network can reduce the efficiency of expensive computing hardware.

Optical fibre is therefore increasingly moving deeper inside the data centre, replacing or supplementing copper connections for longer and higher-capacity links.

As AI models become larger and computing clusters expand, fibre requirements grow not only between individual buildings but also within the server halls themselves.

This creates opportunities for companies capable of manufacturing ultra-high-density cables, pre-terminated assemblies, connectors and complete optical-distribution systems rather than merely producing conventional telecommunications fibre.

$288-Million Contract Includes Risk-Sharing Mechanism

One notable feature of the new hyperscaler agreement is its commercial structure.

Sterlite Technologies said the arrangement establishes a reciprocal risk-sharing framework that defines capped financial liabilities for both parties if actual demand falls below contracted expectations or if STL is unable to provide the required manufacturing capacity.

This structure reflects the unusual challenges associated with long-term supply agreements in a market expanding as rapidly as AI infrastructure.

Hyperscalers want guaranteed access to manufacturing capacity because a shortage of fibre or connectivity products could delay expensive data-centre projects. Manufacturers, meanwhile, need sufficient demand certainty before committing equipment, materials and production capacity several years in advance.

A risk-sharing framework can provide protection to both sides by establishing consequences if one party fails to meet its contractual commitments.

The presence of this mechanism also means the $288-million figure should be understood as the estimated value of the long-term supply arrangement rather than revenue that STL has already earned.

Actual revenue will be recognised progressively as products are ordered, supplied and accepted during the contract period.

Contract Follows Earlier Hyperscaler Agreement Worth More Than $1 Billion

The latest transaction follows an even larger hyperscaler-related agreement announced by Sterlite Technologies earlier this year.

On May 22, 2026, STL announced that one of its subsidiaries had received a multi-year Product Award Letter valued at more than $1 billion for the supply of optical-connectivity products supporting AI data-centre construction. The company said those products would support AI infrastructure being developed in the United States.

STL subsequently reported that this award represented more than ₹10,000 crore of potential business, alongside additional hyperscaler orders exceeding $100 million for products from its Neuralis portfolio.

The new $288-million agreement therefore adds another substantial multi-year commitment to an order pipeline increasingly linked to global data-centre investment.

The company has not publicly established whether the latest agreement involves the same hyperscaler connected with the earlier billion-dollar award, so the two transactions should not be assumed to involve separate customers unless further details are disclosed.

STL Order Book Reaches ₹18,618 Crore

The expansion of hyperscaler business has already begun to change the scale of Sterlite Technologies’ contracted pipeline.

At the end of the quarter to June 30, 2026, the company reported an open order book of ₹18,618 crore, the highest in its history. STL also reported quarterly revenue of ₹1,910 crore, EBITDA of ₹397 crore and profit after tax of ₹197 crore, all representing significant improvements from the corresponding period a year earlier.

The company attributed part of its improved profitability to a higher contribution from the data-centre business and a more favourable product mix.

This is important because specialised AI-data-centre connectivity products can offer manufacturers a different commercial opportunity from conventional commodity optical-fibre cables used for telecommunications roll-outs.

Customers increasingly require complete systems involving high-density cables, connectors, pre-terminated assemblies, panels and engineering support. Suppliers capable of offering those integrated solutions can participate in a larger share of data-centre infrastructure expenditure.

From Glass Preform to Finished Data-Centre Connection

One of STL’s strategic advantages is its vertically integrated manufacturing model.

The company describes its capability as extending from glass preform production and optical-fibre drawing through cable manufacturing and finally connectorisation and finished data-centre assemblies.

Optical fibre begins with an ultra-pure glass preform. The preform is heated and drawn into extremely thin fibre while maintaining precise optical characteristics. Those fibres are subsequently coated, grouped and incorporated into cables before connectors and associated hardware are added according to their final application.

Controlling several stages of this process can allow a manufacturer to optimise fibre properties, cable geometry and finished assemblies as an integrated system.

For hyperscale customers, this can also simplify supply chains because fewer suppliers are needed to provide the different layers of the physical optical network.

STL has increasingly emphasised this end-to-end capability as it shifts towards AI data centres and other high-performance network applications.

Manufacturing Footprint Increasingly Oriented Towards Global Markets

The company’s manufacturing strategy has also become more internationally focused.

Sterlite Technologies operates optical-fibre and cable manufacturing capacity in India and overseas and has been expanding its presence in North America, one of the largest markets for hyperscale data-centre construction.

The United States has become especially important because large technology companies are committing hundreds of billions of dollars to AI infrastructure, data centres, power systems and associated communications networks.

STL launched Neuralis in the US in April 2026 and has been expanding products specifically certified or designed for the North American market.

The company has also announced plans to invest as much as $100 million in additional US manufacturing capacity, reflecting the importance of proximity to hyperscale customers and growing demand for locally available optical-connectivity products.

At the same time, significant portions of its technology, engineering and manufacturing base remain in India, allowing the company to connect Indian manufacturing capability with global digital-infrastructure investment.

Optical Fibre Becoming Strategic Infrastructure for Artificial Intelligence

The rise of AI has placed enormous attention on semiconductor processors such as GPUs, but computing chips represent only one component of an AI data centre.

Large-scale AI systems also require electricity generation, cooling equipment, storage, networking switches and extensive fibre infrastructure connecting thousands of computing devices.

A processor costing tens of thousands of dollars provides little value if it cannot communicate efficiently with the rest of the computing cluster.

The physical network is therefore increasingly becoming one of the constraints determining how effectively AI data centres can scale.

This helps explain why hyperscalers are entering multi-year arrangements with optical-connectivity suppliers. They are not merely purchasing cables for individual construction projects but attempting to secure manufacturing capacity for a prolonged global infrastructure build-out.

For Indian companies such as STL, the opportunity lies in moving beyond conventional telecom fibre and capturing higher-value portions of this emerging network architecture.

Higher Fibre Density Reduces a Physical Constraint

One of the less visible challenges inside large data centres is simply finding enough physical space for connectivity.

Thousands of cables must travel through underground conduits, overhead trays, equipment racks and pathways between buildings. As fibre requirements increase, existing ducts can become congested long before the overall computing campus reaches its ultimate capacity.

Increasing the number of fibres within each cable therefore becomes an important engineering solution.

STL’s Celesta designs with up to 6,912 fibres are intended specifically to address this problem by providing very high fibre density while retaining installation flexibility.

Higher density can allow hyperscalers to increase network capacity without installing proportionately more ducting or expanding cable pathways, both of which can add cost and construction time.

That is why cable density, although far less visible than computing hardware, has become an increasingly strategic characteristic of AI infrastructure.

Green Manufacturing Becomes Another Part of STL’s Strategy

Sterlite Technologies is simultaneously attempting to reduce the environmental footprint associated with producing this infrastructure.

On August 27, the company announced that key manufacturing facilities in Chhatrapati Sambhaji Nagar, Maharashtra, had transitioned to electricity sourced from renewable wind and solar power. STL also said it was moving towards green hydrogen in its Indian manufacturing processes.

The company reported that these changes had reduced the embodied carbon associated with one of its G.657.A2 optical-fibre products from approximately 4.7 kilograms to around 0.9 kilograms of carbon dioxide equivalent, although these are company-reported sustainability figures.

This issue is becoming increasingly relevant to hyperscalers because the environmental footprint of AI infrastructure is no longer limited to the electricity consumed after a data centre begins operating. Large quantities of steel, concrete, electronics and networking equipment also generate emissions during manufacturing.

Technology companies attempting to reduce supply-chain emissions are therefore placing greater pressure on equipment manufacturers to reduce the carbon intensity of their products.

India Moves Deeper Into the Global AI Infrastructure Supply Chain

The $288-million contract illustrates an important change in India’s relationship with the global technology industry.

India’s role in digital services has traditionally been associated with software, IT outsourcing and engineering talent. Companies such as Sterlite Technologies are attempting to build a parallel position in the physical infrastructure that makes the digital economy possible.

AI data centres require optical fibre, cable assemblies, network switches, power equipment, cooling systems and increasingly sophisticated connectivity hardware. Supplying these components allows Indian manufacturers to participate directly in the global capital expenditure created by artificial intelligence.

The latest contract is particularly significant because it is a multi-year agreement extending from 2027 through 2029, potentially continuing for another two years. That provides substantially greater production visibility than a conventional short-duration equipment order.

Combined with STL’s earlier billion-dollar hyperscaler award, its Neuralis product portfolio and an open order book of ₹18,618 crore at the end of June, the agreement demonstrates how rapidly AI-related infrastructure is becoming a central part of the company’s growth strategy.

The larger story is therefore not simply that an Indian company has secured another optical-fibre contract. As AI computing expands across the world, enormous new physical networks must be constructed to connect the processors on which those systems depend. Sterlite Technologies is increasingly positioning Indian-developed optical technology and manufacturing capability inside that global infrastructure build-out.

---