heading · body

Earnings · STLTECH · Telecom Equipment / Optical Fibre

Sterlite Technologies — a fibre turnaround betting the re-rating on AI

Sterlite Technologies Ltd

period Q4 FY25 → Q4 FY26 added 2026-06-09 score 7/10
earnings-call optical-fibre telecom vedanta STLTECH india

The Pulse

Sterlite Technologies (STL) is the Vedanta group’s optical-fibre maker — one of the few companies in the world that runs the full chain from raw glass to finished cable. For five painful years it was stuck in a global fibre downcycle: oversupply, Chinese price deflation and a telco-capex pause dragged a once-20%-margin, dividend-paying business into back-to-back losses, bottoming in FY25 with a ₹123 crore loss and a quarter where margins fell to 5%. FY26 is the first clean recovery year — revenue up 19% to ₹4,745 crore, EBITDA up 39%, profit back in the black at ₹56 crore, order inflows more than doubling, and net debt cut hard. Management delivered convincingly on what it promised this year, especially deleveraging. But here’s the tension: the stock trades at an almost surreal 628 times earnings and 13 times book, because the market isn’t pricing the recovered present — it’s pricing a future where AI and data-centre demand turns STL into a structural fibre winner. That story is plausible but, on the numbers disclosed, still unproven.

The Business

STL describes its value chain as “glass to gigabit”: it takes ultra-pure silicon, makes its own glass preforms (the hard, capital-intensive step very few players do in-house), draws fibre, designs high-density cables, and builds connectivity products. That backward integration is its core claimed cost advantage. It sells to telecom operators, government broadband programmes, enterprises, and — increasingly the whole investment thesis — data-centre and hyperscaler customers. Roughly 80% of revenue comes from outside India, with manufacturing in India plus cable plants in Italy and South Carolina and a fibre facility in China, a multi-geography footprint that helps it dodge tariffs. It holds ~8% of the global ex-China optical-cable market and a real patent estate (780+ patents, genuine next-gen R&D in hollow-core and multi-core fibre).

The corporate shape changed materially in FY25: the services/network-build arm (BharatNet-type deployment) was demerged into a separately listed entity, STL Networks (rebranded Invenia), leaving a slimmer manufacturing-plus-digital STL. A small software leg, STL Digital, contributes ~5–6% of revenue and just turned marginally profitable.

What’s distinctive is the integration and the IP. What’s structurally challenging is the nature of the business itself: optical fibre is cyclical and partly commoditised, with China at roughly half of world supply and chronic deflation risk; STL depends on raw-material chokepoints (germanium and helium, both China-dominated) that management openly admits it cannot fully secure; and it carries real litigation risk — a ~$96.5 million US patent judgment from Prysmian is under appeal. The bull case rests entirely on the demand pivot: if AI and data-centre orders lift volumes and mix enough to push utilisation past 70%, management says margins go from ~14% to 20%.

How Management Thinks

The company is run by Ankit Agarwal (MD, from the Vedanta corporate-development side) with CFO Ajay Jhanjhari, under non-executive chairman Anil Agarwal. The defining trait this year was disciplined execution against clearly stated targets — and on the things they could control, they delivered. The single most-repeated promise across five calls was “get utilisation from ~40–50% to 70%+ and EBITDA goes from 14% to 20%.” Operational margins did climb for six straight quarters. The deleveraging promise was over-delivered: net-debt-to-EBITDA fell from about 3x to 1.3x, beating the “below 2x” target and prompting a new sub-1.2x ambition. Profit turned positive, the turnaround they kept promising actually arrived, and there are small signals of integrity — executive pay was cut in the loss years (and explicitly flagged as “minimum remuneration in view of no profits”), and management refused to chase spiking spot-market fibre prices in China, calling it “a path of discipline.”

But the credibility read needs guardrails. Disclosure is deliberately opaque — management declines, every quarter, to reveal utilisation, capacity, volume-versus-value splits, or data-centre revenue, all on “competitive grounds.” That’s defensible, but it makes the AI story impossible to verify from outside. More tellingly, the headline narrative and the reported numbers are currently moving in opposite directions: management keeps talking up scaling enterprise-and-data-centre revenue to 30% of the total (and keeps pulling the timeline forward), yet that segment’s actual share fell from 23% to 19% over the year, blamed on the copper business. And when an analyst bluntly asked whether STL was losing the hyperscaler race to peers like Corning, Ankit’s denial was firm but evidence-light: “None of these things exist.” The tariff-adjusted margin framing (“ex-tariff we’d be at 18–19%”) is true but conveniently excludes a cost that was real — north of ₹100 crore in FY26.

Where It’s Going

The forward story is the AI/data-centre pivot, explicitly framed as a hedge against telco-capex cyclicality. Management has committed to a 20% reported EBITDA margin by Q4 FY27, enterprise-plus-data-centre revenue reaching 30% in FY27, and about ₹500 crore of capex to upgrade the asset base for high-value data-centre products. The product news supports the ambition: India’s first hollow-core fibre cable, a new US-made data-centre connectivity portfolio (Neuralis), and next-gen fibre wins with European operators. The order book backs the demand signal — inflows up 109% in FY26 to ₹7,687 crore, with large data-centre wins concentrated in North America, where STL’s revenue share has jumped from 25% to 39%.

The genuine tensions: the 20% margin is still a forward promise dependent on tariffs settling and helium/polymer costs behaving; the data-centre mix has to actually start rising rather than falling; the germanium/helium supply risk is unresolved by management’s own admission; the Prysmian judgment has three appeal stages still to run; and hollow-core fibre — the headline innovation — is, per management, expensive to make at low yields with no global standards yet, so mass adoption is years away.

The Four Checks

  1. Quality & moat (gate). This is the hard one, and the honest answer is “narrow and contested.” The glass-to-gigabit integration and patent estate are real differentiators, and very few players make their own preform. But the underlying product is a cyclical, China-dominated, partly-commoditised hardware where STL has earned poor returns on capital for years and just emerged from a multi-year loss. There is a cost-and-technology edge, but not a durable pricing moat — the five-year financial record (negative sales growth, sub-3% ROE) is the evidence. The remaining checks are therefore more academic than for a genuinely high-moat business.

  2. Returns on incremental capital & runway. Weak today, with a credible path to better. ROCE is 7.75% and ROE just 2.24% — these are trough-recovery numbers, not the marks of a high-return business. The argument is that filling existing capacity (the ₹500 crore upgrade aside, management stresses it won’t add gross capacity speculatively) lifts returns sharply because the incremental volume drops through at high margin. If the AI/DC demand and 70%+ utilisation materialise, incremental returns could be genuinely good; if the fibre cycle rolls over again, they won’t. The runway depends almost entirely on the data-centre thesis being right.

  3. Capital allocation for the stage. Rational and improved. Exiting non-core businesses, demerging the services arm to simplify, deleveraging aggressively, cutting executive pay in loss years, and refusing to chase spot prices are all the right moves for a company climbing out of a hole. No dividend is appropriate while it rebuilds. The ~₹1,000 crore QIP that diluted promoters ~10 points was defensible balance-sheet repair. This is the strongest of the four checks.

  4. Price. Extremely demanding — the central risk by a distance. At 628x earnings and 13x book, after the stock re-rated roughly eight-fold off its low, the market has fully priced a multi-year structural AI-fibre super-cycle and a clean march to 20% margins. The actual financials show a company that has only just scraped back to a ₹56 crore profit. Even granting a strong FY27, the valuation leaves no room for the fibre cycle disappointing, the data-centre mix continuing to lag the narrative, or any of the live tail risks (Prysmian, germanium, tariffs) biting. The recovery is real; the price assumes the transformation, not just the recovery.

Sources

Screener snapshot fetched 2026-06-09. Concalls read: May-2025 (Q4 FY25, the cycle bottom), Jul-2025 (Q1 FY26), Nov-2025 (Q2), Jan-2026 (Q3), May-2026 (Q4 & full-year FY26). Annual reports: FY23, FY24, FY25. Note: the latest (Q4 FY26) transcript failed to download on the first pass and was recovered manually from its BSE source. The GSB services business was demerged into STL Networks (Invenia) effective 31 March 2025, so post-FY25 figures reflect a slimmer manufacturing-plus-digital STL — worth bearing in mind when comparing across years. Research dumps in vault/Sources/Earnings/Sterlite Technologies Ltd/.