Business

Business — Fujikura Ltd. (5803)

Figures converted from Japanese yen at historical FX rates — see data/company.json.fx_rates. Ratios, margins, and multiples are unitless and unchanged.

Fujikura looks like a 140-year-old Japanese industrial conglomerate; it now trades like an AI-infrastructure pure play. One segment — Telecommunication Systems, anchored by the AFL US subsidiary's specialty optical fiber for hyperscalers — produced 68% of FY2025 operating profit on 46% of revenue, and is single-handedly responsible for the group's margin going from 8.7% to 13.8% in one year. The rest of the company is a collection of decent-but-ordinary businesses (electronics, auto harnesses, power cables, real estate); the market is paying for the fiber engine, and the investment question is whether that engine's earnings are a peak or a plateau.

1. How This Business Actually Works

The group is five businesses, but one earns the alpha. Telecommunication Systems is a specialty fiber + connectivity franchise sold mostly through AFL into US hyperscalers and Tier-1 telecom carriers; the other four segments are revenue without proportional incremental profit.

The economics inside Telecom are different from the rest of the cable industry. Standard cable is a metals pass-through with a thin spread; Fujikura sells engineered solutions — SWR (Spider Web Ribbon) and WTC (Wrapping Tube Cable) let carriers and hyperscalers cram many more fibers into the same duct, qualified by customer over 12-18 month cycles. AFL co-launched the SDM4 multicore-fiber MSA in March 2026 with Corning, Sumitomo and TeraHop, anchoring four-core fiber as the AI-campus standard. Translation: pricing decoupled from copper, capacity tight, qualification moats real.

The chart captures the thesis: telecom roughly doubled, doubled-again on the same y-axis. Cost structure stayed orthodox — gross margin moved from 21% to 27%, SG&A held at 12-13% of sales, operating leverage did the rest. R&D is $123M (1.9% of sales), capex is $205M (3.1% of sales), free cash flow grew to $581M in FY2025. This is not a capital-intensive business at the group level; it is capital-intensive in a few specific lines (fiber draw towers at Sakura Works) where Fujikura is willing to spend ahead of demand because the qualification cycle is what matters.

2. The Playing Field

Fujikura is mid-sized by group revenue but punches well above its weight on margin and ROE — entirely because of segment mix.

Three things the peer set reveals. First, Fujikura sits with Corning in the upper-right quadrant — the only two cable peers earning specialty-grade operating margins; both are riding the same AI-optical wave. Second, the two pure-cable European leaders trade very differently: Prysmian's 9.8% margin commands 18.5x EV/EBITDA because of a multi-year HVDC subsea backlog (~$19B), while Nexans' 4.8% margin trades at 11x. Multiples in this industry compensate visibility, not raw margin. Third, the Japanese trio (Fujikura, Sumitomo Electric, Furukawa) is the cleanest mix story: same country, same yen, same auto/cable starting point, vastly different outcomes — Fujikura's 14% margin is 3.5x Furukawa's because Fujikura swung harder on AI fiber and has a US-centric AFL footprint.

The right peer for Fujikura is Corning, not Furukawa. Corning's Optical Communications segment grew at hyperscaler-driven double digits, holds gross margins north of 30%, and earns roughly half of Corning's op profit. Fujikura's Telecom segment is structurally similar — optical fiber + connectivity, US-anchored, hyperscaler-leveraged — but smaller. The gap between a "specialty fiber" multiple and a "Japanese cable conglomerate" multiple is what the SOTP section below tries to size.

3. Is This Business Cyclical?

Yes — and the cycle is idiosyncratic to demand vector, not to GDP. Fujikura's history shows three distinct stress events in the last decade, each driven by a different lever, plus the current burst that has come almost entirely from one segment.

The lesson from the FY2019–FY2021 episode is the relevant one. Operating margin fell from a normalized 4-5% to 0.5% in FY2020 because hyperscale was not yet meaningful, telecom carrier capex paused globally, copper ran the wrong way against working capital, and goodwill on prior acquisitions was impaired. The dividend went to zero in FY2021. Recovery came from mix shift into specialty fiber and the AFL hyperscaler ramp, not cost-cutting alone. The current 13.8% margin is therefore not a "cycle peak" in the classic sense; it is the result of one segment's revenue almost doubling on a roughly fixed cost base. The cyclical question is consequently very specific: how durable is hyperscaler optical-fiber capex? If hyperscale capex flatlines, telecom segment growth slows, operating leverage runs in reverse and group margin compresses fast. The FY2020 evidence says Fujikura's downside in a real demand shock can be substantial — fixed-cost structure, US dollar mix, and metals working capital all flex against the company simultaneously.

A second smaller cycle to watch is Auto. Management flagged in November 2025 that EV harness demand is slowing relative to expectations, and the segment's 3.3% margin gives essentially no cushion. Auto is not the alpha source, but it is the most likely segment to print a loss in a soft year and create a non-cash impairment that lands in headline EPS.

4. The Metrics That Actually Matter

Group revenue and group EPS both mislead here. Group revenue moves with copper and yen FX; group EPS moves with one-offs. The metrics below are what professionals actually track.

Operating margin at the segment level is the single most decisive number. Tracking the consolidated 13.8% misses that the durability question is entirely about whether 20.4% in Telecom is real. Watch the segment table in every quarterly disclosure; the rest is secondary.

5. What Is This Business Worth?

This is a rare case where sum-of-the-parts is genuinely the right lens, not a forced exercise. The five segments earn dramatically different margins (3.3% to 20.4%), face different multiples, and one (Telecom) is structurally a pure-play specialty-fiber business that the rest of Fujikura partially obscures. Consolidated multiples that average a 14% Telecom-fiber business with a 4% auto-harness business are a category error.

The exercise is directional, not precise. Telecom's ~$616M of operating profit at a mid-teens specialty-fiber multiple accounts for the bulk of group enterprise value at current market price (~$11.4 billion). The other four segments together generate roughly $300M of operating profit and would be worth perhaps $2.3-3.0B at generous traditional cable / electronics multiples. Net financial position is roughly neutral. In practice: at today's price, the market is implicitly valuing the Telecom franchise close to a Corning Optical Communications multiple. The stock is cheap from here only if Telecom op profit compounds toward $800M+ and holds; it is expensive if Telecom op profit reverts to $400-470M.

A second valuation-relevant feature: there is no listed-subsidiary discount to chase. AFL is wholly owned, real estate is held in segments, and Fujikura is not a holding company. So unlike Sumitomo Electric (which holds many cross-shareholdings) or true Japanese conglomerates with listed sub-stakes, the SOTP here is purely operational segment economics, not balance-sheet financial-stake unwind.

6. What I'd Tell a Young Analyst

Forget the consolidated income statement. Open the segment table first, every quarter. Telecom revenue and margin are 80% of the story; group revenue is mostly metals and yen.

The set-up is asymmetric in a way that requires real opinion. The operating leverage that drove FY2025's margin from 8.7% to 13.8% can run hard in reverse. If hyperscaler capex digestion is real (a known risk into 2026-2027), Telecom revenue can flatten or fall in a single year and consolidated margin can drop 400-600bps. Sustained 17%+ Telecom margin is the bull thesis; reversion to 12-13% is the bear case — and the answer depends on hyperscaler optical capex through the next two years, plus whether SDM4 multicore-fiber adoption holds the qualification moat.

The market is plausibly wrong in both directions at once: overestimating how durable the current Telecom margin is for any given year (operating leverage cuts both ways) and underestimating how structurally advantaged Fujikura's position is across multiple years (SWR/WTC, fusion-splicer franchise, SDM4 MSA, US-onshored AFL plant footprint into a tariff regime that favors domestic content). The trade-able edge sits around the FY2026 guide — flat telecom revenue guidance prices in reversion; double-digit growth into a known capex digestion is a real disagreement to underwrite.

Signals that would change the view:

• Bullish trigger: Verizon WTC-style design wins from a second hyperscaler tier; SDM4 production-scale orders; capex guidance up sharply for telecom plants
• Bearish trigger: Hyperscaler capex guidance cuts in Q1 of any major buyer; Auto segment swings to operating loss; inventory days up sharply alongside softening orders
• Trigger that gets misread: a copper or aluminum spike that creates a quarterly working-capital miss — historically a buying opportunity, not a thesis break