Home

Deck

RS Technologies · 3445 · TSE

RS Technologies is Japan's #1 reclaimed-silicon-wafer maker — it strips and re-polishes scrap test wafers from foundries like TSMC and Samsung at 37% segment margins — bolted to a Chinese prime-wafer subsidiary and a low-margin equipment rollup.

¥6,640
Price
¥176B
Market cap
¥76.7B
Revenue (FY25)
¥76B
Net cash (43% of mcap)
Listed Aug 2016 at split-adjusted ~¥600; ground sideways in a ¥2,500–4,000 range for four years, then broke out in April 2026 to ¥6,640 — +150% in twelve months, at all-time highs.
2 · Three businesses, one ticker

Reclaim is 36% of revenue but 71% of segment operating profit.

36.9%
Reclaim op margin (FY25) 9-yr band: 35.1%–40.6%
19.9%
Prime wafer (China) margin 14.5-pt cyclical band
5.3%
Equipment rollup margin optical + camera + battery
71%
Of segment profit from reclaim alone

Reclaim is the engine: the customer ships its own used test wafer back, RST strips the films with proprietary chemistry, re-polishes, and sells it back for one-quarter to one-half of a new wafer. Feedstock is free; through nine years of cycles the segment has never printed below 35% or above 41%. Prime wafer is a capital-heavy Chinese 8-inch oligopoly running on state subsidies. Equipment is a low-margin rollup of optical pickups, camera modules and vanadium-flow batteries. Valuing RST on consolidated margins averages a quality compounder with a commodity ramp and a passthrough.

3 · The tension

One number — reclaim segment margin — separates a ¥9,500 SOTP from a ¥4,200 stress case.

  • Bull SOTP arithmetic: ¥9,500. Carve reclaim at a quality-materials multiple (12–13× EBIT, like specialty chemicals), prime at 6× on China-policy discount, equipment at 5×, add ¥76B net cash, and the model lands at ~¥220B of equity vs ¥176B market cap — a 25% gap the blended 8.7× EV/EBITDA does not pay for.
  • Bear stress case: ¥4,200. Management itself authored a 560 bps margin walk-down — consolidated op margin from 22.1% (FY24) to 16.5% (FY28). Reclaim segment margin has already slid 370 bps over three years from a 40.6% FY22 peak to 36.9% in FY25. Three mid-term plans have been quietly retired in three years on the non-reclaim side.
  • The shared fact. Reclaim segment margin sits in a nine-year 35.1%–40.6% band that no listed wafer peer approaches. Cushion to the published moat-breaker (sub-35% for two quarters) is ~200 bps and the direction is wrong. Q1 FY26 rebounded to 38.5% on full Sanbongi/Tainan utilization — the first datapoint that tilts the live debate toward bull mechanics.
Same metric, two reads. Two consecutive prints at 37%+ would validate the bull mechanics; two below 35% would trigger the bear's own thesis-breaker. The nine-year SOTP gap is the variable being resolved — in one direction or the other.
4 · What just happened

A parabolic +150% one-year rally meets management's own four-year margin walk-down.

  • The breakout. Sideways ¥2,500–4,000 for four years, then ¥4,550 → ¥6,640 in 30 days (April 14 → May 15). Weekly RSI hit 87 — the 10-year max. Golden cross active since August 2025. Realized vol 67%, in the stressed band. Sanbongi Plant 7 capex was pulled forward a full year to FY26 because customer LTAs are pre-signed.
  • The disconnect. Only one published sell-side target sits on the tape (¥4,702) — 30% below market and unrefreshed since the rally began. No new initiations. The market price has become the consensus signal, and is implicitly trusting the FY28 ¥115B / ¥19B plan at face value.
  • The single hard date. Q2 FY26 earnings on 7 August 2026 — the only hard-dated event in the next 90 days. The segment table on page 5 prints reclaim margin under live Sanbongi 7 ramp dilution. One number will set the trajectory for the next two quarters.
Capex on a commodity input is rarely accelerated. An LTA-backed pull-forward is the moat showing up as contracted demand — or the precursor to a low-utilization commissioning if the demand forecast misses.
5 · Bull & Bear

Lean watchlist — moat is real, entry is poor, and one August earnings print resolves the argument.

  • For. Reclaim margin band 35.1%–40.6% across nine years; rose in the FY19 memory glut and held 35.1% through FY20 COVID — the cleanest counter-cyclical evidence in listed wafers. Peers traversed 25+ point swings in the same window.
  • For. Sanbongi Plant 7 was pulled forward a full year because customer LTAs are pre-signed. ¥40.5B capex commitment lifts reclaim capacity from 690K to 1.19M wafers/month by 2028. Net cash ¥76B is 43% of market cap — capex self-funds.
  • Against. Management itself walks operating margin down 560 bps through FY28 — the FY28 endpoint sits below FY24. Reclaim segment margin has slid 370 bps over three years; cushion to the published moat-breaker is ~200 bps.
  • Against. Three mid-term plans retired in three years on the non-reclaim side. The Feb 2024 ¥131B Upside Plan was replaced with ¥84B FY26. LE System's ¥24B FY26 target became <¥1B actual — a 96% miss. The FY28 ¥115B plan earns the same discount.
At ¥6,640 with RSI 87 and 49% above the only published consensus mark, paying through August's segment table is poor risk control. Two reads ≥37% with Sanbongi 7 above 85% utilization would shift the setup toward long; two prints below 35% would shift it toward avoid.

Watchlist to re-rate: Q2 FY26 reclaim segment operating margin (7 Aug 2026, ~80 days out); Sanbongi Plant 7 commissioning utilization through H2 FY26; the Feb 2027 mid-term plan refresh — whether it walks back the FY28 ¥115B / ¥19B targets a fourth time, or reaffirms with a first material buyback.