Record sales and a 340bp margin reset in the same year — transitory first-unit cost or structural mix-down.

• Order book at the cycle high. FY2026 revenue printed a record ¥54.4B with book-to-bill 1.10 and backlog around 55% of next-year sales.
• Margin at the cycle low. Gross margin fell 340bp to 33.8% and operating margin from 16.6% to 12.7% — the lowest in five years — on first-unit costs for HBM and panel-level packaging tools plus a transfer-heavy mix.
• The third 'transitory' year in a row. Management framed FY2023 as FX, FY2025 as memory-delay, FY2026 as first-unit cost — the recurring transitory is the cycle if Q4's snapback doesn't carry into 1H FY2027.

The market is queueing for November when Q4 already ran the experiment.

• Q4 FY2026 is the print everyone is waiting for. Shipped ¥17.44B at an 18.5% operating margin — the highest quarter of the year and above the FY2024–FY2025 through-cycle band. Annualized, that's ¥69B revenue and ¥12.9B operating profit, above management's own FY2027 guide of ¥64B / ¥10.2B.
• Consensus is doubly-discounting. Trailing 42× P/E built on a base that includes Q1 FY2026's ¥581M operating loss; the 24% five-session post-print drawdown to ¥2,579 is the market paying full price to wait six months for a confirmation that has already cleared in Q4.
• INNOMS pricing is in the multiple at zero. The next-generation compression platform targets a +20–30% ASP premium and is in field evaluation, with a first commercial order expected this fiscal year. Both prior platform launches — 2009 original compression and FY2024 CPM1080 — held their targeted premium pricing. Consensus models nothing for the third.

A clean, conservative balance sheet absorbing one ugly cash year that is mostly inventory and capex.

Operating cash flow halved to ¥4.1B as inventory built to ¥19.8B for the FY2027 ramp and receivables stretched from 80 to 110 days; capex stepped up to ¥5.5B to fund Kyushu and Malaysia capacity. There is no acquisition baggage, no goodwill, no stock-based comp of note, no factoring — the cash hole is mechanical and growth-funded. The FY2027 guide of ¥64B sales and ¥10.2B operating profit needs gross margin to snap from 33.8% back toward 37% to land.

Highest single-step share in the peer set, lowest multiple — by a factor of four.

• 21× EV/EBITDA vs 87× BESI and 126× Hanmi. The same HBM cycle drives all three. TOWA owns ~63% of global molding share — up from 59% in FY2022 — and effectively all of the high-end compression sockets that encase every HBM stack shipping today.
• The discount has a real reason. AI capex is flowing to bonders, not molders. TCB bonder TAM is growing 30% CAGR to $1.6B by 2028; BESI has installed over 150 hybrid bonders with a mid-case 350 by 2030. Hybrid bonding eventually removes the encapsulation step at the highest-end sockets, capping the multiple even if the cyclical recovery lands.
• But share is still rising, not falling. Book-to-bill 1.10 is the highest in the peer set; no compression socket has yet been lost at SK Hynix, Samsung, or Micron. The market is paying for share loss that hasn't happened — and a confirmed share-durability print through HBM4E is the watchpoint for whether the discount narrows or persists.