We disclose and quantify humanoid revenue, not positioning. We have secured AI-robot orders as customers shift toward mass production, broken out separately from industrial robots. But social implementation among the makers we target has slowed considerably versus our initial expectations. We were too dependent on a single AI robot; the new plan diversifies away from that. AI robots remain a small, early part of the business; industrial robots and semiconductor equipment are the volume today.

We reported that the year ended March 31 2026 with net sales of ¥59,557 million, up 7.0%, and operating profit recovering to ¥2,567 million from a near‑zero ¥6 million the year before. Orders received rose 16.2% to ¥61,613 million, giving a book‑to‑bill above one‑to‑one, while factory utilization and orders for industrial robots and semiconductor equipment improved. Net profit fell 53.7% only because the prior year included a one‑time ¥5,779 million securities gain, and the recovery was aided by a cost‑innovation project saving ¥1.2 billion and the first price revision in about 15 years.

Beyond our humanoid work, we supply specialty precision components for semiconductor manufacturing equipment, and sales there have risen as data‑center and generative‑AI demand expanded, with a rapid order turnaround since early 2026. We note that this semiconductor leg is cyclical, having troughed last year, and we also see aerospace, space and defense as a growth pillar—having met a ¥3 billion space‑revenue target and observing rising defense budgets and easing export rules, especially in Germany.

We see pressure to commoditize reducers, so we are moving up the value chain, shifting from selling reducers alone to higher‑value‑added actuators with built‑in drivers that can serve as a platform for all axes, so customers need not design reducers and motors from scratch. We target substantial output in about 1.5 years, and we are also developing an ultra‑compact gearbox dexterous hand with Minebea Mitsumi, shown at CES 2026. Whether this added value holds as challengers narrow the precision gap remains an open question we flag.

Beyond humanoids we supply precision components into semiconductor equipment. Sales rose on data-center and generative-AI demand, with orders turning rapidly since early 2026. Semiconductors are a medium- to long-term growth trend, especially in the US, though the cycle troughed last year. The second pillar is aerospace, space and defense: we met a 3 billion yen space-revenue target last year, parts fly on Airbus, and rising defense budgets plus easing export rules — Germany furthest on dual-use — support the outlook. This is broad dual-use aerospace, not drone-specific.

Making a strain-wave gear well demands high-precision grinding machines limited worldwide, specialized flexspline metallurgy, micron tolerances, and long qualification cycles — which is why independent sell-side analysis ranks the harmonic reduction gear the highest humanoid subsystem at 16 of 18, maximum on content value, technology barrier, manufacturing barrier, and humanoid-driven demand, and names us top pick at that layer. The gear sits in the precision-grinding and metallurgy tier, above the actuator-assembly layer. These scores carry verify-at-the-filings status.

Our strengths sit on the balance sheet: 72.2% equity ratio, net cash, roughly 19.1 billion yen cash with long-term borrowings being repaid. Operating cash flow stayed positive through the down-cycle at about 6.4 billion yen. Capex of about 5.7 billion yen went into recovery, including a US-subsidiary land purchase for future capacity. The weakness is in the P&L: operating margin 4.3%, ROE 2.0%, both far below our through-cycle targets. Recovery is underway but unproven against the new plan.