We have written a lot about who benefits from AI capital expenditure. Samco supplies the equipment that deposits films on chips. FUJI Co. assembles the machines that place components on circuit boards. Tomen Devices distributes the memory modules that populate AI servers. The supply chain is long, and most of it is invisible to equity investors focused on the hyperscalers.
Nittobo (TSE:3110) is near the beginning of that chain, and its Q1 FY2027 results make the case that proximity to the raw material layer is, right now, extremely profitable.
What Nittobo Actually Makes
Nittobo is formally classified as a glass fiber and specialty materials company — a category that sounds like 1970s industrial Japan. The reality in 2026 is different. Its electronic materials division produces two specific types of specialty glass:
- Low-dielectric specialty glass: used in PCB substrates inside AI servers to reduce signal propagation delay as data moves at higher frequencies
- Low thermal-expansion specialty glass: used in semiconductor package substrates to maintain dimensional stability as chips heat up during operation
Neither product is a commodity. The physical properties required — specific dielectric constants, controlled expansion coefficients — are difficult to replicate at scale. Nittobo has been refining these materials for decades and holds a niche position that newer entrants cannot quickly displace.
The Numbers
| Metric | Q1 FY2027 | Q1 FY2026 | YoY |
|---|---|---|---|
| Revenue | JPY 34.0bn | JPY 28.2bn | +20.6% |
| Operating Profit | JPY 7.94bn | JPY 4.30bn | +84.8% |
| Operating Margin | 23.3% | 15.2% | — |
The headline numbers are strong. The segment breakdown is where the story lives.
Segment Breakdown: One Division Carries the Company
| Segment | Revenue | YoY | Op. Profit | YoY | OPM |
|---|---|---|---|---|---|
| Electronic Materials | JPY 14.6bn | +29.5% | JPY 7.04bn | +70.0% | 48.2% |
| Medical | JPY 3.8bn | +6.8% | JPY 0.71bn | +66.8% | 18.5% |
| Insulation | JPY 4.4bn | +24.2% | JPY 0.25bn | swing to profit | 5.7% |
| Other | JPY 4.8bn | +11.8% | JPY 0.21bn | +90.8% | 4.5% |
| Materials & Chemical | JPY 2.6bn | +15.4% | JPY 0.22bn | +103.3% | 8.5% |
| Composites | JPY 3.8bn | +15.9% | JPY 0.00bn | swing to profit | 0% |
Electronic Materials accounts for 42.9% of revenue but 88.6% of operating profit. Its 48.2% operating margin is not a rounding error — it reflects genuine pricing power in a supply-constrained niche.
The remaining five segments combined generated JPY 1.39bn in operating profit. They matter for revenue diversification, and several swung from loss to profit this quarter, but they are not the investment thesis.
The Backlog Question
Nittobo does not disclose order backlog in its quarterly filings — standard practice for Japanese specialty materials companies. This is a genuine limitation for investors trying to assess demand durability.
The available proxy is management guidance behavior. At the Q1 stage — with only one quarter of data — Nittobo raised its full-year operating profit forecast by 15.4%:
| Previous Guidance | Revised Guidance | Change | |
|---|---|---|---|
| H1 Operating Profit | JPY 12.6bn | JPY 15.3bn | +21.4% |
| Full-Year Operating Profit | JPY 26.0bn | JPY 30.0bn | +15.4% |
The implied H2 operating profit is JPY 14.7bn — a +29% increase over H2 FY2026. Management revised up the back half of the year, not just the front half they could already see. That is the closest available signal to backlog confidence.
The qualitative language in the filing supports this: demand for AI server specialty glass is described as “continuing to expand” with no hedging language about cyclical risk.
The Net Profit Distortion
Full-year net profit guidance of JPY 20.0bn represents a -52.1% decline versus FY2026 actual net profit of JPY 41.8bn. This looks alarming and warrants explanation.
FY2026’s net profit was inflated by substantial one-time gains outside core operations. The operating profit trajectory — FY2026 actual JPY 20.8bn → FY2027 guidance JPY 30.0bn (+44%) — tells the operational story accurately. Investors focused on the net profit line are reading the wrong metric for this company in this period.
Where This Fits in the AI Supply Chain Thesis
This site’s analysis of the AI capex wave has followed a consistent thread: the most interesting beneficiaries are often not the companies building AI models or selling compute, but the companies whose specialized products are embedded in the physical infrastructure those models require.
Samco makes the deposition equipment. FUJI assembles the machines. Nittobo makes the glass that holds the circuit boards together. The further upstream you go, the harder it is to substitute the supplier — and in Nittobo’s case, a 48% operating margin suggests the market agrees.
The risk is cyclicality. AI server demand has driven an exceptional Q1, and management’s guidance revision implies confidence through FY2027 March. What happens in FY2028 depends on whether hyperscaler capex sustains at current levels or normalizes. Nittobo has no disclosed backlog to anchor that answer; the guidance revision is the best available proxy, and it is currently pointing up.
Source: Original filing (TDnet) | Earnings report (TSE:3110) | 日本語版
Disclaimer | This article is for informational purposes only and does not constitute investment advice.