Home/Reports/Evolve DD Report — 2026-06-04

Evolve DD Report — 2026-06-04

2026-06-04 19:16 · 12.6 KB

Type: Deep Evolution

Stocks updated: Nitto Boseki (3110.T), Daikin Industries (6367.T), Rohm (6963.T)

Report date: 2026-06-04

Method: T-Glass Supply Chain Bottleneck Analysis + ReACT verification


Executive Summary

Three stocks evolved today spanning AI packaging materials (Nitto Boseki), data center cooling infrastructure (Daikin), and power semiconductor consolidation (Rohm). All three had stale data from April 2026; May earnings have now been incorporated. Key findings: Nitto Boseki's T-glass monopoly is strengthening with dual price+volume expansion; Daikin's DC cooling thesis is real but still below the concept-stock revenue threshold; Rohm's JV talks are materially delayed and the stock is priced well above analyst fundamental consensus.


1. Nitto Boseki (3110.T) — T-Glass Monopoly

Theme: AI / Restocking

Thesis status: HOLD

Data date: 2026-06-04 (previously 2026-04-21)

FY2025 Actuals (Year ended March 31, 2026 — announced May 12, 2026)

MetricFY2025 ActualYoY
Revenue¥118.2B+8.4%
Operating Profit¥20.8B+26.6%
Net Income¥41.8B+225% (asset sales — not recurring)
FY2026 OP Guidance¥26.0B+24.9% guided
FY2027 OP Guidance¥36.0B+38.5% guided

Source: TDnet FY2025 決算短信, May 12, 2026

Key Developments

Supply chain confirmed:

  • **Nan Ya Plastics partnership** (confirmed, Nov 2025): Nan Ya handles 20% of Nittobo's specialty weaving by 2027, providing near-term supply relief while Fukushima ramps. Source: TrendForce, 2025-11-28
  • **Nvidia, Google, Amazon competing for supply** (confirmed, Nov 2025): Named as direct customers competing for Nittobo T-glass. Order books full through end-2026. Source: TrendForce, 2025-11-24
  • Ibiden ABF substrates (probable): T-glass is explicit spec requirement for ABF substrates; no public contract.

Pricing & Volume:

  • T-glass price raised +20% August 2025; JPMorgan models +25% pricing from 2026 onward
  • Shipment growth guided +20-30% YoY — rare dual price+volume expansion simultaneously
  • Nvidia GPU substrate area growing from 3,190mm² (Hopper) to ~8,000mm² (Rubin) + more layers → multiplies T-glass content per chip

Capacity: Fukushima factory (¥15B, 3x capacity, METI ¥2.4B subsidy) production start Jan-Mar 2027. Nan Ya partnership provides interim relief.

Analyst actions: JPMorgan upgraded to Overweight (T-glass pricing +25% from 2026); Morgan Stanley upgraded to Overweight. 5/0 Buy/Sell ratio. 5-for-1 stock split effective June 29, 2026.

Anti-Pattern Check

Anti-PatternStatus
Peak Earnings TrapCLEAR: OP +26.6% with FY2026 +24.9% guided. NI distorted by asset sales; operating trend intact.
Concept StockCLEAR: T-glass ~20-30% of electronic materials (~40% of revenue) → well above 5% threshold
Capacity HangoverFLAG 2027+: 3x Fukushima + Nan Ya + competitor qualifications could flip supply. Not imminent.

Risk Flags

  • Competitor qualification 2027+: Apple qualifying Grace Fabric Technology + Taiwan Glass; Chinese quartz cloth reportedly passed Nvidia certification at small scale. Structural watch for 2027+.
  • Intel glass-core substrate: Bypasses T-glass entirely for Xeon 6+ — medium-term threat to non-GPU packaging addressable market.
  • Capacity hangover 2027+: If 3x Fukushima + Nan Ya + alternatives all materialize while AI server demand plateaus.

Inversion

Thesis breaks if: Competitor achieves full-scale Nvidia certification for T-glass alternative (Apple+Grace Fabric or Chinese quartz cloth at volume), OR Fukushima ramp delayed >6 months, OR AI server demand plateaus causing T-glass pricing to reverse.


2. Daikin Industries (6367.T) — AI Data Center Cooling

Theme: Data Center Ecosystem

Thesis status: HOLD

Data date: 2026-06-04 (previously 2026-04-14)

FY2026 Actuals (Year ended March 31, 2026 — announced May 12, 2026)

MetricFY2026 ActualYoY
Revenue¥5.02T+5.5% (record)
Operating Profit~¥435B+5% (record)
Net Income¥275.2B+4.0% (record)
OP Margin8.7%
Tariff impact absorbed¥42Bfully absorbed
FX headwind¥71.5Bfully absorbed

Source: Daikin FY2026 4Q Financial Results, May 12, 2026

FUSION 30 Strategic Plan (announced May 12, 2026)

  • FY2028 target: 10% operating margin (vs 8.7% today), 12% ROE (vs 9.26%)
  • FY2030 aspirational: 12% OP margin, 15% ROE
  • Data centers explicitly named as core growth pillar

Source: FUSION 30 Press Release, May 12, 2026

Supply Chain Confirmed

Customer/EntityProductEvidenceDate
Delta ElectronicsCDU co-development MOU (ASEAN-Oceania, 100kW–3,000kW rack density range)ConfirmedDaikin PR, 2026-05-11
Chilldyne (subsidiary)Negative-pressure liquid CDU — eliminates leak risk for GPU racksConfirmedPR Newswire, 2025-11-04
DDC Solutions (subsidiary)Modular high-density server rack cooling cabinets (white-space layer)ConfirmedDDC Solutions, 2025-08

Critical gap: No named hyperscaler contracts (AWS, Google, Microsoft) publicly confirmed. Daikin uses "hyperscale" terminology in marketing.

Valuation Update

MetricOld (Apr 14)New (Jun 4)
PE22.7x25.16x
Forward PE22.1x23.48x
PB1.84x2.1x
ROE8.1%9.26%
OP Margin8.4%8.7%
Div Yield1.6%1.55%
Market CapN/A¥6.9T

Source: StockAnalysis.com TYO:6367

Anti-Pattern Check

Anti-PatternStatus
Peak Earnings TrapMILD FLAG: FY2026 set records in revenue + OP + NI. However, FUSION 30 growth path is credible and tariff headwinds become neutral in FY2027. Not a cyclical peak.
Concept StockFLAG: DC cooling ~3-5% of group revenue — at the concept-stock threshold. Products are real (not vaporware) but revenue not yet material.
Capacity HangoverMILD FLAG: Vertiv, Schneider, Mitsubishi, and Daikin all expanding DC cooling simultaneously. Market large enough for multiple winners (projected CAGR 16.9% to 2036) but pricing pressure risk rises.

Risk Flags

  • Concept stock threshold: DC cooling at ~3-5% of group revenue — investors are paying for optionality not yet in earnings. Named hyperscaler contracts absent.
  • Vertiv lead: USD 15B order backlog (+109% YoY), 3-5yr head start in hyperscaler relationships. Daikin competes for investor capital against a pure-play with better visibility.
  • Valuation: At PE 25.16x with concept-stock element, any guidance miss in FY2027 would pressure the stock.

Inversion

Thesis breaks if: Q1 FY2027 (Aug 2026) DC cooling revenue stagnates AND no named hyperscaler contract materializes AND FY2027 full-year guidance disappoints vs analyst consensus.


3. Rohm (6963.T) — SiC Power Semiconductor + JV

Theme: EV / Restocking

Thesis status: HOLD (elevated caution — do not add at current levels)

Data date: 2026-06-04 (previously 2026-04-20 in restocking, 2026-05-03 in EV)

FY2025 Actuals (Year ended March 31, 2026 — announced May 12-13, 2026)

MetricFY2025 Actualvs. Nov 2025 Guide
Revenue¥481.1BBeat (¥460B guide)
Operating Profit¥10.8BBeat (¥5B guide)
Net Income-¥158.4BMiss (¥9B guide) — SiC impairment
FY2026 OP Guidance¥30.0B+176% from FY2025
FY2026 Revenue Guide¥510.0B+6.0%

Source: Rohm FY2025 決算短信, May 12, 2026; BigGo/TDnet, 2026-05-13

Note: The ¥158.4B net loss (vs. previously known ¥57B FY2024 loss) is a kitchen-sink impairment on SiC fixed assets driven by BEV market slowdown and Chinese competition. Post-impairment, depreciation burden falls sharply — the ¥30B OP recovery in FY2026 is mechanically credible.

JV Status — DELAYED

  • March 27, 2026: MoU signed by all 5 parties (Rohm, Toshiba, Mitsubishi Electric, JIP, TBJH). Combined power semi global rank would be #2 (~9.7-11% market share, above onsemi).
  • April 28, 2026: Denso withdrew ¥1.3T acquisition bid. Rohm preferred JV path.
  • May 15, 2026: Rohm president warned JV talks "more complex and slower than initially expected." No formal equity structure yet.

Sources: Toshiba MoU, 2026-03-27; TrendForce JV delay, 2026-05-15

Supply Chain Confirmed

CustomerProductEvidenceDate
Toyota bZ5SiC MOSFETs via HAIMOSIC JV (~360K units/yr)ConfirmedRohm IR, 2025-06-23
SchaefflerSiC bare chips for HV inverter brickConfirmedRohm PR, 2025-09-04
InfineonSiC package MOU (second-source, EV/DC/PV)ConfirmedInfineon PR, 2025-09-25

Valuation Update

MetricOldNew
PE12.4 (now Loss)Loss (NI -¥158.4B)
PB1.37x2.56x
ROE3%Loss
Market Cap¥1.35T¥1.32T
Analyst consensus PT¥2,944 (9 analysts)
Market price~¥1,350~¥5,400
Implied analyst downside-47%

Source: StockAnalysis.com TYO:6963; Investing.com consensus

Anti-Pattern / Guardrail Flags

CheckStatus
P/B at cyclical peakFLAG: 2.56x vs historical 1.3-1.4x norm. Above estimated 80th percentile.
EV validationFLAG: Analyst consensus PT ¥2,944 vs market ¥5,400+ = >30% divergence.
Still loss-makingFLAG: ROE negative; NI still in loss; first clean profitable quarter not yet achieved.
JV speculation premiumFLAG: Stock 3x'd from lows on M&A; primary catalyst (Denso bid) withdrawn.

Risk Flags

  • JV delay risk: Three-party integration of Rohm + Toshiba TDSC + Mitsubishi Electric involves complex overlapping product lines, manufacturing footprints, and corporate cultures. Rohm president acknowledged complexity; no timeline confirmed.
  • P/B at historical peak: At 2.56x, well above pre-crisis 1.3x norm. Risk of mean reversion if JV stalls.
  • Analyst consensus -47% from market: Fundamental analysts not buying the JV/M&A premium at current levels.
  • SiC impairment worsened: FY2025 NI -¥158.4B worse than FY2024 -¥57B. Management credibility gap.

Inversion

Thesis breaks if: JV talks formally collapse, OR FY2026 OP misses ¥30B target by >30%, OR SiC impairment charge recurs in FY2026, OR Chinese competitor wins a major EV platform customer from Rohm.


Pipeline Status

  • JSON files updated: dcpower_companies.json, ai_companies.json, restocking_companies.json, ev_companies.json
  • Stocks updated: 6367.T (dcpower), 3110.T (ai+restocking), 6963.T (restocking+ev)
  • Kairos data regenerated: Yes
  • Trace file: data/research_traces/2026-06-04_evolve.jsonl

Sources

All sources cited inline above. Tier 1-2 sources used for all metrics:

  • TDnet/Nikkei (Tier 1): Nitto Boseki FY2025, Rohm FY2025, Daikin FY2026 決算短信
  • StockAnalysis.com (Tier 2): All valuation ratios
  • Company press releases (Tier 2): Rohm IR, Daikin IR, Toshiba MoU, Infineon MoU
  • TrendForce (Tier 3): T-glass market context, JV delay report
  • Digitimes (Tier 3): SiC market, substrate upcycle

*This report is for internal research tracking only. Not investment advice.*