DD Evolution: 9434.T / 6988.T / 6506.T
2026-06-07 19:16 · 14.4 KB
Date: 2026-06-07 | Type: Evolve | Stocks: 3
Executive Summary
This evolve round updated three stocks with the oldest data in the portfolio. Key findings:
- 9434.T SoftBank Corp: UPGRADED to MEDIUM-HIGH — FY2025 record results (Rev ¥7.04T, NI ¥550.8B), first dividend increase in 5 years, GPU Cloud Oct 2026, supply chain filled from 0→7 links (NVIDIA, Oracle, OpenAI, Ericsson, Qualcomm, Cosmos Lab).
- 6988.T Nitto Denko: HOLD — CRITICAL thesis correction: T-glass attribution was wrong (T-glass = Nittobo/3110.T, not Nitto Denko). Actual thesis: semiconductor process tapes (NF735 for HBM/2.5D/3D packaging) + optical films + IBM JDA.
- 6506.T Yaskawa Electric: HOLD — Dash 35 plan confirmed (OP ¥100B / 15.4% margin by FY2029). NVIDIA Physical AI partnership and SoftBank MOU captured. July 9 US tariff expiry is a BINARY risk event.
9434.T — SoftBank Corp (UPGRADED: MEDIUM → MEDIUM-HIGH)
Thesis Update
SoftBank Corp is executing a credible transformation from pure telco to Japan's sovereign AI infrastructure platform.
FY2025 results (ended March 31, 2026) — All four targets beaten (after two upward revisions):
- Revenue: ¥7,038.7B (+8% YoY, record high)
- Operating Income: ¥1,042.6B (+5% YoY)
- Net Income: ¥550.8B (+5% YoY, record high)
- Adjusted FCF: >¥600B (target achieved); Capex: <¥330B (target achieved)
FY2026 Guidance: Revenue ¥7,500B (+7%), OP ¥1,100B (+6%), NI ¥560B (+2%)
New Medium-Term Plan (FY2026–FY2030):
- Operating Income target: ¥1.7T by FY2030
- Net Income target: ¥700B by FY2030
- Dividend: ¥8.80 FY2026 (first increase in 5 years), targeting ¥10/share by FY2030
Key Developments Since Last Review (April 2026)
1. IDC Frontier absorbed April 1, 2026 — Integrated DC+AI platform complete
2. Tomakomai Hokkaido DC: ~60-70% complete as of Feb 2026; Phase 1 (50MW) targeting FY2026 operational; full 300MW+ planned
3. AI Data Center GPU Cloud (October 2026 launch): NVIDIA GB200 NVL72, Infrinia AI Cloud OS, sovereign AI positioning for Japanese enterprises
4. Osaka Sakai AX Factory: 140MW AI DC + 110 ExaFLOPS compute announced; former Sharp LCD plant
5. Crystal Intelligence (SB OAI Japan JV): First deployer across 2,500 internal systems; commercial rollout to enterprises ongoing
6. Oracle Alloy Cloud PF Type A: East Japan live April 2026, West Japan October 2026 — 200+ OCI services
7. Zinc-halogen battery factory (Cosmos Lab + DeltaX JV): Mass production FY2026, ¥100B+ revenue target FY2030
Supply Chain (0 → 7 links)
| Customer/Partner | Product | Evidence | Date |
|---|---|---|---|
| NVIDIA | GB200 NVL72 GPUs for AI Data Center GPU Cloud | confirmed | May 2026 |
| Oracle | Alloy sovereign cloud, 200+ OCI services in SoftBank DCs | confirmed | Oct 2025 |
| OpenAI (SB OAI Japan JV) | Crystal Intelligence enterprise AI, 2,500 internal deployments | confirmed | Feb 2026 |
| Ericsson | Cloud-native 5G Core modernization + AI-powered Massive MIMO | confirmed | Mar 2026 |
| Qualcomm Technologies | 5G Advanced/L4S commercial trial partner | confirmed | Jan 2026 |
| Cosmos Lab + DeltaX | Zinc-halogen battery JV, Sakai factory | confirmed | May 2026 |
| Japan enterprise sovereign AI buyers | GPU Cloud + Oracle Alloy target market | probable | May 2026 |
Metrics Update
| Metric | Old (Apr 2026) | New (Jun 2026) | Source |
|---|---|---|---|
| PE (TTM) | 16.4x | 18.80x | StockAnalysis/SoftBank IR |
| Forward PE | 19.5x | 17.50x | StockAnalysis consensus |
| P/B | 2.33x | 3.51x | mlq.ai |
| ROE | 14.4% | 17.42% | SoftBank FY2025 IR |
| Op. Margin | 17.0% | 14.76% | StockAnalysis |
| Adj. FCF | +¥124B | +¥600B | SoftBank FY2025 IR |
| Market Cap | N/A | ¥10.2T | derived |
| Dividend Yield | 3.9% | 4.16% | SoftBank IR |
Key Upcoming Catalysts
- Jul 2026: Q1 FY2026 earnings — Enterprise segment +20% trajectory check
- Aug 2026: Tomakomai Phase 1 (50MW) operational status
- Oct 2026: AI DC GPU Cloud commercial launch (NVIDIA GB200); Oracle Alloy West Japan
- FY2026: Battery factory mass production start
Risk Flags
- GUARDRAIL — P/B 3.51x: Elevated for telecom sector. Override justified: DC transformation (IDC Frontier, GPU Cloud, Tomakomai, Osaka Sakai) represents structural shift from pure telco to AI infrastructure operator. FY2025 record results confirm execution.
- CONCEPT STOCK WATCH: DC/AI revenue still <10% of ¥7T total. Not flagged as concept stock (telco is the stable core funding DC buildout), but monitor for DC segment revenue disclosure to confirm thesis is materializing.
- Thesis breaks if: Tomakomai construction delayed >6 months past FY2026 target; GPU Cloud <50% utilized by March 2027; D/E exceeds 4.0x without revenue offset.
6988.T — Nitto Denko Corporation (HOLD — Thesis Corrected)
CRITICAL CORRECTION
The original thesis attributed T-glass manufacturing to Nitto Denko (6988.T). This is INCORRECT.
- T-glass (low-dielectric E-glass for AI PCB substrates) is made by Nittobo / Nitto Boseki (3110.T) — a completely separate company
- Nitto Denko (6988.T, 日東電工) makes optical polarizing films, semiconductor process tapes, and industrial adhesives
- The corrected thesis below replaces all T-glass references
Corrected Investment Thesis
Nitto Denko's legitimate AI investment case:
1. NF735 plasma dicing tape for HBM (HBM3/HBM3E), 2.5D/3D packaging at TSMC and Samsung → direct AI infrastructure exposure through wafer-level process consumables
2. IBM JDA (Sep 2025) for advanced chiplet packaging polymeric materials (warpage control, thermal management, crosstalk suppression) → R&D-stage upside confirmed by primary press release
3. Dominant OLED/LCD polarizing films (~71% oligopoly with Sumitomo Chemical + LG Chem) for Apple/Samsung → stable high-margin revenue base
FY2025 Results (ended March 31, 2026)
- Revenue: ¥1,028.2B (+1.4% YoY)
- Operating Profit: ¥183.6B (-1.1% YoY); OP Margin: 17.9%
- Net Income: ¥133.5B (-2.7% YoY); EPS: ¥197.20 (post 5-for-1 split basis)
- FCF: ¥84.7B (down from ¥102.8B prior year)
- Dividend raised: ¥60→¥64/share
FY2027 Guidance: Revenue ¥1,065B (+3.6%), OP ¥193B (+5.1%), NI ¥141B (+5.6%)
Supply Chain (corrected and expanded)
| Customer | Product | Evidence | Date |
|---|---|---|---|
| Apple | Optical polarizing films (NPF) for iPhone/iPad OLED/LCD | confirmed | ongoing |
| IBM Research | Polymeric materials for advanced chiplet packaging (JDA) | confirmed | Sep 2025 |
| Samsung (HBM/Pyeongtaek) | NF735 plasma dicing tape for HBM3/HBM3E + 2.5D/3D | probable | 2024-present |
| TSMC | Dicing tapes, back-grinding tapes, DAF films | probable | 2024-present |
| Samsung Display | Circular polarizers, retardation films for OLED | probable | ongoing |
Metrics Update
| Metric | Old (Apr 2026) | New (Jun 2026) | Source |
|---|---|---|---|
| PE (TTM) | 15.5x | 15.82x | StockAnalysis |
| Forward PE | 16.0x | 11.35x | TipRanks consensus |
| P/B | 1.96x | 2.0x | CompaniesMarketCap |
| ROE | 12.8% | 12.2% | Nitto FY2025 IR |
| Op. Margin | 18.8% | 17.9% | FY2025 actual |
| FCF | +¥49B | +¥84.7B | FY2025 IR |
| Dividend Yield | 1.9% | 2.05% | ¥64/share |
| Market Cap | N/A | ¥2.1T | derived |
Morningstar fair value estimate: ¥5,179 vs market ~¥3,115 (~66% upside)
Key Upcoming Catalysts
- Jul/Aug 2026: Q1 FY2027 earnings — semiconductor tape revenue growth and Optronics segment recovery
- 2026-2027: IBM JDA commercialization announcement (if R&D successful)
- Ongoing 2026: China export control escalation — formal restrictions on semiconductor materials exports would be a significant negative
Risk Flags
- CRITICAL: Thesis corrected — T-glass attribution removed. Prior MEDIUM conviction was based on incorrect company identification. Corrected thesis (semiconductor process tapes) is legitimate but narrower and less direct than a T-glass monopoly would have been.
- CHINA EXPORT CONTROL RISK: Nitto is on China's watch list. Asia-Pacific ~60% of revenue. Formal restrictions on semiconductor tape exports to Chinese chipmakers = material downside (HIGH severity).
- ROE below target: 12.2% vs 15% company goal. FCF declined ¥102.8B→¥84.7B YoY — rising capex intensity.
- Concept stock partial flag: True AI-adjacent revenue (HBM dicing tape, IBM JDA) is likely 10-20% of total — not <5% threshold, but not a pure-play AI company.
- Thesis breaks if: China formally restricts Nitto semiconductor material exports; NF735 loses TSMC/Samsung design wins to competitors (Lintec, Mitsui); Apple sources OLED polarizers from alternate suppliers.
6506.T — Yaskawa Electric (HOLD — Positive Bias)
Thesis Update
World's #1 servo motor maker (~30% global share) is transitioning to a Physical AI platform, anchored by the Dash 35 mid-term plan.
FY2025 results (ended Feb 28, 2026) — NI decline is FX/cost distortion, not structural:
- Revenue: ¥542.1B (+0.8% YoY)
- Operating Profit: ¥47.3B (-5.7% YoY); OP Margin: 8.7%
- Net Income: ¥35.2B (-38.2% YoY) — driven by FX translation losses + abnormal large low-margin projects (non-recurring)
- FCF: ¥11.8B (OCF ¥49.2B minus capex ¥37.4B)
FY2027 Guidance: Revenue ¥580B (+7%), OP ¥60B (+27%), NI ¥47B (+33%)
Dash 35 Mid-Term Plan (May 22, 2026) — CONFIRMED
- FY2029 targets: Operating Profit ¥100B (2.1x FY2025), Operating Margin 15.4% (vs 8.7% today)
- Vision 2035: OP Margin 20%+, Dividend Payout Ratio 40%+
- Humanoid robots: R&D stage — no unit/revenue targets disclosed; Tokyo Robotics acquisition (Jul 2025) for actuator development; CEO: "commercial value in 2-3 years"
- Primary KGIs: Operating profit and operating margin (most important metrics to track)
- Sources: Dash 35 PDF, Vision 2035 PDF
Physical AI Differentiation — Confirmed
1. NVIDIA MOTOMAN NEXT (GTC 2026, March 2026): Integrates Isaac Manipulator (zero-shot manipulation), Omniverse (digital twin commissioning), Jetson (real-time AI inference) — confirmed press release
2. SoftBank MOU (December 1, 2025): Physical AI robots for office/hospital/school using AI-RAN + VLA model; demonstrated at iREX 2025
3. Tokyo Robotics acquisition (July 1, 2025): Humanoid actuator R&D (Waseda spinout); developing wheeled humanoid with dual arms
Tariff Risk Assessment (MEDIUM)
- Americas revenue: ~18-20% of total (~¥94B in 9M FY2025)
- July 9, 2026 BINARY EVENT: 90-day US tariff pause expiry — 24% reciprocal tariff on Japan goods potentially resumes
- Mitigation: $200M+ US local manufacturing (Franklin, WI + Miamisburg, OH); management has incorporated tariff uncertainty into FY2027 guidance
- Worst case: ~4-5% revenue at risk; partially mitigated by local production ramp
Supply Chain (Updated)
| Customer | Product | Evidence | Date |
|---|---|---|---|
| NVIDIA | MOTOMAN NEXT — Isaac Manipulator + Omniverse + Jetson | confirmed | Mar 2026 |
| SoftBank | Physical AI MOU — office/hospital/school robots, AI-RAN+VLA | confirmed | Dec 2025 |
| Toyota | SFA method — robot replication of skilled worker techniques | confirmed | 2025-ongoing |
| Tokyo Robotics (acquired) | Humanoid actuator R&D (Waseda spinout) | confirmed | Jul 2025 |
| GM / VW / auto OEMs | Industrial robots for automotive manufacturing | inferred | ongoing (unverified) |
*Note: GM/VW downgraded from "confirmed" to "inferred" — no direct press release confirmation found in 2025-2026 searches.*
Metrics Update
| Metric | Old (Apr 2026) | New (Jun 2026) | Source |
|---|---|---|---|
| PE (TTM) | 38.88x | 21.38x | StockAnalysis |
| Forward PE | N/A | 29.16x | StockAnalysis consensus |
| P/B | 2.77x | 2.80x | Investing.com |
| ROE | 7.84% | 7.84% | FY2025 actual |
| FCF | +¥5.9B | +¥11.8B | FY2025 actual |
| Market Cap | ¥1.37T | ¥1.86T | derived at ¥7,140 |
| Dividend Yield | N/A | 0.95% | ¥68/share at ¥7,140 |
*PE TTM distorted by FY2025 NI collapse. Forward PE 29.16x on FY2027 EPS is most relevant.*
Key Upcoming Catalysts
- Jul 3, 2026: Q1 FY2026 earnings (May 31 quarter) — CRITICAL: confirm order-to-revenue acceleration; upgrade trigger if >¥140B quarterly revenue
- Jul 9, 2026: BINARY EVENT — US 90-day tariff pause expiry; 24% tariff reinstatement risk
- H2 2026: MOTOMAN NEXT Physical AI commercial launch + first bookings
- H2 2026: Kitakyushu factory ramp (+50% capacity); demand absorption test
- Oct 2026: Q2 FY2026/H1 FY2026 earnings — ¥580B revenue guidance tracking
- Dec 2026: iREX 2026 — Physical AI and humanoid actuator showcase
Risk Flags
- GUARDRAIL — Forward PE 29x: Well above Japan machinery average ~15x. Market is pricing full Dash 35 execution. Override justified: Dash 35 is confirmed and credible (¥100B OP = 2.1x recovery from depressed FY2025 base, not speculation). However, any execution shortfall = significant multiple compression.
- GUARDRAIL — FCF growth cap: FCF from ¥11.8B base could grow >15% p.a. if OP reaches ¥100B. Override: growth is recovery from distorted FY2025 NI base (FX+cost anomaly), not extrapolation of elevated rates.
- Capacity expansion risk (AP02): New Kitakyushu factory +50% capacity. All 4 major robot OEMs expanding. Monitor utilization H2 FY2026.
- Thesis breaks if: Dash 35 targets revised down before FY2028; China revenue declines >20% YoY for 2 consecutive quarters; July 9 tariff reinstated at 24% AND US manufacturing not sufficiently ramped by year-end.
Methodology
Research conducted June 7, 2026 using:
- Tier 1 sources: Company IR filings (TDnet/EDINET), SoftBank IR press releases, Yaskawa FY2025 annual results, Nitto Denko JPX TDnet filing
- Tier 1 cross-validation: J-Quants data (used for base metrics); StockAnalysis.com (PE, PB, market cap, split-adjusted)
- Tier 2 sources: TipRanks, Investing.com, Quartr (earnings summaries), NVIDIA press releases, Oracle press releases
- Supply chain: Primary press releases for confirmed links; secondary industry reports for probable; no evidence for inferred
- Anti-patterns: All three stocks checked against AP01 (peak earnings), AP02 (capacity hangover), AP03 (concept stock)
- Guardrails: All valuation guardrails from
data/config/valuation_guardrails.jsonchecked; overrides documented inline
*This report is for internal research purposes only. Not investment advice.*