Panasonic Energy

Phase A — Understand the business

Lens 1 · Company Overview

What it is. Panasonic Energy Co., Ltd. is the wholly-owned Energy operating company of Panasonic Holdings, headquartered in Moriguchi/Osaka, led by President & CEO Kazuo Tadanobu. It is the descendant of the Sanyo battery business Panasonic absorbed, and it is the company that has supplied Tesla's cells since the original Gigafactory Nevada partnership. The unit makes its money three ways:

1. Automotive lithium-ion cells — the core. Cylindrical cells (2170 format) for EVs, historically ~Tesla-dominated, now diversifying. This is the revenue engine and the source of nearly all the volatility.
2. Industrial / consumer batteries — dry cells, smaller Li-ion, devices.
3. Energy-storage systems (ESS) — the fast-emerging leg: data-center backup (battery backup units / capacitor backup units) where the unit claims ~80% share of the specialized space-saving high-output niche.

The business model in plain terms. This is a capital-intensive, customer-concentrated cell manufacturer competing in a commoditizing-from-below market. Cells are sold under multi-year supply agreements to automakers; the dominant historical buyer is Tesla (Nevada cells; Kansas De Soto plant built largely for Tesla demand). Revenue is volume × cell-ASP, and both legs are under pressure: Tesla US volume softened through 2025, and cell ASPs are falling industry-wide as Chinese LFP scale collapses prices.

Scale markers (parent, the lens that has numbers). Panasonic Holdings FY3/25 (yr-end Mar 2025): net sales ¥8.46T, operating profit ¥426.5B (+18.2% YoY). FY3/26 (yr-end Mar 2026): net sales ~¥8.05T (−4.84% YoY), net income ¥189.5B (−48.2% YoY), adjusted operating profit ¥447.4B. The Energy segment is one of the three named profit pillars (with Connect and Industry) — but it absorbed a one-time ~¥40B automotive-battery defect charge in FY3/25 that reversed in FY3/26.

Contract structure / payment terms. Automotive cell supply is take-or-pay-adjacent multi-year volume commitments, but the 2025 Kansas ramp delay shows the practical limit: when the anchor customer's demand softens, the supplier eats idle capacity even with incentives intact (see Lens 5). Data-center ESS is being pre-sold — Panasonic reportedly has already sold future data-center battery output, a far better contract posture than the spot-exposed auto book.

Lens 2 · Supply Chain

Map: upstream materials → Panasonic Energy cell/module plants → automaker/data-center buyer.

Upstream (named):

• Cathode active material — nickel-rich NCA/NCM. Suppliers historically include Sumitomo Metal Mining (long-standing NCA partner) and others; nickel/cobalt sourcing is the geopolitical pressure point.
• Anode — graphite today; silicon anode is the differentiator. Panasonic signed a silicon-anode purchase agreement with Sila Nanotechnologies for next-gen energy density. The roadmap also references SiOx (TRL 7–8) vs advanced silicon composites (TRL 5–6) as the live tension.
• Separators, electrolyte, copper/aluminum foil, can/casing — multi-source, but Chinese-origin material is now a regulatory chokepoint under OBBBA's "prohibited foreign entity / material assistance" 45X restriction (Lens 10).

Midstream (the company's own plants — named):

• Nevada (Gigafactory, with Tesla) — original 2170 line.
• De Soto, Kansas — $4B, ~30 GWh initial 2170 line, opened 2025 ("world's largest battery plant") but ramp delayed.
• Wakayama, Japan — 4680 mass-production line (preparations complete, awaiting confirmed customer PO).
• Osaka — vehicle line being repurposed to Li-ion + supercapacitors for data-center demand.
• Mexico — second/third battery-module sites planned for ESS.

Downstream (named buyers):

• Tesla — the anchor; Model 3/Y (2170), Cybertruck/next-gen (4680 intended).
• Mazda, Subaru — alliances finalized late 2025 (diversification).
• Data-center / hyperscaler operators — ESS backup buyers (output pre-sold).

Chokepoints / single-source dependencies:

1. Customer single-source: Tesla. The dominant chokepoint is demand-side — one buyer historically drove the auto book. This is the structural fragility (Lens 13).
2. Chinese material assistance — 45X eligibility now conditions on excluding "prohibited foreign entity" content; cathode/anode/separator supply chains run through China, so qualifying US-made cells for the full credit is a live, unresolved supply-chain engineering problem.
3. Silicon-anode scale-up — Sila/SiOx is single-vendor-ish at the frontier; bridging TRL 5–6 to production is the energy-density critical path.

Lens 3 · Competitive Advantages (moats)

Where the moat is real:

• Cylindrical-cell process IP + manufacturing yield. Decades of Sanyo+Panasonic cylindrical know-how; ~3,616 cylindrical-cell patents cited. Premium nickel-rich high-energy-density chemistry is genuinely hard to match on quality/safety — the 2170 still tests at ~269 Wh/kg vs Tesla-Austin 4680 ~244 Wh/kg.
• Data-center ESS dominance — ~80% share of the specialized high-output backup niche. This is the most defensible position the unit holds anywhere right now: high-spec, space-constrained, reliability-critical, and the output is pre-sold. It is a real, under-appreciated moat.
• "Made-in-USA, ex-China" positioning — for IRA-era domestic-content buyers and for any customer wanting to de-risk Chinese supply, Panasonic is one of very few non-Chinese, US-manufacturing cell makers at scale. Geopolitics is, perversely, a moat.

Where the moat is weak (be honest):

• No cost moat in volume EV cells. Against CATL/BYD LFP at scale, Panasonic is a price-taker. Chinese players hold ~69% of global EV battery installs and set the price floor. Premium NCA does not win the mass-market cost war.
• Bargaining power over its anchor customer is low — Tesla can second-source (CATL, LG, its own 4680) and did. The 4680 PO delay is the moat-weakness made visible: Panasonic prepared a line and is waiting for the customer to commit.

Net: moat = process/quality + the ESS niche + ex-China geography. It is not a cost or a customer-lock-in moat in autos. The durable edge is migrating out of autos and into data-center ESS.

Lens 4 · Segments

Panasonic Holdings reports operating-company segments, not a clean standalone "Panasonic Energy" P&L in the English summaries I can source — so segment-level figures here are n/a at the granularity the lens ideally wants (segments.csv is empty). What is sourceable:

Trend & cause: Auto-battery revenue decelerated in FY3/25–FY3/26 (Tesla US softness + the ¥40B defect charge). The segment story is a mix shift: autos flat-to-soft, data-center ESS inflecting hard (targets imply ESS going from a rounding error to ~half the unit's ¥2T FY2029 sales). Geographically the bet is North America-heavy (Kansas, Nevada, Mexico modules) plus a Japan cell-tripling by FY2029. The directional read: the unit is re-segmenting itself away from its own legacy — the growth is no longer "more Tesla cells."

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print — FY3/26, reported May 2026)

Headline (parent). Net sales ~¥8.05T, −4.84% YoY; net income ¥189.5B, −48.2% YoY; adjusted operating profit ¥447.4B. The net-income halving is the eye-catcher and it is restructuring-driven, not operational collapse.

What drove it:

• ¥174.5B structural-reform / restructuring costs (the 10,000-job group reform) hit the P&L.
• Energy positive swing: automotive batteries benefited from the reversal of the prior-year ¥40B defect charge plus expanded North American plant sales reaching 46 GWh. So the operating trend in Energy was up even as group net income fell — the decline is reform expense + a weaker top line elsewhere.
• Mid-year guidance cut (FY3/26): Panasonic had cut its FY profit forecast by ~13.5% in Oct 2025 citing a weak US EV battery market — the auto-demand reality.

Guidance / outlook (the important part):

• FY3/27 group adjusted OP target ¥600B (up from ¥447.4B). The earlier "≥¥600B in FY3/27 via management reform" framing is intact.
• Battery-unit profit guided +39% for the relevant year per Reuters — management is signaling Energy profit acceleration.
• Tone shift: 2025 was branded "Group Management Reform"; 2026 is branded "Achieve a Turnaround for Growth". CEO Kusumi returned ~40% of his comp — an accountability signal.

Balance-sheet / capital flags:

• Panasonic sold its entire Tesla equity stake (~$3.6B) to fund strategic investment (incl. the $7.1B Blue Yonder acquisition). Monetizing the Tesla stake while also being Tesla's supplier is a notable de-linking.
• Big capital commitments ahead: ¥500B (~$3.1B) AI-infrastructure budget, of which ¥350B (~$2B) to Energy for data-center ESS across FY2026–28.

Market reaction / what's priced: The stock sits at ¥3,800 (Jun 14 2026) with analyst consensus ¥3,608 — below the price. The market is not paying up for the turnaround; if anything consensus targets imply mild downside. That is the setup: a unit-level acceleration story inside a stock the Street is lukewarm on.

Unusual vs own history: the ¥40B single-customer defect charge (FY3/25) and its reversal (FY3/26) is the anomaly — it flattered the FY3/26 Energy comparison and should be normalized out when judging underlying trend.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk (transcripts/ empty) — sentiment is reconstructed from sourced commentary, labeled ``.

Tonal arc across the last ~4 reporting points:

• FY3/25 results (May 2025): defensive — restructuring announced, 10,000 jobs, CEO pay cut, ¥40B defect charge. Language of "Group Management Reform," accountability, cost-out.
• Mid-FY3/26 (Oct 2025): cautious-to-negative — forecast cut 13.5% on "weak US EV battery market"; Kansas ramp delayed (Jul 2025) on Tesla softness.
• FY3/26 results (May 2026): pivot to optimism — "Achieve a Turnaround for Growth," ¥600B FY3/27 OP target, battery profit +39% guide, ¥500B AI bet unveiled.
• Investors Day (Jun 8 2026): most bullish — ¥350B AIDC storage, ~¥1T data-center sales by FY2029, 80% niche share, "quadrupling AI-related sales".

What management started saying: "AI data center," "energy storage systems," "battery backup / capacitor backup units," "ex-China supply chain," "turnaround for growth." What they stopped saying: the EV-cell-volume-uber-alles / Tesla-centric framing. The Tesla stake sale is the symbolic full stop. Sentiment shift = from cyclical-auto-supplier defensiveness → structural-diversification offense, concentrated on data-center power.

Lens 7 · Comps

Peer set pulled from _index.json (topic = electrification) + obvious global cell makers.

2025 installed-share anchor (well-sourced): CATL 39.2% (464.7 GWh) / BYD 16.4% (194.8 GWh) / LGES 9.2% (108.8 GWh) / … Panasonic 3.7% (≈44 GWh of 1,187 GWh total market) ranked #7.

Read: Panasonic trades at a 46x trailing P/E — but that is the conglomerate multiple distorted by the ¥174.5B restructuring crushing FY3/26 net income (E is depressed → P/E inflated). It is not a clean read on the battery unit, and EV/EBITDA — the multiple that would matter for a capital-intensive cell maker — is not sourceable at the segment level. Conclusion: the comp table is directionally useful on share (Panasonic is a small, premium, non-China #7) but the valuation cells are mostly unsourced — do not anchor a price target on them.

Lens 8 · Stock-Price Catalysts (what actually moves 6752.T)

Move-drivers over the recent window, all ``:

• Battery / EV-demand headlines — the FY3/26 guidance cut on "weak US EV battery market" (Oct 2025) and the Kansas ramp delay (Jul 2025) are the negative catalysts; Tesla demand is a direct swing factor.
• Restructuring announcements — the 10,000-job reform (May 2025) and ¥600B OP target framing move sentiment on the cost lever.
• AI / data-center pivot — the ¥500B AI budget (May 2026) and ¥350B AIDC storage / Investors Day (Jun 2026) are the new positive catalyst class.
• One-off charges — the ¥40B defect charge (FY3/25).
• Macro/FX — as a Japanese exporter with USD-heavy US plants, yen/dollar and US policy (IRA/OBBBA, tariffs) swing the economics.

Pattern: for 6752.T the market reacts most to (a) the EV-demand / Tesla cycle and (b) cost-reform credibility — and is only now starting to price the data-center ESS optionality. The thing the market "reacts to" has historically been the auto cycle; the bull thesis is that the reaction function is shifting to ESS.

---

Phase C — Judge people & books

Lens 9 · Management

• Group CEO: Yuki Kusumi (Panasonic Holdings). Architect of the "Group Management Reform" → "Turnaround for Growth" arc. Skin-in-the-game signal: voluntarily returned ~40% of his compensation over the job cuts — a genuine accountability gesture, rare among Japanese conglomerate CEOs. Track record is the hard, unglamorous restructuring of a sprawling group (holding-company conversion, operating-company autonomy, portfolio pruning — exited US residential solar/home storage 2025; bought Blue Yonder $7.1B for supply-chain software).
• Energy CEO: Kazuo Tadanobu (President & CEO, Panasonic Energy Co.). Runs the cell + ESS business; the public face of the Kansas build-out and the data-center pivot.
• Energy CTO/EVP: Shoichiro Watanabe — CTO, IP, environment, procurement; the voice on the solid-state pilot ("mass production next fiscal year," pilot batch first) and 4680/silicon roadmap.
• Group structure change: new operating-company presidents effective April 1, 2026; Tetsuro Homma moved to Corporate Advisor (Jun 2026).

Capital-allocation history (the real test):

• Disciplined moves: monetized the Tesla stake (~$3.6B) rather than ride it; exited unprofitable US residential solar/home storage; redirected capital into higher-spec industrial/ESS. This is a management team pruning to the defensible — a good sign.
• The big bet: ¥500B AI / ¥350B AIDC storage — concentrated, conviction capital allocation into the one place they have 80% share. Coherent.
• Red flag(s): the ¥40B single-customer battery defect is an operational-quality miss that cost real money and credibility; the Kansas $4B plant built ahead of confirmed sustained demand is the classic capex-ahead-of-orders risk that already produced an idle-ramp embarrassment.

Archetype: professional-manager restructurers, not founders. For a mature conglomerate unit in transition, that is the right archetype — the value here is portfolio discipline and cost-out, and they are delivering both. The question is whether professional managers can grow (the ESS bet) as well as they can cut.

Lens 10 · Forensic Red Flags

No filings on disk; analysis is web-grounded + the pre-fetched regulatory file. Label every figure.

Accounting / quality risks:

• Revenue/charge lumpiness: the ¥40B defect charge in FY3/25 reversing in FY3/26 mechanically flatters the FY3/26 Energy YoY comparison. Anyone reading "Energy operating profit up" must normalize out the ~¥40B base effect — underlying auto-battery profitability is weaker than the optics.
• Restructuring below the line: ¥174.5B reform costs (FY3/26) are real cash, but framed as one-off "structural reform" — watch that "one-off" doesn't recur annually.
• Capex-ahead-of-revenue: Kansas (~$4B) running below intended ramp = idle fixed-cost absorption depressing unit economics until volume arrives; depreciation runs regardless of utilization.
• IRA credit dependency in the model: the unit's ≥20% EBITDA target is stated ex-IRA credits, which is honest — but the reported economics lean on 45X (~$1B+/yr, up to ~$6.8B cumulative). A subsidy that flows through cost of goods / other income flatters margins; the FY2031 ex-credit target is the figure to trust.
• Segment opacity: no clean standalone Energy P&L in English disclosure → genuine transparency limitation for outside analysis (segments.csv empty). Not fraud — just hard to audit from outside.

Regulatory findings (required sub-section). Per the pre-fetched regulatory/regulatory-findings.md (generated 2026-06-18):

• SEC (EDGAR LR + AAER): zero findings — and not applicable. Panasonic Energy has no CIK; as a Japanese subsidiary it does not file with the SEC, so no EDGAR enforcement search is possible.
• Non-SEC web check ("Panasonic Energy" (FTC OR DOJ OR FDA OR consent decree OR settlement OR fine OR penalty) enforcement): no material US enforcement action surfaced in the deep-dive searches. The most material quasi-regulatory exposures are policy, not enforcement — (a) OBBBA 45X "prohibited foreign entity / material assistance" restriction (tax-years after Jul 4 2025) that could disqualify credits if Chinese-origin material content isn't engineered out; (b) 30D consumer EV credit sunset Sep 30 2025, a demand headwind, not a penalty; (c) Kansas state/local incentive packages contingent on hiring/investment milestones.
• Parent-level note: as part of Panasonic Holdings (TSE Prime), the unit is governed under Japanese securities/J-SOX disclosure — outside the SEC enforcement net analyzed here.
• Verdict: No material regulatory or legal enforcement findings — verified via SEC EDGAR EFTS (LR, AAER, returns zero for a no-CIK subsidiary) and web search as of 2026-06-18. The live risks are policy-driven (45X PFE rules, EV-credit sunset), not enforcement.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Panasonic Energy has no standalone EPS (it's a unit, not a listed share), so the scoreable projection is parent EPS for 6752.T plus the unit-level operating-profit path that actually carries the thesis. All ``, arithmetic shown; no Brier forecast.ts create (breadth/watchlist loop — skipped per instructions).

Parent anchors: FY3/26 net income ¥189.5B (depressed by ¥174.5B reform). FY3/27 group adjusted OP target ¥600B. Shares ~ market cap ¥8.87T ÷ ¥3,800 ≈ 2.33B shares.

Unit operating-profit path (the thesis engine):

• FY3/26 (actual-ish): Energy OP positive, flattered by ¥40B reversal + 46 GWh NA volume.
• FY2029 (mgmt interim): Energy sales ~¥2T, adj OP >¥300B.
• FY2031 (mgmt target): Energy sales >¥3T, EBITDA margin ≥20% ex-credits.

Parent EPS, base / bull / bear (FY3/27 → FY3/29),:

Honest caveat: these EPS lines are `` built on a sourced OP target and an estimated share count; consensus EPS for 6752.T was not cleanly sourceable beyond the FY3/26 actual and the ¥3,608 price target — so treat the table as a structure, not a precision forecast. The variable that matters is whether Energy OP reaches ~¥300B by FY2029; everything else is conglomerate noise.

The single scoreable claim (logged narratively, not via forecast.ts): Panasonic Energy adjusted operating profit ≥ ¥300B in FY2029 — base-case p ≈ 0.45 (it requires the ESS ramp to deliver and autos to stop bleeding; both are contested).

Lens 12 · Bull vs Bear

Bull case. Panasonic is quietly executing the right pivot at the right time. It is abandoning the unwinnable commodity-EV-cell cost war and concentrating on two defensible positions: (1) premium/ex-China auto cells for customers who must de-risk China (a structural, policy-driven tailwind), and (2) data-center backup ESS, where it already holds ~80% share and has pre-sold output into the single hottest demand curve in the economy (AI power). Management is credible on cost (¥174.5B reform, CEO pay cut, exited losing solar) and is putting conviction capital (¥350B) exactly where it has share. The FY2029 unit target (¥2T sales, >¥300B OP) implies a doubling-plus of energy profit. The stock is at a 46x optically-depressed P/E on reform-crushed earnings with the Street at a below-market target — i.e., the turnaround is un-priced. If Energy OP compounds toward ¥300B and group OP hits ¥600B, the normalized P/E is far lower than 46x and the re-rating is the trade.

Bear case (permanent-impairment risks).

1. Auto-cell structural marginalization. Chinese LFP at ~69% share sets the price; premium NCA loses the volume war permanently; Tesla second-sources its 4680. The auto book becomes a low-return, capital-trapped legacy that never earns its cost of capital. This is the impairment risk.
2. The ESS bet is a hope, not yet a business. 80% of a small specialized niche is not the same as a ¥1T business — the targets (¥1T by FY2029) require the niche to explode and Panasonic to hold share against CATL/BYD/Samsung all now charging into AIDC storage. Share in a small pond rarely survives the pond becoming an ocean.
3. Policy whiplash. 30D gone, 45X subject to PFE material-assistance rules; a hostile US policy turn could strand Kansas/Nevada economics.

Pre-mortem (18 months out, thesis broke): Tesla 4680 PO never came at scale; Kansas ran sub-50% utilization through 2027 bleeding depreciation; AIDC storage competition (CATL especially) crushed ESS pricing before Panasonic scaled module output; 45X credits got clipped on Chinese-material content; the "turnaround for growth" became "another year of reform." The stock de-rated to a sub-1x P/S Japanese-conglomerate-discount and the ESS optionality evaporated into a crowded commodity.

Are multiples too high? The 46x P/E is an artifact of depressed E, not a real premium — on normalized earnings the parent is probably cheap-ish for a turnaround. But "cheap conglomerate with a battery unit" is a value trap unless the ESS inflection is real. The multiple isn't the risk; the earnings durability is.

Contrarian view (what the market refuses to see): The Street still prices Panasonic as a tired Tesla-battery supplier in a demand downcycle (consensus below the price). What it is under-weighting is that the most defensible thing Panasonic owns is no longer the Tesla relationship — it's an 80%-share, pre-sold position in AI-data-center backup power. The market is valuing the legacy and ignoring the option. That gap is the thesis — and also the trap if the option doesn't convert.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case. "80% share of data-center backup" is the bull's whole story — and it is 80% of a niche that barely existed yesterday and that every battery giant on earth is now invading. CATL, BYD, and Samsung are explicitly targeting the same $2.2B AIDC storage prize. First-mover share in an exploding category is the most competed-away kind of share. By the time it's a ¥1T market, Panasonic's 80% is a memory.

Where revenue is concentrated / what breaks it. The auto book is still Tesla-shaped, and Tesla has (a) sold-off Panasonic's own alignment (Panasonic dumped the stake), (b) its own 4680, and (c) softening US volume that already forced a Kansas ramp delay and a guidance cut. A single-customer demand air-pocket already happened once in 2025 — it can happen again, and Kansas is a $4B fixed-cost monument to it.

Why the moat is weaker than bulls think. The cylindrical-patent / energy-density edge is real but shrinking — Tesla's own 4680 and Chinese cells are closing the gap, and energy density doesn't win when LFP wins on $/kWh and safety-at-cost. "Ex-China" is a policy moat, and policy is exactly what OBBBA just made fragile (PFE material-assistance, 30D sunset).

Most dangerous competitor bulls underestimate: CATL — not in autos (that war's lost) but in AIDC storage, where CATL's cost structure and scale could undercut Panasonic's premium ESS before it scales, turning the one good moat into another price war.

Worst capital-allocation moves: building Kansas $4B ahead of confirmed sustained demand (idle ramp), and the ¥40B quality defect that signals execution risk in the very premium-quality story the bull case rests on.

What must hold for today's price: that the ESS pivot converts to real, defensible, high-margin revenue and autos stop bleeding and 45X survives. Three independent "ands."

Valuation if growth disappoints 20–30%: if Energy OP lands at ~¥200B (not ¥300B) by FY2029 and ESS sales come in at half-target, the "turnaround" is just a conglomerate muddling at a structurally low ROE — and 6752.T re-rates down toward the Japanese-industrial value-trap multiple, not up. Consensus is already below the price; the asymmetry from here may favor the bear.

Single permanent-impairment scenario (most plausible): commodity-ization of AIDC storage by Chinese scale before Panasonic scales — simultaneously killing the one growth moat and leaving Kansas/Nevada as stranded premium-auto-cell capacity. Plausibility: moderate — it's the consensus-of-the-bears, which is itself a (weak) reason it might be partly priced.

Lens 14 · Management Questions (ordered by information value)

1. Energy unit FY2029 OP of >¥300B — what fraction comes from data-center ESS vs autos, and what ESS market-share assumption underlies the ¥1T sales target as CATL/BYD/Samsung enter? (answer most changes the thesis)
2. On AIDC storage: how much of the FY2029 ¥1T target is already contracted / pre-sold vs pipeline, and at what gross margin vs the auto book?
3. What is the current utilization of Kansas (De Soto), and what confirmed-demand threshold triggers the full 30 GWh (and the 4680 expansion)?
4. 4680: is there a confirmed customer purchase order at scale yet, or is Wakayama still capacity-in-waiting? Which customer, what volume, what year?
5. Under OBBBA's 45X "prohibited foreign entity / material assistance" rule, what share of your US-made cell bill-of-materials is currently China-sourced, and what's the cost to engineer it out to preserve the credit?
6. Quantify the ex-IRA-credit unit economics today — what is auto-cell EBITDA margin excluding 45X, and how far is it from the ≥20% FY2031 target?
7. With the Tesla stake sold, how would you characterize the Tesla relationship's durability and your customer concentration trajectory (Mazda/Subaru ramp timing and volumes)?
8. Solid-state: realistic timeline from pilot to automotive-qualified mass production, and how do you avoid being late vs CATL's 2027 condensed/solid-state roadmap?
9. The ¥40B defect charge — root cause, customer, and what changed in quality systems to prevent recurrence in the premium-quality franchise?
10. Capital allocation: with ¥500B going to AI infrastructure, what is the hurdle rate / expected ROIC on the ¥350B ESS investment, and what kills it?
11. How structural vs one-off is the ¥174.5B reform cost — is FY3/27 genuinely clean, or should we expect further "structural" charges?
12. Silicon-anode (Sila/SiOx): what energy-density and $/kWh does it unlock, by when, and is supply single-sourced?
13. FX/policy sensitivity: how much of the unit's reported profit is yen-weakness and US-subsidy dependent, and what's the normalized picture?
14. Why keep the auto-cell business at all if it can't win on cost — what's the strategic rationale vs harvesting it for cash and concentrating on ESS?
15. What is the one competitive development (a CATL AIDC product, a Tesla in-housing move) that would most force a strategy rethink?

---