4062 / 4062

4062 — Ibiden Co., Ltd.

WIKI stockadvanced-packagingpassives-mlcc updated 2026-08-29

🔴 Live ¥20,475 (yf:4062.T) · 2026-09-05 · research written 2026-05-15 Consensus ¥10,500–¥30,000 (mean ¥24,850, 12mo, Yahoo consensus (16 analysts)) · 2026-09-05 your re-engage <¥13,000–¥14,000 (own-DD, on-pullback) — +46% away

Thesis

Verdict: WATCH, not BUY (as of 2026-05-15 refresh). Ibiden is the global #1 maker of high-end ABF / FCBGA IC substrate at the precise moment AI accelerator package areas crossed organic substrate's prior physical comfort zone — exactly the regime where Ibiden's process moat is structurally widest. The business quality is real and the trajectory is consensus-backed; what is contested is the multiple. The stock trades at 74.6x TTM / 52.0x forward P/E, and even after a 3% pullback the base-case DCF target (JPY 14,350) sits ~11% below spot. This is a quality franchise priced for perfection, not an asymmetric entry.

What has to be true for the bull case. (1) The substrate franchise keeps compounding through the Vera Rubin / Rubin Ultra cycle, with Ibiden defending flagship share against Unimicron; (2) the glass-core transition (Intel GlassCore, 2027-2030 commercial) becomes a bridge Ibiden builds rather than one built around it; (3) sell-side targets re-rate up to the real JPY 14-18k consensus range, removing the structural overhang of stale screener marks.

Why WATCH and not BUY at JPY 16,065. The FundamentEdge hard rules split: revenue-growth primacy PASSES (FY3/26 +34%, FY3/27E +18%), second-derivative PASSES (accelerating), estimate-revision direction PASSES (upward, with the caveat sell-side has been slow), but valuation-as-thesis FAILS (rich multiple, no margin of safety) and quality-as-risk is MIXED (oligopolistic, but the Unimicron share-loss dynamic is genuine). The asymmetry favors waiting. STF Research's coverage reinforces this independently: they treat Ibiden as the consensus center of the AI substrate value chain but route their actual stock picks (Uyemura, MEC, Taiyo, Kitagawa Seiki, Maruwa) around Ibiden — the implicit message that Ibiden's quality is real but the price has caught up and the alpha now sits in the under-covered chemistry / equipment names that feed it.

Conviction: Medium — high quality of business plus high quality of trajectory, but valuation has not reset enough to make this an entry-priced compounder. Flip to BUY on a pullback to JPY 13-14k (mid-30s forward P/E), a sell-side re-rate above JPY 17k, a Rubin / Rubin Ultra sourcing confirmation, or an FY3/27 Q1 beat with margin expansion.

Snapshot

Ibiden Co., Ltd. — ticker 4062.T, Tokyo Stock Exchange Prime. Global #1 ABF / FCBGA IC substrate maker; Intel server CPU + Nvidia data-centre GPU (Hopper / Blackwell / Vera Rubin) + AMD + hyperscaler substrate franchise. Oono (Fukui) plant phase 2 commissioning; glass-core transition leader. Founded 1912 as Ibigawa Electric, renamed Ibiden 1982. HQ Ogaki, Gifu Prefecture; ~14,000 group employees.

Valuation snapshot (2026-05-15 close)

Metric Value Δ vs 2026-05-12
Spot JPY 16,065 –2.9% (from JPY 16,550)
Market cap JPY 4.49 tn (USD ~29.9 bn at JPY 150/USD) –2.8% (from JPY 4.62 tn)
Enterprise value JPY 4.64 tn +8.4% vs JPY 4.28 tn (net cash drawdown on capex)
P/E (TTM) 74.6x down from ~77x
P/E (Fwd, FY3/27E) 52.0x down from ~54x
EV/EBITDA (TTM) ~26-28x –1 turn
EV/Revenue (TTM) ~9-10x unchanged
52-week range JPY 2,477 – JPY 18,365 new ATH JPY 18,365 printed May 12-14
Dividend yield 0.24% unchanged (token, capex-prioritised)
Beta (5Y, JPY) 1.56 unchanged
Shares out 279.2M unchanged
Yfinance consensus PT JPY 10,441 STALE — do not anchor

Stock printed a new 52-week high of JPY 18,365 in the May 12-14 window before fading to JPY 16,065 — a ~12.5% drawdown in two sessions on no specific news, read as mechanical / positioning rather than fundamental.

Business

Three segments. Electronics ~70-75% (the IC-substrate franchise), Ceramics ~20-25% (diesel particulate filters / DPF), Others ~3-5% (includes a specialty graphite sub-line). Latest IR materials are the FY3/26 Q4 / full-year deck released late April 2026 (https://www.ibiden.com/ir/library/).

The moat — substrate process leadership

Ibiden's edge is in high-end ABF / FCBGA substrate process: warpage management at large panels, the multi-build-up dielectric ladder, embedded passives, and panel-level yield compounding. The thesis turns on package-area physics: AI accelerator package area crossed 5,625mm² (B200) → 8,100mm² (Vera Rubin), and B300 NVL16 pushes beyond the JEDEC 120×120mm package size envelope — the regime where Ibiden's process moat (low warpage at large panels, build-up cycle yield) is structurally widest. Glass-core substrate is the next bridge, and Ibiden is positioned as Intel's lead glass-core development partner.

Customers

Top customers: Nvidia (~15-25%; Hopper sole-supplier / Blackwell majority / Vera Rubin qualified), Intel (~20-25%; server CPU multi-decade relationship), AMD (~10-15%), hyperscalers AWS / GOOGL / MSFT / Meta combined ~10-15%, with Apple / Marvell / Broadcom / MediaTek as a ~5-10% long tail. Diesel DPF automotive ~20% (Cummins, Daimler Truck, Stellantis). Concentration flag: Nvidia + Intel combined approach ~50% of electronics revenue and ~35-40% of group revenue — real single-customer risk, partly offset by the Unimicron dual-sourcing dynamic (which protects against Nvidia squeezing prices via take-or-pay leverage, since Ibiden can substitute Nvidia volume into Rubin-tier supply at richer pricing).

Competitive position

Top-5 organic ABF substrate makers globally: Ibiden, Unimicron, Shinko Electric (renamed Resonatec post-Toshiba spin), Nan Ya PCB, Kinsus — with AT&S the European challenger, mainly Intel-aligned. Ibiden remains #1 by AI / data-center mix and process moat. Unimicron is the credible #2 and has been the share-taker at Blackwell GPU substrate (per SemiAnalysis 2024). Industry structure is a consolidated tier-1 oligopoly with 4-5 players capable of building flagship-tier substrate.

Value chain (upstream nodes)

Key upstream suppliers feeding the substrate: Ajinomoto (ABF resin), Mitsui Kinzoku (MicroThin / HVLP Cu foil), MEC (CZ adhesion chemistry), plus Atotech and Uyemura on plating / surface treatment. Per STF Research, Uyemura's ENEPIG final-pad surface-treatment share is ~50-55% in Japan (Ibiden, Shinko Electric) vs ~70% in Taiwan — implying the Japan substrate ecosystem is more chemistry-source-diversified than Taiwan, marginally reducing single-source upstream risk.

Financials

Income statement & margins (consensus from 17-broker average per QUICK / Bloomberg compile, late April; spread is wide):

Metric FY3/24 (FY-2) FY3/25 (FY-1) FY3/26 (FY0, latest) FY3/27E (FY+1E)
Revenue (JPY B) 369 369 ~495 (+34%) ~585 (+18%)
Operating margin % ~14 ~14 ~17-18 ~19-21
Net income (JPY B) ~35 33.7 ~58 ~80
EPS (JPY) ~125 121 ~209 ~287
FCF (JPY B) negative –79.6 turns positive 2H FY3/26 clearly positive

EPS roughly doubles FY3/25 → FY3/27E (JPY 121 → ~JPY 287). Latest reported quarter is FY3/26 Q4, disclosed late April 2026. No new earnings between May 12 and May 15.

Balance sheet & cash flow

Net cash; capex self-funded (cash on hand + operating cash flow + bank facility). FCF was negative in FY3/24 and FY3/25 (–79.6B) on the Oono build, turns positive in 2H FY3/26, and is clearly positive in FY3/27E. FCF yield is currently negative — Oono phase 2 capex still flowing; P/FCF is n/a on negative LTM FCF. The Oono plant (Fukui) carries a JPY 250-300B capex commitment, with phase 1 operating since early 2024 and phase 2 commissioning across 2026-2027.

Note on EV. The enterprise value move from JPY 4.28tn → JPY 4.64tn at a lower market cap is consistent with a net cash drawdown of ~JPY 350-400B over the 3-day window — possibly a capex tranche payment to Oono phase 2 contractors or a bank-facility drawdown for working capital. Flagged for the next earnings reconciliation, not actionable today.

Dilution

Dilution risk: low. No ATM, no shelf, no convertible overhang. The capex is self-funded. Incremental-margin analysis over the prior 8 quarters shows operating leverage building at the Electronics-segment level.

Industry landscape

High-end ABF / FCBGA IC substrate is a consolidated tier-1 oligopoly — Ibiden, Unimicron, Resonatec (ex-Shinko Electric), Nan Ya PCB, Kinsus, with AT&S as the European / Intel-aligned challenger. Demand is inflecting on AI accelerator package-area growth (5,625mm² → 8,100mm² → JEDEC-exceeding); supply is tight; Unimicron is taking structural share at Blackwell even as absolute volumes grow for everyone. The next architectural shift is glass-core substrate (2026-2030 depending on source). Upstream bottleneck nodes — Ajinomoto (ABF resin), Mitsui Kinzoku (HVLP Cu foil), MEC, Uyemura — are themselves under-covered beneficiaries.

See sector page: advanced-packaging (ABF/FC-BGA IC-substrate; also a major MLCC peer — see passives-mlcc)

Management

Leadership. President / CEO Koji Kawashima (since June 2017; prior CTO / COO; substrate-engineering lifer with a Toyohashi University of Technology PhD background). CFO is listed as Takeshi Aoyama in the profile fragment and as Yoshitaka Hidaka in the mgmt-dd fragment — DISCREPANCY between source fragments; both names retained pending verification against the FY3/26 securities report. Chairman Takeshi Ozawa (transitioned from the CEO role). Other named executives: Kazuhiro Sakai (EVP) and Hiroshi Nakamura (Electronics segment lead). Not founder-led — professional, internally-promoted management with multi-decade Ibiden tenure (typical Japanese corporate model).

Ownership & alignment

Insider ownership low — combined directors + ESOP <2% of outstanding, typical for a Japanese listed. Top holders (Japan structure): Master Trust Bank of Japan ~12-15%, Custody Bank of Japan ~5-8%, Toyota / Toyota-affiliated entities (small), foreign institutionals (BlackRock, Vanguard, JPM, Fidelity via passive + active sleeves) combined ~20-25%, self-held / ESOP ~2-3%. No 13D-style activist position disclosed; no recent material insider transactions in EDINET.

Compensation

Compensation is heavily cash (base + cash bonus) per the FY3/25 securities report; stock-based comp is modest (~JPY 100-300M per NEO range); no golden-parachute structures. METI / TSE governance reform is pushing TSE Prime listeds toward equity-linked compensation, but this is not yet substantive at Ibiden. Verdict on alignment: aligned-by-default given career concentration, weakly aligned-by-design given the cash-comp emphasis.

Capital allocation & governance

Capital allocation graded A−: the Oono capex is a clear, bet-the-house capacity expansion at the right point in the cycle; no M&A history of concern; modest annual buybacks at TSE-typical rates; no major equity issuance in the last decade. Guidance tendency is conservative-to-straight-shooter, with multiple FY3/25 → FY3/26 in-line beats and no weasel-language pattern. Board: ~11 directors, ~4 outside (independent) — below US best-practice on independence ratio but compliant with the revised Japan Corporate Governance Code; board stable, no activist involvement. Overall mgmt grade: A− / Green-with-yellow-on-disclosure-granularity — the yellow is structural (Japanese listed disclosure regime, no Form-4-equivalent same-day insider filings, no 10b5-1-equivalent plans), not Ibiden-specific. A team to trust with capital; the open question is the price.

Next governance reads

FY3/26 Annual Securities Report (有価証券報告書) expected June 2026 — the next material governance read (updated NEO comp tables, related-party summary, beneficial-owner list, board attendance, audit-firm tenure). FY3/26 AGM late June 2026 — watch for any activist / minority push on capital allocation, equity-linked comp, or board independence.

Catalysts & risks

Catalysts (bull)

Near-term (0-90 days): FY3/27 Q1 earnings (late July / early Aug 2026) — first print under Oono phase 2 partial commissioning; watch Electronics-segment operating-margin trajectory, any Vera Rubin substrate ASP / volume guidance, and Unimicron co-supply commentary. This is the next material datapoint. Nvidia GTC Fall 2026 (late October) — Rubin Ultra detail; SA's 2025 GTC piece flagged "beyond JEDEC 120mm" package sizes, and confirmation here is bullish on Ibiden's per-package dollar content. Intel earnings late July — Granite Rapids volume + GlassCore qualification timeline.

Medium-term (3-18 months): Vera Rubin / Rubin Ultra substrate sourcing confirmation (Q3/4 2026 watch); Oono phase 2 commissioning milestones (multiple announcements through 2H 2026 / 1H 2027); hyperscaler custom-silicon ramp (Trainium 2, TPU v6/v7, MAIA 200, Axion 2); METI advanced-packaging subsidy updates; glass-core Intel qualification gating. No new contract awards or capacity commitments disclosed in the 72-hour window.

Risks (bear)

  1. Valuation / consensus-ahead risk — THE NUMBER ONE RISK. Stock is 12.5% off ATH on no news; the marginal buyer is becoming sensitive to multiple. Until sell-side actually re-rates targets to the real JPY 14-18k range, the stock is structurally exposed to multiple compression on any minor disappointment. Likelihood Medium-High in 3-6 months. Not closeable except through sell-side update cycles and continued execution.
  2. Dual-source structural acceleration to Unimicron. The single most important fundamental risk. Mitsui Kinzoku (5706) primer flagged Unimicron and AT&S pulling MicroThin Cu foil volume that previously went exclusively to Ibiden, weakening Ibiden's de facto exclusivity at flagship Cu-foil specs. Likelihood Medium-High over 12-24 months. Partially closeable — Ibiden can defend at glass-core / next-gen but cannot reverse organic ABF share loss.
  3. Capex overrun / Oono phase 2 commissioning execution slip.
  4. JPY / USD FX — Ibiden books in JPY, sells in USD-tied pricing; a strong yen compresses reported margins.
  5. Glass-core substrate transition execution — leadership on organic ABF does not automatically translate. Discrepancy / open risk: the base deep-dive assumed Intel GlassCore commercial 2027-2030 with Ibiden as the bridge; STF Research's January-2026 datapoint (Absolics — the SKC / Applied Materials JV — shipping volume glass-core samples to AMD for MI400, commercial plausibly 2026-2027) suggests glass-core may commercialise 12-24 months earlier and via AMD-through-Absolics rather than Intel-through-Ibiden. If it proves out, Ibiden's "bridge to glass core" optionality weakens materially. Watch AMD MI400 substrate sourcing announcements in 2H 2026.

Dilution risk: low (no ATM/shelf/convertible). Key-person risk: unchanged — Kawashima tenure secure, no succession event flagged.

Behavioral traps: watch for FOMO (most run-up Japanese semi name, easy to chase) and narrative seduction (the AI substrate story is genuinely good but the price has caught up).

Valuation / DCF

Three-scenario forward P/E framework on FY3/27E EPS of JPY 287:

Scenario Forward P/E Target vs spot JPY 16,065
Bear 35x JPY 10,050 ~–37%
Base 50x JPY 14,350 ~–11%
Bull 60x JPY 17,200 ~+7%

At spot JPY 16,065 the stock sits between base and bull but closer to bull — risk-reward is asymmetric to the downside on multiple-compression risk. Base-case expected return over a 12-18 month horizon is ~–11%; the 3-day pullback closed only ~2pts of the gap (from –13% at the 2026-05-12 write). The DCF does not work at the current multiple — this is not a margin-of-safety play.

Deeper bear scenario. If the multiple compresses to the peer mid-cycle high-20s / low-30s forward P/E, the stock prints JPY 8,000-10,000 (–40% to –50%). Trigger: Unimicron displaces Ibiden as primary Rubin Ultra source, OR Oono phase 2 slips ≥6 months, OR JPY strengthens to JPY 140/USD.

On the sell-side consensus. The yfinance mean PT of JPY 10,441 is STALE / un-rerated sell-side data, not a real consensus mark — it reflects analysts who have not caught up with the AI substrate re-rating after the ~6.7x TTM run. Bloomberg / QUICK house-view consensus circulates JPY 14-18k. Do not anchor on the JPY 10.4k figure for any sizing or signal work; verify against the real consensus number on the next earnings cycle. A re-rate of the sell-side mean above JPY 12k would signal the catch-up is underway; above JPY 17k would remove a structural overhang and is a BUY trigger.

Position sizing. Conviction Medium; suggested action watchlist, not buy. If forced to enter today, 0.5-1.0% portfolio weight, scaling strictly only on dips to the JPY 13-14k range (mid-30s forward P/E).

Decision log

  • 2026-05-12 — Base research cycle: full profile, deep-dive (~58k words across 17 sections), mgmt-dd, and pre-buy checklist. Checklist verdict: WATCH (quality business, consensus-ahead valuation, wait for better entry). Spot JPY 16,550; base-case target JPY 14,350 implied ~–13%. Mgmt grade A−.
  • 2026-05-15 — 3-day delta refresh across all four research files (part of the Japanese-semi GAA-ramp swarm). Spot JPY 16,065 (–2.9%); 52-week high reset to JPY 18,365 (printed May 12-14), stock ~12.5% off ATH on no news. TTM P/E recompressed 77x → 74.6x; fwd P/E ~54x → 52.0x. Base-case target gap improved to ~–11%. Verdict unchanged: WATCH (still not BUY). No new IR disclosure, no insider EDINET filing, no SA mirror post, no governance change in the 72-hour window. Mgmt grade A− carries.
  • 2026-05-15 — flags raised for next cycle: (a) yfinance mean PT JPY 10,441 is stale, do not anchor; (b) STF Research Absolics→AMD MI400 glass-core datapoint suggests faster glass-core timeline (2026-2027) via a non-Ibiden path — track AMD MI400 substrate sourcing; (c) EV jumped JPY 4.28→4.64tn on a ~JPY 350-400B net-cash drawdown, reconcile at next earnings; (d) re-fetch STF "Substrate Wars" and "Kitagawa Seiki" posts (fetch failed this session).

Triggers to flip to BUY

  1. Pullback to JPY 13-14k (mid-30s fwd P/E) — resets multiple to defendable level.
  2. Sell-side consensus target re-rates above JPY 17k — catch-up done, overhang removed.
  3. Rubin / Rubin Ultra sourcing confirmation (likely late Q3 2026) — gates dual-source risk.
  4. FY3/27 Q1 earnings beat with operating-margin expansion (late July / early Aug 2026).

Triggers to flip to PASS (permanently uninteresting)

  1. Unimicron confirmed as primary Rubin Ultra source.
  2. Oono phase 2 commissioning slips ≥6 months.
  3. Multiple stays above 50x fwd P/E through FY3/27 Q2 — the consensus-ahead problem becomes chronic.

Next material datapoint: FY3/27 Q1 earnings, late July / early August 2026 (Japanese fiscal calendar). Calendar reminder worth setting; also FY3/26 Annual Securities Report (June 2026) for the next governance forensic read.

Sources

Consolidated from the following vault research fragments (all 2026-05-15 three-day delta refreshes pointing back to a fuller 2026-05-12 base cycle):

  • 4062-profile (2026-05-15 profile)
  • 4062-buy-checklist (2026-05-15 pre-buy checklist, verdict WATCH)
  • 4062-mgmt-dd (2026-05-15 management DD, grade A−)
  • 4062-deep-dive (2026-05-15 deep dive, Section A delta refresh)

External / cross-referenced sources cited within the fragments:

  • Yfinance (2026-05-15 quote, valuation multiples, beta, mean PT)
  • Ibiden IR page (https://www.ibiden.com/ir/library/); TDnet / EDINET — no fresh disclosure in window
  • Consensus from 17-broker average per QUICK / Bloomberg compile (Nomura, Daiwa, SMBC Nikko, Mizuho, Mitsubishi UFJ Morgan Stanley, Goldman, Morgan Stanley, JPM, Citi, Jefferies, CLSA, Macquarie, BofA, Bernstein, Tokai Tokyo)
  • SemiAnalysis mirror at ~/Dropbox/Wafflebun/KB/wiki/semianalysis/ — no May 2026 posts on 4062 (prior framing: sole-supplier 2023 → Unimicron share-loss 2024 → JEDEC-exceeding package size 2025)
  • STF Research (@stfbutnou, paid sub): "Uyemura: Another Invisible Beneficiary Behind Every Substrate" (May 11, 2026, fully read); "The Substrate Wars: A Reality Check + Investor's Playbook" (fetch failed, re-try); "AI Supply Chain: Shortage Intensifying Toward 2027" (Apr 7, 2026, Founding-Member tier, locked); "Kitagawa Seiki: Hidden Champion" (fetch failed); "Finding the Last Cheap AI Stock in the Japanese Rally" (Mar 2026, referenced — STF did NOT pick Ibiden)
  • Related primers: packaging-glass-substrate-primer (2026-05-03), cu-wiring-resin-sub-layer-primer (2026-05-12), ai-server-pcb-primer (2026-05-03)
  • Prior canonical page 4062 (created 2026-05-12, updated 2026-05-15)

Consolidation queue (merged 2026-05-30)

The following research fragments were folded into this canonical page on 2026-05-30. They stay live pending Pink's archive confirmation.

  • [ ] 4062-profile.md
  • [ ] 4062-buy-checklist.md
  • [ ] 4062-mgmt-dd.md
  • [ ] 4062-deep-dive.md
  • [ ] 4062.md

Source updates (auto-maintained)

Drop/Memory (Jun 1, 26) - GS Memory

Goldman Sachs's 31 May 2026 memory note argues the DRAM/NAND/HBM upcycle extends into 2028 on AI-server demand and constrained supply, projecting HBM TAM of $56bn/$116bn/$168bn for 2026E/2027E/2028E; Ibiden is not directly covered in the visible portion of the report.

Relevant to your thesis: A sustained multi-year HBM cycle supports AI accelerator volume that drives Ibiden's substrate demand, but does not address the valuation overhang — the watch-not-buy stance is unchanged.

Source: dropfile://Memory/GS Memory.pdf

Drop/Google Infra (Dec 30, 25) - Inside Google’s Ironwood TPU v7 Supply Chain - by Phabian

Ibiden is named as one of the core ABF substrate vendors gating AI accelerator production across the industry, explicitly listed as a direct beneficiary of each incremental TPU v7 wafer alongside Unimicron, Kinsus, Nan Ya PCB, and Shinko.

Relevant to your thesis: Confirms Ibiden's substrate franchise extends to Google's Ironwood TPU supply chain, reinforcing the "multiple AI customer" moat argument and partially offsetting the Nvidia/Intel concentration flag in the wiki.

Source: dropfile://Google Infra/TPUs/Inside Google’s Ironwood TPU v7 Supply Chain - by Phabian.pdf

Drop/GPUs (Dec 21, 25) - SA_ Deep Dive_GB200 Hardware Architecture - Component Supply...

The SA GB200 supply-chain deep-dive includes a dedicated "Substrate, PCB, and CCL" section (content truncated in excerpt); the visible portions confirm GB200 deploys across four rack-scale form factors (NVL72, NVL36x2, Ariel, Miranda) at volume covering the top 50 hyperscale buyers.

Relevant to your thesis: Confirms the Blackwell deployment breadth driving Ibiden's FY3/26 revenue acceleration; the specific Ibiden vs. Unimicron substrate-share split sits in the truncated substrate section.

Source: dropfile://GPUs/SA_ Deep Dive_GB200 Hardware Architecture - Component Supply Chain & BOM.pdf

Drop/2. Frontend (Dec 30, 25) - Inside Google’s Ironwood TPU v7 Supply Chain - by Phabian

Ibiden is listed among the ABF substrate vendors — alongside Unimicron, Kinsus, Nan Ya PCB, and Shinko — confirmed as feeding Ironwood TPU v7's MCM packages through TSMC and the major OSATs, with the article noting this tier directly translates each incremental TPU wafer into utilization and pricing power for the ABF ecosystem.

Relevant to your thesis: Corroborates the demand-side view that Ibiden's substrate moat spans beyond Nvidia/Intel into Google's TPU stack, reinforcing the revenue-growth and second-derivative PASS legs while leaving the valuation-as-thesis FAIL unchanged.

Source: dropfile://2. Frontend/Google Infra/TPUs/Inside Google’s Ironwood TPU v7 Supply Chain - by Phabian.pdf

Drop/2. Frontend (Dec 21, 25) - SA_ Deep Dive_GB200 Hardware Architecture - Component Supply...

SemiAnalysis's full GB200 BOM breakdown covers Blackwell rack architectures (NVL72 at 120 kW, NVL36×2 at 132 kW) across 50+ subcomponents including a dedicated "Substrate, PCB, and CCL" section where Ibiden is named; the substrate-specific passages fall in the truncated portion of the article.

Relevant to your thesis: The rack-scale architecture detail — 36–72 B200 GPUs per NVLink domain at extreme power densities — corroborates the package-area inflection argument (Hopper → Blackwell → Vera Rubin) that anchors the substrate-moat bull case.

Source: dropfile://2. Frontend/GPUs/SA_ Deep Dive_GB200 Hardware Architecture - Component Supply Chain & BOM.pdf

Vik's Newsletter (May 6, 26) - Vik's Newsletter

TSMC named Ibiden as one of two partners in its first CoPoS glass validation data, alongside Innolux on the glass panel side, confirming Ibiden's role in TSMC's glass-core substrate program.

Relevant to your thesis: Confirms the wiki's claim that Ibiden is positioned as a glass-core development partner, though the TSMC relationship adds a dimension beyond the Intel-focused framing — Ibiden is inside both glass programs simultaneously.

Source: https://www.viksnewsletter.com/p/battle-for-advanced-packaging-tsmc-intel-amkor

Drop/Bottleneck (Jun 17, 26) - 0 - JPM 150626 - MLCC substrates

JPM's June 2026 substrate deck places Ibiden's capacity at 82/95/107 (normalised units, 2025/2026e/2027e) within an industry it characterises as "entering unprecedented supply tightness," driven by AI chip body-size migration and CoWoS interposer-area expansion that JPM maps explicitly against substrate layer-count growth.

Relevant to your thesis: Supply tightness and confirmed capacity ramp directly support the bull case that Ibiden's process moat widens as package areas grow — the watch condition remains valuation, not demand visibility.

Source: dropfile://Bottleneck/MLCC/0 - JPM 150626 - MLCC substrates.pdf

Drop/2. Frontend (Jun 1, 26) - GS Memory

Goldman Sachs (31 May 2026) raises HBM/DRAM/NAND price forecasts and argues the memory upcycle extends into 2028, driven by AI server demand, supply discipline, and LTAs — with HBM TAM reaching USD 168bn by 2028E; Ibiden is mentioned only tangentially as part of the AI supply chain context, not covered directly.

Relevant to your thesis: Structurally bullish for Ibiden's substrate demand runway (Blackwell/Rubin cycle sustained longer), reinforcing the revenue-growth PASS but not addressing the valuation-multiple FAIL that keeps the verdict at WATCH.

Source: dropfile://2. Frontend/Memory/GS Memory.pdf

Intake (May 12, 26) - cu-wiring-resin-primer

Ibiden (4062) is the substrate fabricator that consumes MicroThin Cu foil and ABF resin — it owns the process integration and customer qualification slot, while the chemistry moat sits upstream with Mitsui Kinzoku, MEC, and Ajinomoto.

Relevant to your thesis: Reinforces the existing wiki's upstream-supplier framing (Uyemura, MEC, Taiyo as the alpha-rich plays) and the STF Research read that Ibiden's quality is real but the value capture is split with chemistry suppliers.

Source: intakefile://cu-wiring-resin-primer.md

from the vault · open in obsidian ↗