Japan Electronic Materials (TSE: 6855)

---

Identity

Item	Detail
Full Legal Name	Japan Electronic Materials Corporation (日本電子材料株式会社)
Ticker	6855.T (TSE Standard Market)
Sector	Semiconductor Equipment — Test & Measurement
Headquarters	2-5-13 Nishinagasu-cho, Amagasaki, Hyogo 660-0805, Japan
Founded	April 1960
Probe Cards Since	1970 (tech partnership with Rucker & Kolls, US)
Listed	TSE First Section (2006); transitioned to TSE Standard Market (2022)
FY End	March 31
Employees	~1,154 (consolidated)
Subsidiaries	10 (Japan, US, Taiwan, Hong Kong, China x2, France, Korea, Thailand, Singapore)
Market Cap	~¥103B (~$690M) pre-dilution; ~¥122B post-Feb 2026 PO
Enterprise Value	~¥100B (~$670M)
Website	jem-net.co.jp

JEM is the fifth-largest probe card manufacturer in the world, the #1 in Japan by its own account, and arguably the purest publicly traded bet on HBM testing demand.

---

Thesis

JEM is a pure-play probe card manufacturer riding the explosive HBM demand cycle driven by AI infrastructure buildout. As memory makers (Samsung, SK Hynix, Micron) ramp HBM3/HBM4 production, every HBM die must be individually tested before stacking — the "Known Good Die" requirement — and each test burns through probe cards with 150,000+ microscopic needles. JEM's MEMS M-type probe cards are purpose-built for this.

Revenue is recovering sharply from the FY2024 memory trough, with record profits expected in FY2026 and a new factory under construction to meet demand. Q3 FY2026 results confirmed the thesis: +40% revenue, +78% operating profit year-over-year, and an upward earnings revision.

The bear case is real: the stock has already rallied ~578% from its April 2025 low, management just diluted shareholders by 15.8% through an equity raise they arguably didn't need (sitting on ¥7B+ net cash), and the memory cycle is notoriously boom-bust. JEM is also a small player (~5-7% global share) competing against much larger rivals with deeper R&D budgets.

Metric	Value
Current Price	~¥8,490 (Mar 2026)
Target Price	¥8,300-10,400 (fair value, 20-25x FY2027E EPS of ¥417)
Bull Case	¥12,500+ if HBM cycle extends and new factory ramps on schedule
Bear Case	¥3,000-4,000 if memory cycle turns (as in FY2024)
Conviction	Medium — compelling HBM thesis with forward earnings support, but dilution, cycle risk, and Korean competition warrant caution

Single most important thing that must go right: HBM production volumes must continue to grow. JEM's fortunes are directly tied to memory semiconductor capex and production ramp.

What would invalidate the thesis:

• HBM production growth decelerating below 20% year-over-year
• A major customer publicly switching to Korean competitors for next-gen HBM probe cards
• New factory delays or cost overruns consuming the equity raise proceeds
• Operating margins failing to reach 23%+ as guided for FY2026

---

Business

What This Company Actually Does

JEM makes probe cards — precision circuit boards studded with thousands of microscopic needles that press against semiconductor chips while they're still on the silicon wafer to test whether each chip works before it gets cut and packaged. Think of it as the quality-control stamp at the end of the chip assembly line, except the stamp has 150,000 hair-thin metal pins and costs up to $1 million. The needles wear out after millions of touches, so customers keep reordering — making probe cards a consumable, not a one-time capital purchase.

This is essentially a single-product company. Everything that matters is probe cards.

Segment	Revenue Share	What It Does
Semiconductor Inspection Components	~99%	Probe cards (cantilever, vertical, MEMS), cleaning sheets, package test adapters
Electron Tube Components	~1%	Legacy: CRT heaters, filaments, heat-resistant metal processing (tungsten, molybdenum)

The Consumable Business Model

Probe cards are consumables, not capital equipment. This is the single most important thing to understand about JEM's business model:

1. Each new chip design requires a custom probe card (non-recurring engineering + production units)
2. Cards wear out after millions of touchdowns — needles deform, contact resistance drifts, contamination builds up
3. HBM burn-in testing accelerates wear dramatically — aggressive Known Good Die testing physically degrades MEMS needles faster
4. Technology transitions (HBM3 → HBM3E → HBM4) require entirely new probe card designs

This creates a recurring revenue dynamic that compounds with both wafer volume AND design complexity. As HBM stacks get taller (8-high → 12-high → 16-high) and pin counts increase, the dollar content of probe cards per wafer grows — even if wafer starts stay flat.

Revenue is project-based and somewhat lumpy — a new chip design qualification drives a burst of orders, then follow-on replacement cards as the probes wear out. Not a SaaS-style predictable revenue stream, but the repeat-purchase dynamic is real.

Probe Card Technology Types

JEM manufactures three main types, with MEMS being the growth driver:

Cantilever Probe Cards (legacy): The oldest technology. Fine metal needles extend horizontally like wheel spokes, contacting pads at an angle. Simple, cheap, and reliable for large-pad, low-pin-count applications (analog, driver ICs, legacy logic). JEM's historical strength — but this market is mature and shrinking. JEM's CE series targets automotive-grade power devices where test requirements are stricter than consumer.

Vertical Probe Cards (VT/VS Series): Needles mount perpendicular to the wafer surface, enabling denser arrays, smaller probe marks, and less pad damage. A step up from cantilever for mainstream memory and logic testing. VE series hybrid cards include light source openings for CMOS image sensor testing (Sony, Samsung).

MEMS Probe Cards (M-Type) — the growth engine: Uses semiconductor lithographic processes to fabricate micro-springs with sub-micron precision. MEMS probes deliver the best planarity (all needles at exactly the same height — critical when 150,000 needles must touch simultaneously), highest pin counts, longest lifetimes (10 million+ touchdowns vs. 500K for cantilever), and lowest contact resistance variation. MEMS is the required technology for HBM testing.

JEM's MEMS revenue mix has increased steadily: ~50% of revenue in 2019, >60% in 2020, and likely 70-80%+ in FY2026. This mix shift is the primary margin expansion driver — MEMS cards cost 3-4x more than legacy cantilever cards but also carry higher margins.

Why HBM Changes Everything for Probe Cards

A standard DRAM chip has roughly 40,000 test pads. An HBM die has 150,000+ pads because of the through-silicon vias (TSVs) that connect stacked layers. Building a probe card with 150,000 needles that all make simultaneous contact with 45-55 micrometer pitch (barely wider than a human hair) is an extraordinary engineering challenge:

1. Planarity: All 150,000 probes must be within ±12.5 micrometers of the same plane. Any needle that's too high or too low won't make proper contact — and at 45μm pitch, there's almost no room for error.
2. Thermal management: HBM stacks generate concentrated heat during burn-in testing. Different thermal expansion rates between the probe card PCB and the silicon wafer cause alignment drift at elevated temperatures.
3. Signal integrity: Testing at multi-GHz speeds across 150,000 channels requires controlled impedance, minimal crosstalk, and low insertion loss. The probe card isn't just a mechanical contact — it's a high-frequency electrical interconnect.
4. Wear: Aggressive burn-in testing for KGD physically degrades the MEMS needles faster than normal testing. This accelerates the replacement cycle — great for recurring revenue, but the probes must be engineered for maximum durability.

The net result: HBM probe cards cost 3-4x more per card than standard DRAM cards, wear out faster, and are needed in growing quantities as HBM production scales.

Metric	Standard DRAM	HBM	Why It Matters
Pin count	~40,000	150,000+	3.75x more needles = much higher ASP
Pitch	60-80 μm	45-55 μm	Tighter pitch requires MEMS precision
Card cost	$200-300K	$500K-1M+	Direct revenue uplift per card
Touchdown lifetime	5-10M (MEMS)	Lower (aggressive burn-in)	Faster replacement = more recurring revenue
Planarity	±25 μm	±12.5 μm	Tighter spec = higher engineering content

How the Test Flow Works

ATE (Tester)  →  Test Head  →  Probe Card  →  Wafer (on prober chuck)
 (Advantest,      (electrical     (PCB with       (vacuum-mounted,
  Teradyne)        connection)     1000s of        aligned by camera)
                                   needles)

The wafer prober (made by Tokyo Electron or Tokyo Seimitsu) vacuum-mounts the wafer on a precision stage, uses cameras to optically align the probe card's needle tips with the die's bond pads, then raises the stage to bring the wafer into contact with the probes. Electrical signals flow through the probes to the ATE (Advantest or Teradyne), which runs test programs and records pass/fail results for each die.

For a full-wafer probe card (used in memory testing), all dies on the wafer are contacted simultaneously in a single touchdown — requiring 50,000-150,000+ probes to make perfect contact at the same time.

Product Segments in Detail

Segment A: MEMS Probe Cards (M-Type) — ~70-80% of revenue (estimated)

JEM's M-type MEMS probe cards are the growth engine. Specifically designed for memory testing — DRAM, NAND, and critically, HBM. The MEMS fabrication process uses semiconductor lithography to create micro-spring probe structures with sub-micron dimensional control, enabling the ultra-fine pitch and massive pin counts HBM requires.

• Customers: SK Hynix, Samsung, Micron (inferred — JEM doesn't disclose names, but IRBank data shows Micron entities account for ~70% of major disclosed customer revenue)
• Growth: Driving the +40% year-over-year revenue growth in 9M FY2026
• ASP trend: Rising as HBM share of memory testing increases
• Competitive position: Strong in Japan and gaining share in Korea/Taiwan through HBM demand

Segment B: Cantilever & Vertical Probe Cards — ~20-30% of revenue (estimated)

Legacy products serving mature semiconductor applications: analog ICs, driver chips, power devices, older logic nodes. Revenue is flat to declining as the industry shifts toward MEMS for advanced applications.

• Customers: Diverse — domestic Japanese IDMs, foundries, analog chip makers
• Growth: Flat to slightly declining
• Competitive position: Historical strength but facing commoditization pressure

Segment C: Cleaning Sheets & Accessories — Small but high-margin

Consumable cleaning sheets used to maintain probe card and test socket contact quality. A small niche product that contributes ~27% "services-like" revenue characteristics.

Value Chain Position

[ATE Tester] → [Test Head] → [PROBE CARD] → [Wafer Prober] → [Wafer]
  Advantest       Interface      JEM, FORM,       TEL, Tokyo       TSMC,
  Teradyne                       MJC, MPI         Seimitsu        Samsung,
                                                                   SK Hynix

Probe cards sit in a unique position: they're a consumable interface between two capital equipment systems. Unlike testers or probers (purchased once per fab line), probe cards are consumed in production and replaced regularly. Revenue is driven by design starts (new chips need new cards) and production volume (more wafers tested = faster card wear) — a dual growth driver. A single fab may use hundreds of different probe card designs simultaneously.

Key ATE relationships:

• Advantest (~55-60% ATE share) has a minority stake in FormFactor — ensuring interoperability
• Teradyne (~30% ATE share) invested $516M in Technoprobe — same strategy
• JEM has no ATE partner — independent but potentially disadvantaged in winning new platform qualifications

Geographic Revenue Mix & Operations Footprint

JEM operates globally with heavy Asian concentration, consistent with where the memory fabs are:

• Japan: Largest single market — domestic memory fabs (Kioxia, Micron Hiroshima)
• Taiwan: ~¥5.5B — serving TSMC and other foundries
• China: ~¥3.3B
• South Korea: SK Hynix and Samsung — likely the fastest-growing region
• North America & Europe: Smaller shares

Manufacturing Sites:

Location	Function	Notes
Amagasaki HQ (Hyogo)	Main R&D and manufacturing	Corporate headquarters
Sanda Factory (Hyogo)	Primary probe card production	Main manufacturing facility
Kumamoto Plant	Manufacturing	Near TSMC's Kumamoto fab (JASM); benefits from local content mandates
New Factory (under construction)	MEMS M-type capacity expansion	Funded by Feb 2026 equity raise (~¥14B)
Fremont, California	Manufacturing + sales	Serves North American customers
Chupei, Taiwan	Probe manufacturing	Serves Taiwan foundry/fabless ecosystem
Shenzhen, China	Manufacturing	Serves mainland China market
Chonburi, Thailand	Manufacturing	Regional manufacturing base

Sales offices: Shin-Yokohama, Shanghai, Hong Kong, Seoul (JEMCO Korea), Grenoble (France), Singapore.

TAM & Market Share

The global probe card market was approximately $2.1-2.6B in 2024 and is projected to reach $3.3-5.4B by 2030-32, growing at a CAGR of 6-10%. The HBM market driving probe demand is projected at $54.6B in 2026 (+58% YoY per BofA).

JEM ranks approximately #5 globally with an estimated 5-7% global share. The top three players (FormFactor ~30%, Technoprobe ~18-20%, Micronics Japan ~14%) control roughly 60% of the market. JEM's realistic SAM is narrower — probably $800M-1.2B — reflecting its stronger position in Japan/Asia and limited penetration in the most advanced logic MEMS segments.

Joint Ventures & Strategic Partnerships

None. JEM operates independently without JV structures. Unlike FormFactor (Advantest minority stake) or Technoprobe (Teradyne invested $516M), JEM has no strategic investor from the ATE ecosystem. This is both a weakness (no guaranteed interoperability partnership) and a strength (no dependency on a single ATE platform).

JEM benefits from implicit partnerships through supplier qualification programs at major foundries — once qualified as a probe card vendor for a specific process node, JEM enjoys a sticky commercial relationship. But these are standard vendor relationships, not formal strategic alliances.

Secular Tailwinds

1. HBM production scaling: SK Hynix has sold out its entire 2026 HBM supply. Samsung and SK Hynix shipped >250M HBM stacks in 2025. The HBM market is projected to grow 600% cumulatively through 2030.
2. Rising test intensity: Advanced packaging (chiplets, 3D stacking) makes KGD testing mandatory. More test insertions per device = more probe card demand per wafer.
3. HBM generational upgrades: HBM3 → HBM3E → HBM4 → HBM4E — each generation increases pin count, tightens pitch, and requires new probe card designs. A structural design-start multiplier.
4. Japan fab investment: Japan's ¥2 trillion semiconductor subsidy program (TSMC Kumamoto, Rapidus, Micron Hiroshima) creates domestic sourcing mandates that benefit JEM directly.
5. Automotive electrification: Growing EV and ADAS content drives demand for tested power semiconductors, where JEM's automotive-grade CE series cards have an edge.
6. Node migration: Smaller process nodes have tighter electrical requirements, pushing adoption of more expensive advanced probe cards.

TerraPROBE (6627.T) Relationship

TerraPROBE Inc. is a key customer — one of the largest independent semiconductor test houses in Japan. They buy probe cards as consumables for their testing operations and are likely a top-5 customer for JEM. TerraPROBE's order backlog and utilization rates are a leading indicator for JEM's probe card demand. Both companies benefit from the same TSMC Kumamoto / Japan semi investment wave.

TerraPROBE is both a growth catalyst and a concentration risk. If they shift to a competing probe card supplier, the revenue impact could be significant. Monitor TerraPROBE's earnings as a cross-check on JEM's demand trajectory.

---

Competitive Landscape

The probe card market is more fragmented than most semiconductor equipment markets because of extreme customization requirements, regional supply preferences, and technology segmentation. The top 5 hold ~73%, but no player dominates the way ASML dominates lithography.

Company	Ticker	Revenue	Share	Focus	HBM Exposure
FormFactor	FORM	~$800M	~30%	Logic + Memory	High (HBM rev 4x'd in 2024)
Technoprobe	TPRO.MI	~$400M est.	~18-20%	Logic + Memory	Medium-High
Micronics Japan	6871/6871	6871.T	~$400M est.	~14%	Memory #1
JEM	6855.T	~$180M	~5-7%	Memory	High
MPI Corporation	6223.TW	~$200M est.	~5-7%	Mixed	Medium

JEM's competitive position: The smallest of the top 5 by revenue, but the purest memory/HBM play. While FormFactor and Technoprobe split attention between logic and memory, JEM is all-in on memory probe cards. This concentration is both a strength (deep customer relationships, focused R&D) and a vulnerability (no diversification when memory cycles turn).

Moat Analysis

JEM's competitive moat is moderate — narrower than typical semiconductor equipment companies:

Moat sources:

1. Customer qualification lock-in (moderate): Probe cards must be qualified per device and fab, creating switching costs for current production. But customers can switch vendors for the next device generation relatively easily.
2. Japan domestic preference (moderate): Japanese fabs prefer Japanese suppliers, and Japan's ¥2T subsidy program includes sourcing mandates.
3. MEMS fabrication expertise (moderate): Not all probe card makers can produce MEMS at scale, but FormFactor, Technoprobe, and MJC all can.
4. 55-year probe card heritage (weak): Reputation helps, but doesn't prevent customer defection.

Moat weaknesses:

• No ATE partnership (FormFactor has Advantest; Technoprobe has Teradyne)
• Smaller R&D budget than top 3 competitors
• No unique technology differentiation that competitors can't replicate

Porter's Five Forces

• Supplier power (Low-Medium): Probe card components (substrates, wire, MEMS materials) are specialty but not single-sourced. No supplier dominance.
• Buyer power (Medium-High): JEM's customers are enormous foundries and IDMs (TSMC, Samsung, SK Hynix, Micron) with significant negotiating leverage. Customer concentration is extreme — decisions of 3 memory makers determine JEM's fate.
• Threat of new entrants (Low): Probe card manufacturing requires deep domain expertise, precision engineering, and years of customer qualification — barriers to entry are high. Chinese entrants are trying but the technology gap for HBM-grade MEMS is significant.
• Substitute threat (Low): There is no substitute for probe cards in wafer-level testing. Contactless testing methods remain experimental and years from commercial viability.
• Competitive rivalry (High): The top three players are investing aggressively in MEMS technology and capacity expansion, putting pressure on smaller players like JEM. Korean domestic suppliers are emerging as direct competitors.

Emerging Threats

Korean probe card makers (Korea Instrument, TSE): The most immediate threat. Samsung and SK Hynix — JEM's primary customers — are actively cultivating domestic Korean probe card suppliers. TSE's probe card revenue grew 90% year-over-year in 2025. This is the most credible competitive threat to JEM in its core market.

FormFactor's memory expansion: FormFactor's HBM revenue quadrupled in 2024, and the company is investing aggressively in memory probe cards. With $800M in revenue and 15%+ R&D intensity, FormFactor has the resources to compete in any segment it targets.

Chinese entrants (MaxOne, Shanghai Zenfocus, JUNR/Wuxi Junr): Currently ~2% global share, competing on speed and price in the low-end market. The technology gap for HBM-grade MEMS cards is significant, but China's self-sufficiency push could accelerate.

Contactless testing: Long-term, if the industry develops non-contact testing methods (optical or electromagnetic), probe cards become obsolete. A distant risk — no viable alternative exists today — but worth noting for a 5+ year horizon.

Contrarian read — JEM may be NAND-only, not HBM-qualified (source: 2026-05-08)

A market source pushed back on JEM's HBM positioning: claims JEM's probe card business is NAND-only, with only MJC (6871) and FormFactor qualified on HBM probe cards specifically. NAND probe cards are commoditizing under Chinese price competition; HBM is the segment with pricing power and customer lock-in.

If true, this materially changes the thesis: JEM is competing in a commoditizing segment (NAND) while the wiki frames it as a pure HBM play. The "valuation gap" vs MJC at 9.4x P/E (already noted in Peer Trading Comps below) may then reflect a real product-mix gap, not a market dislocation.

Verification needed: confirm via (a) JEM IR / FY2025 segment disclosure on HBM-qualified products, (b) Samsung/SK Hynix supplier qualification lists, (c) MJC and FORM filings naming JEM as a competitor in HBM specifically. Until verified, treat HBM exposure in this wiki page as unconfirmed.

Peer Trading Comps

Metric	FormFactor (FORM)	JEM (6855)	MJC (6871)	MPI (6223.TW)
Market Cap (USD)	$6.6B	$670M	$3.4B	$8.5B
Revenue (LTM, USD)	$785M	~$176M	~$468M	~$392M
P/E (Forward)	~41x	~14x	~9.4x	~51x
EV/EBITDA (NTM)	~20x	~11x	~15x	~35x
EV/Revenue (NTM)	~6.9x	~2.5x	~6.0x	~13x
Gross Margin	39%	40%	48%	56%
Operating Margin	7% (GAAP)	20%	23%	29%
PEG	~4.0x	~1.5x	~1.2x	~1.9x
Dividend Yield	0%	1.6%	1.0%	0.6%

JEM trades at a significant discount to all peers on forward P/E and EV/Revenue. The discount reflects smaller scale, Standard Market listing, recent dilution, and single-analyst coverage. Notably, MJC at 9.4x forward P/E is actually cheaper than JEM at 14x, despite larger scale, higher gross margins (48% vs. 40%), and a TSMC 2nm logic qualification JEM lacks. Either MJC is undervalued relative to JEM, or the market assigns a higher HBM purity premium to JEM than the fundamentals justify.

Industry Cycle Position

The probe card market is in the mid-to-late expansion phase of a semiconductor test cycle:

• Trough: FY2024 (JEM revenue -16%, OP margin collapsed to 5.8%)
• Recovery: FY2025 (revenue +36.5%, sharp margin rebound)
• Expansion: FY2026 (record profits expected, new capacity under construction)
• Peak risk: FY2027-28 — if AI capex moderates or HBM production overshoots demand

Historical semiconductor test cycles typically last 3-4 years from trough to peak. We are approximately 2 years into the recovery. The HBM-specific demand may extend this cycle beyond historical norms, but don't assume the cycle has been abolished.

---

Management

CEO: Teruhisa Sakata (坂田輝久)

Detail	Info
Born	April 11, 1962 (age 63)
Appointed President	June 2023
Joined JEM	April 1986 (40 years at the company)
Shares Owned	15,000 (0.12%, ~¥105-127M / $700-850K)
Education	Not publicly disclosed

Career progression: Pure lifer — joined JEM in 1986 straight from school. Rose through Development Division → Technical Strategy → Sales & L-Products → MEMS Division → Quality Assurance → President. He covered the full product lifecycle before elevation: R&D, sales, manufacturing (MEMS), and quality control.

How he got the role: Classic Japanese internal grooming. Ohkubo stepped down as president in May 2023 and handed it to Sakata. Not founder family — a career technocrat promoted by the family-connected chairman.

Assessment: Deep institutional knowledge but zero outside career experience. He knows JEM's operations cold but may lack perspective on strategic pivots or governance evolution. For a probe card company where technology drives competitive advantage, an engineering-track CEO makes sense. His MEMS division experience is directly relevant to JEM's growth engine. But his ownership stake is tiny — financial alignment is negligible; career risk is his only incentive.

Chairman: Kazumasa Ohkubo (大久保和正)

Detail	Info
Born	March 17, 1955 (age 71)
Role	Director and Chairman (lost "representative" designation June 2024)
Tenure	Joined JEM 1985; President 2017-2023; Chairman since May 2023
Shares Owned	511,000 (~4.03%, ~¥3.6-4.3B / $24-29M)
Education	San Jose State University, Materials Science (1981)

Career: Started at Tokyo Electron (1983-1985), then joined JEM in 1985. Ran European and American subsidiaries (Chairman of JEM Europe 2003, JEM America 2004). VP from 2008, President from 2017, Chairman from 2023.

Founding family connection: The Ohkubo surname is shared with the founding family entity Ohkubo Kogyosha Co., Ltd. (大久保興産), which holds 3.37%. Combined with Kazumasa's personal 4.03% and Hidemasa Ohkubo's 1.68%, the Ohkubo family bloc is ~9.1% — the de facto controlling family group. The prior patriarch Masao Ohkubo held ~20% at peak.

Assessment: The shadow power behind the throne. He stepped down from "representative director" in June 2024 — technically a governance positive — but retains the Chairman title, the largest individual shareholding, and family influence. His international experience (US-educated, ran overseas subsidiaries) adds useful perspective for a company with growing Asian customer relationships. The real question is how much operating authority Sakata actually has.

Other Key Leaders

Name	Role	Background
Yoshiyuki Miyamoto (宮本佳幸)	Managing Director, heads MEMS Division	Age 67. Came from Renesas/Hitachi. The lone external hire on the executive board — brought in for MEMS fabrication expertise JEM lacked internally.
Keiichi Tatsu (龍圭一)	Board Director (since June 2025), Product Technology	Ex-Toshiba Memory/Kioxia. Second meaningful external hire at the technology layer.
Yasutaka Adachi (足立安孝)	Audit Committee Chair	Age 74, board since 2009 (16+ years). Long tenure raises independence questions.
Moriyasu Sawai (澤井守康)	Sales Division Head	Heads commercial operations.
Akihiko Fujii (藤井昭彦)	Production Division Head	Manufacturing operations.

8 executive officers total below the board. Leadership is insider-dominated with two meaningful external hires (Miyamoto from Renesas, Tatsu from Kioxia) at the technology layer.

Management Track Record

JEM has navigated semiconductor cycles with mixed but generally solid results:

Fiscal Year	Revenue (¥B)	OP (¥B)	OP Margin	Notes
FY2021 (COVID recovery)	18.5	2.7	14.4%	Recovery year
FY2022 (memory boom)	23.6	5.0	21.0%	Peak cycle
FY2023	20.8	3.2	15.4%	Memory downturn managed
FY2024 (trough)	17.5	0.87	5.0%	Brutal but remained profitable
FY2025 (recovery)	23.8	4.6	19.2%	Strong rebound
FY2026E (HBM boom)	28.1	6.5	23.1%	Record year expected

The company stayed profitable even in the FY2024 trough, which is notable for a cyclical semiconductor equipment company. They maintained dividends and R&D through the downturn — both discipline signals.

Ownership Structure

Category	Approximate %
Institutional shareholders	~25-33%
Insiders (Ohkubo family + management)	~10-19%
Public / retail	~50-56%

Key institutional holders:

• Nomura Asset Management: ~6.5-9.35% (largest institutional)
• Japan Master Trust Bank: ~14.7% (passive/supportive)
• Mitsubishi UFJ Financial Group: ~2.44% (possibly policy shareholding)
• UBS AG: reported holder (foreign institutional)

No controlling shareholder or parent company. JEM is independent. The Ohkubo family collectively holds meaningful influence but not a controlling stake. Float is adequate given ~50%+ public ownership, though the TSE Standard Market listing means lower liquidity than Prime.

Insider Ownership & Skin in the Game

Name/Entity	Role	Shares	% Outstanding	Est. Value
Kazumasa Ohkubo	Chairman	511,000	4.03%	~$24-29M
Ohkubo Kogyosha Co.	Family holding co.	427,000	3.37%	~$20-24M
Hidemasa Ohkubo	Director/VP	213,000	1.68%	~$10-12M
Yasutaka Adachi	Audit Committee Chair	32,000	0.25%	~$1.5-1.8M
Teruhisa Sakata	President & CEO	15,000	0.12%	~$700-850K
Yoshiyuki Miyamoto	Managing Director	5,000	0.04%	~$230-280K
Ohkubo family total		1,151,000	~9.1%	~$54-65M
Professional mgmt total		~53,000	~0.42%	~$2.5-3M

The family has real skin in the game — ~$54-65M concentrated in JEM. Professional management does not. CEO Sakata's 15,000 shares worth ~$700-850K is trivial relative to his influence over a $580-700M market cap company. His ownership comes entirely from restricted stock grants (RSU plan introduced May 2022), not personal purchases. No open-market purchases by professional management identified.

A restricted stock plan was introduced in May 2022, with annual RSU grants each June (confirmed 2022-2025). Dilution is minimal (+0.07% YoY share count change). But the absence of meaningful equity incentives remains a governance gap.

Okubo Kogyosha has a stock lending agreement with Japan Securities Finance Co. capping at 527,000 shares — could obscure effective control dynamics.

Where Is Their Money Really?

Name	Role	JEM Holdings	Where Is Majority of Wealth?
Kazumasa Ohkubo	Chairman	$24-29M, 4.03%	Likely JEM + Ohkubo Kogyosha — significant concentration
Hidemasa Ohkubo	Director	$10-12M, 1.68%	Likely JEM-concentrated
Teruhisa Sakata	President	~$700-850K, 0.12%	Almost certainly elsewhere — JEM stake is trivial
Yoshiyuki Miyamoto	Managing Dir	~$230-280K, 0.04%	Elsewhere — 40+ year career at Renesas/Hitachi

The Ohkubo family's wealth is meaningfully concentrated in JEM — this is generational commitment, not a temporary position. Professional management has negligible personal capital at risk — alignment is career-based, not financial. No evidence of executives holding conflicting positions in customers, suppliers, or competitors.

Board & Governance

Director	Type	Independent?	Tenure	Background
Kazumasa Ohkubo	Internal (Chairman)	No	40 years at JEM	Founding family, SJSU + Tokyo Electron
Teruhisa Sakata	Internal (President)	No	39 years at JEM	Career lifer
Yoshiyuki Miyamoto	Internal (Managing Dir)	No	~5 years at JEM	Ex-Renesas/Hitachi
Keiichi Tatsu	Internal (new 2025)	No	~4 years at JEM	Ex-Toshiba Memory/Kioxia
Yasutaka Adachi	Internal (Audit Chair)	No	16+ years on board	JEM career insider
Yukikazu Hamada	Outside (Audit)	Yes	Since 2017	Tax accountant
Erika Chiba-Sakura	Outside (Audit)	Yes	Since 2023	NY attorney, IP specialist
Wataru Miyajima	Outside	Yes	Pre-2025	Attorney, Frontier Law

Board independence: 3 of 7-8 directors = 37.5-42.9%. Female representation: 1 of 7-8 (12.5-14.3%).

JEM uses the "Company with Audit and Supervisory Committee" (監査等委員会設置会社) governance model. Two of three Audit Committee members are independent. Voluntary advisory Nomination and Compensation Committees were established in March 2022.

JEM chose the TSE Standard Market in the 2022 restructuring — not the more demanding Prime Market. This avoids stricter governance requirements (higher free float, enhanced English disclosure, more rigorous governance code). This is a governance-ambition signal.

Anti-Takeover & Shareholder Activity

No formal poison pill or takeover defense plan identified. De facto protection comes from:

• Ohkubo family bloc (~9.1%)
• Japan Master Trust Bank (14.7%, passive/supportive)
• Mitsubishi UFJ Bank cross-holding (2.44%)

No activist campaigns, contested elections, or shareholder proposals found.

CRITICAL: External Auditor — Hibiki Audit Corporation (PKF Japan)

Date	Event
Jan 2023	CPAAOB recommended disciplinary action — partners falsified audit working papers after receiving inspection notice and concealed this from inspectors
Mar 2023	FSA issued Business Improvement Order #1
Jun 2024	FSA issued Business Improvement Order #2 — triggered exodus of 13 clients including three of Hibiki's largest

JEM did not switch auditors. Audit fees are only ¥26M — extremely low for a company with ¥24B+ revenue. This is the single most concerning governance finding. Why does JEM tolerate a discredited auditor when alternatives exist?

Compensation

Category	Total (¥M)	# Persons	Per Person (¥M)	Per Person (USD)
Directors (non-audit)	106	3	~35	~$230K
Audit Committee (internal)	18	1	18	~$120K
External Directors	17	3	~6	~$40K

CEO compensation of roughly ¥35M is modest by global standards but typical for a Japanese TSE Standard company. Total board pay of ¥141M for 7 directors is at the low end. CEO-to-median-worker ratio estimated at ~3.5-5x (extremely low globally). No director earns enough to trigger the ¥100M individual disclosure threshold.

Compensation has three components: fixed salary, performance-linked cash bonus (genuinely declined ¥56M from FY2024 to FY2025 when performance weakened), and restricted stock grants (since 2022). No golden parachutes, no excessive perks, no related-party leases identified. This is not a management team extracting value through compensation.

Employee Culture Flag

OpenWork score: 2.54/5.0 with "long-term talent development" at 2.0/5. Reviews describe a facade of competence externally while new employees leave quickly. 100+ overtime hours/month reported for some roles. For a technology company dependent on engineering talent, this is structural risk.

Management DD Verdict

Dimension	Rating	Key Finding
Skin in the Game	Yellow	Ohkubo family ~9.1%, ~$54-65M (good). CEO 0.12% (weak). Split verdict.
Holdings Concentration	Green	Family wealth concentrated in JEM. No conflicting interests.
Shell / Cross-Holdings	Green	Simple structure. Ohkubo Kogyosha is a passive family vehicle. No complex entity webs.
Capital Allocation	Red	Repeated dilutive equity raises from net cash; no buybacks ever.
Compensation Alignment	Green	Modest, performance-linked, declined with earnings. RSU plan introduced 2022.
Governance Quality	Red	Auditor Hibiki has TWO FSA orders for fabricating documents — JEM kept them. Standard Market listing.
Litigation / Enforcement	Green	Clean — no lawsuits, no regulatory actions, no patent disputes. 65-year clean record. 853 patents (~$157M portfolio).
Overall Management Grade	Yellow / C+	Honest and modest management. Founder alignment is real; professional alignment is weak. Capital allocation and the auditor issue are the biggest concerns.

Management Green Flags

• Ohkubo family has ~$54-65M concentrated in JEM — generational commitment
• No litigation, no regulatory actions, no scandals — 65-year clean record
• Compensation is genuinely modest — total board pay ¥141M for 7 directors
• Dividend maintained through FY2024 earnings collapse (81% payout ratio)
• Simple corporate structure — no opaque SPVs, no layered holdings
• Minimal dilution from RSUs — share count essentially flat YoY (the PO is a separate issue)
• Fortress balance sheet — ¥6.3B+ net cash, 73% equity ratio
• External hires at technology layer (Miyamoto from Renesas, Tatsu from Kioxia) — recognizing gaps
• 853 patents with ~$157M estimated portfolio value
• R&D maintained through downturns — no cuts when earnings collapsed

Management Yellow Flags

• CEO Sakata owns only 0.12% — financial alignment is negligible
• Standard Market listing — chose to avoid Prime Market governance requirements
• Kumamoto capex expansion — ¥4B+ bet on logic-focused TSMC adjacency for a memory company
• No buyback program — missed the FY2024 trough (stock at ¥1,252 vs. ¥7,000-8,500 now)
• Incentive comp metrics not fully disclosed
• JEMCO domestic subsidiary function not publicly disclosed
• Revenue cyclicality — 80% OP decline FY2022 to FY2024

Management Red Flags

• Auditor Hibiki received TWO FSA Business Improvement Orders for fabricating audit working papers and concealing evidence. Thirteen clients fled. JEM stayed. Audit fee of ¥26M appears too low. This is the single biggest red flag.
• Employee culture deterioration — OpenWork score 2.54/5.0 with 100+ overtime hours/month reported for some roles. New employees leave quickly. For a technology company dependent on engineering talent, this is structural risk.
• Repeated dilutive equity raises from a net cash position without ever buying back shares — 37% share count growth since FY2021.

Bottom line: You're betting on a family-influenced, conservative, technically capable but culturally insular Japanese small-cap — and the auditor issue means you can't fully trust the numbers until they switch to a credible firm. Trust the family's alignment, but verify everything through the auditor lens.

---

Financials

Income Statement & Margins

JEM's fiscal year ends March 31. All figures in JPY billions unless noted.

Metric	FY2022	FY2023	FY2024	FY2025	FY2026E
Revenue	23.6	20.8	17.5	23.8	28.1
Revenue growth	—	-11.9%	-16.0%	+36.5%	+17.9%
Gross profit	—	7.4	5.2	9.5	—
Gross margin	—	35.6%	30.0%	39.7%	—
Operating profit	5.1	3.3	1.0	4.6	6.5
Operating margin	21.6%	16.1%	5.8%	19.2%	23.1%
Net income	~3.5	~2.2	~0.6	~3.5	4.3
Net margin	~14.8%	~10.6%	~3.4%	~14.5%	~15.3%
EPS (¥)	~280	~175	~48	~274	~340 (pre-dilution), ~299 (post-dilution)
DPS (¥)	40	40	40	70	80

The cyclicality is extreme. From peak (FY2022) to trough (FY2024), operating profit fell 80% and margins collapsed from 22% to 6%. The recovery has been equally dramatic — revenue rebounded 36% in FY2025 and is guided up another 18% in FY2026, with operating margins expanding past the prior peak. This is classic operating leverage in a high-fixed-cost manufacturing business.

The 9-month actual FY2026 results (revenue +40%, OP +78% YoY) are tracking ahead of full-year guidance, suggesting management is being conservative on Q4.

Financial Health Assessment

Dimension	Grade	Notes
Revenue Growth	Strong	+47% TTM; +18% NTM guided
Margins	Expanding	Gross 40%, Operating 23%, both at/near records
Balance Sheet	Excellent	Net cash ¥7B+, current ratio 4.2x, D/E 0.22
Cash Flow	Investment phase	Negative FCF in FY2025 due to capex ramp — acceptable given context
ROIC vs. WACC	Value creating	13-23% ROIC vs. 8-9% WACC = positive spread
Cyclicality	High risk	Revenue swung ¥24B → ¥17B → ¥30B in 3 years
Accounting	Concern	Discredited auditor (Hibiki) — cannot fully verify
Overall Financial Health	B+	Strong fundamentals with cyclical and audit risk

No restatements, no going concern issues, no accounting red flags beyond the auditor question.

Key financial red flags to monitor:

• Cyclicality amplitude: Revenue swung from ¥24B (FY2023) → ¥17B (FY2024 trough) → ¥30B (FY2026E). This is a high-amplitude cyclical business — don't mistake a cyclical upswing for secular growth.
• Capex cycle risk: Heavy investment in new factory may produce negative FCF through FY2027, though the investment is strategically sound if HBM demand materializes.
• Post-PO EPS dilution: FY2026 EPS drops from ~¥340 pre-dilution to ~¥299 post-dilution. All forward valuation must use the post-dilution share count (~14.4M).

Incremental Margin Analysis (9M FY2026 vs. Prior Year)

Metric	9M FY2025	9M FY2026	Delta	Incremental Margin
Revenue	¥14,739M	¥20,675M	+¥5,936M	—
Operating Profit	¥2,833M	¥5,028M	+¥2,195M	37.0%
Net Income	¥2,052M	¥3,468M	+¥1,416M	23.9%

Incremental operating margin of 37% on ¥5.9B of incremental revenue is very strong — new revenue is coming in at significantly higher margins than the existing business. This is exactly what you'd expect from the MEMS/HBM mix shift. If this holds as the new factory ramps, blended operating margins could approach 25-27% at full capacity.

Cash Flow & Balance Sheet

Metric	FY2023	FY2024	FY2025	Latest (Dec 2025)
Operating cash flow	3.3B	2.3B	1.8B	—
Capex	-0.9B	-1.9B	-3.5B	—
Free cash flow	2.4B	0.4B	-1.7B	—
Cash & deposits	—	—	12.2B	13.2B
Total debt	—	—	5.9B	6.1B
Net cash	8.5B	8.3B	6.3B	7.1B (pre-raise)
Debt/Equity	—	—	—	20.2%
Current ratio	—	—	—	4.19x
Equity ratio	—	—	73.1%	—

FY2025 showed negative FCF despite record-level revenue because capex nearly doubled to ¥3.46B — JEM is investing heavily in manufacturing capacity. This is investment-phase spending, not business deterioration.

Detailed Balance Sheet

Metric (¥M)	TTM (Dec 2025)	FY2025	FY2024	FY2023
Cash & Equivalents	13,244	12,148	14,217	12,779
Total Debt	6,137	5,920	5,981	4,300
Net Cash	7,107	6,228	8,236	8,479
Shareholders' Equity	30,380	27,913	24,669	24,241
Total Assets	41,572	39,859	34,769	32,691

The balance sheet is fortress-like. Post equity raise, net cash could reach ~¥21B before factory construction spending begins. No financial distress risk. The company has ample room to fund expansion — arguably too conservatively levered.

Metric	Value
ROIC (current)	~13-23% (sources vary)
ROE (TTM)	17.2%
Estimated WACC	~8-9% (beta 2.19)
ROIC-WACC spread	+5-14% — creating value

R&D Investment

Fiscal Year	R&D (¥M)	% of Revenue
FY2021	1,447	7.8%
FY2022	1,414	6.0%
FY2023	1,498	7.2%
FY2024	1,634	9.4%
FY2025	1,538	6.5%
FY2026 TTM	1,662	5.6%

R&D spending remarkably stable at ¥1.4-1.7B annually — maintained even during the FY2024 downturn. This is a positive signal showing long-term technology investment discipline. 853 patents with ~$157M estimated portfolio value.

Capital Allocation

Dividend History

Fiscal Year	DPS (¥)	Payout Ratio	Notes
FY2021	15-28	~8-15%	Low base, COVID period
FY2022	40	~13%	Significant increase
FY2023	40	~19%	Maintained through downturn
FY2024	40	~81%	Maintained despite weak earnings — shows commitment
FY2025	65-70	~23-25%	Increased with recovery
FY2026E	80	~24-26%	Further increase; interim ¥50 declared (ex-date Mar 30, 2026)

Consecutive dividend increases ongoing. 5-year dividend CAGR of ~23%. Payout ratio target appears to be 20-30% during normal years — conservative but leaves room for growth investment.

Buybacks: None. Ever. JEM has never conducted a share buyback. Treasury stock: 15,900 shares (0.13%). Missed a significant opportunity during the FY2024 trough when stock traded at ~¥1,252 (vs. ~¥7,000-8,500 now).

M&A: None. JEM has grown entirely organically, which is unusual for probe cards (FormFactor has been acquisitive) but consistent with conservative Japanese management culture.

The February 2026 Equity Raise — a case study in suboptimal capital allocation:

On February 25, 2026, JEM announced a public offering of 1.74M new shares (15.8% dilution) to raise ~¥14B for a new MEMS probe card factory. The stock hit limit-down on the announcement.

The strategic rationale is sound — HBM demand is surging and JEM needs capacity. But:

• JEM was sitting on ¥7B+ net cash at the time
• Debt-to-equity was only 20.2% — ample room for debt
• A ¥14B debt raise at Japanese corporate rates (~1-2%) would cost ¥140-280M/year in interest — far cheaper than ~¥700M+ annual earnings dilution from 15.8% more shares
• This is the second significant dilution in recent years (shares grew ~37% from 10.7M in FY2021 to 12.6M pre-PO)

This pattern — equity raises from a net cash position with no buyback history — signals either excessive balance sheet conservatism or a management team that doesn't internalize dilution as a cost.

Capex & Capacity Expansion

Year	Capex (¥M)	Revenue (¥M)	Capex/Revenue
FY2021	914	18,521	4.9%
FY2022	860	23,599	3.6%
FY2023	921	20,781	4.4%
FY2024	1,918	17,461	11.0%
FY2025	4,015	23,829	16.8%

Capex surged 4x from FY2023 to FY2025 — Kumamoto factory expansion + MEMS/HBM capacity build. The Feb 2026 equity raise of ¥14B is specifically earmarked for a new factory to expand M-type probe card production capacity for HBM testing. This represents roughly one-third of current total assets.

Kumamoto expansion note: ¥4B+ investment targeting TSMC adjacency, but JEM is memory-focused while TSMC Kumamoto is logic (12-28nm). Strategic fit is indirect and requires qualification of JEM products for logic testing — a different competitive dynamic where FormFactor and Technoprobe are stronger.

Dilution History

Year	Shares Outstanding	Change
FY2021	10.7M	—
FY2022	12.2M	+14%
FY2024-25	12.6M	+3%
Post-PO (2026)	~14.4M	+14%
Total since FY2021		+37%

37% share count growth over 5 years is a meaningful drag on per-share value creation.

Valuation

Current Multiples

Multiple	Value	Assessment
P/E (TTM)	~27x
P/E (FY2026E, post-dilution)	~28x	Pricing in "things go right"
Forward P/E (FY2027E consensus)	~14x
EV/Revenue (FY2026E)	~3.6x	Discount to peers
EV/EBITDA (TTM)	~11x	Attractive
EV/EBIT	12.2x
PEG Ratio	~1.0-1.5x	Fair to moderate
P/FCF	~37x	Elevated (capex-heavy phase)
P/B	~4.1x
Dividend yield	~0.9-1.6%

Historical Comparison

Metric	Current	5Y Avg	Premium/Discount
NTM P/E	23-28x	~15x	+54-87% premium
Justified?	Partially — HBM growth cycle justifies elevated multiples, but 54%+ above history requires continued execution

Growth Implied by Current Price

At ¥8,490 and FY2026 EPS of ~¥299 (post-dilution), the stock trades at ~28x. For this to be "fair," EPS needs to grow at ~15%+ annually for the next 3 years. Forward consensus estimates:

• FY2027E EPS: ¥371.5 (+9.3%)
• FY2028E EPS: ¥403.1 (+8.5%)

This implies ~8-9% EPS CAGR, meaning the current 28x P/E is pricing in more growth than consensus expects. Either consensus will be revised up (likely given the recent beat), or the stock is somewhat expensive on current estimates.

DCF & Third-Party Fair Value Estimates

Source	Fair Value	vs. Current Price
Alpha Spread (base case)	¥4,226	Overvalued by ~48%
Alpha Spread (DCF only)	¥4,389	Overvalued by ~46%
Morningstar Quantitative	¥8,592	Roughly fair (~+1%)
Kabuyoho (PER-based)	¥3,815	Overvalued by ~52%

Note: Alpha Spread and Kabuyoho estimates likely use pre-revision (pre-Feb 6, 2026) earnings inputs. Morningstar's ¥8,592 is the most current and suggests fair value near current price.

Valuation Verdict

Fair value: ¥6,500-7,500 based on 22-25x post-dilution FY2026E earnings. Incorporates dilution overhang, cycle risk, and competitive threats.

Bull case: ¥10,000-12,000 if FY2027 OP reaches ¥8-10B (new factory at capacity, HBM cycle extends), implying 20-25x on ¥400-500 EPS.

Bear case: ¥3,000-4,000 if the memory cycle turns. At 15x trough-adjusted earnings, the stock could retrace 60-70% from current levels.

At ¥8,490 (~28x post-dilution FY2026E P/E), the market is pricing in FY2026 record profits delivered as guided, continued HBM growth through FY2027-28, successful factory ramp, no significant customer losses, and no cycle downturn. This is a "things go right" price with limited margin of safety.

Would you still buy at 10-15% higher (¥9,200-9,700)? No. At ¥9,500, the forward P/E stretches to ~32x, which is expensive for a cyclical business with 15.8% fresh dilution. The margin of safety disappears entirely.

Valuation Grade: C+ — Not cheap, not expensive at current levels. Fair value territory based on Morningstar, but well above historical averages. The PEG ratio of ~1.0-1.5x is reasonable, but the premium to historical P/E and cyclical risk temper enthusiasm.

---

Catalysts & Risks

Catalysts

Near-term (0-12 months):

• May 19, 2026: FY2026 full-year results and FY2027 guidance — if guidance exceeds ¥30B+ revenue and ¥7B+ OP, stock could re-rate
• New factory construction progress updates
• HBM4 design wins or disclosed probe card orders
• Analyst initiation (only 1 analyst covers — new coverage increases institutional awareness)
• Memory maker capex announcements (Samsung and SK Hynix typically announce mid-year)

Medium-term (1-3 years):

• New factory ramp (FY2027-28) — at capacity, revenue could step-change to ¥35-40B+ with 25%+ OP margins
• HBM4 production start (expected 2027), requiring new probe card designs with even higher pin counts
• Potential TSE Prime Market upgrade — would increase institutional eligibility and liquidity

Risks

Risk	Likelihood	Impact	Mitigation
HBM capex cycle peak — memory makers cut capex, probe card demand collapses	Moderate (12-18 months)	Very High — revenue could drop 30-40%, OP margin collapse to single digits (happened in FY2024)	Monitor Samsung/SK Hynix/Micron capex guidance quarterly
Korean competitor share gains (Korea Instrument, TSE growing 90% YoY)	Medium-High	Medium — gradual share erosion in JEM's core market	Japan domestic preference; 55-year customer relationships
Customer concentration — Samsung, SK Hynix, Micron = most of revenue	High (structural)	High — capex decisions of 3 customers determine JEM's fate	3 is better than 1, but still concentrated
Post-PO dilution overhang	Certain (already happened)	Medium — 15.8% EPS dilution	Closes as new factory ramps and earnings grow into higher share count
New factory execution risk	Low-Medium	Medium	JEM has built factories before; ¥14B+ in cash
Competitive displacement by MEMS leaders	Medium	Medium	Investing in MEMS technology; capacity expansion
Japan policy reversal	Low	Medium	Broad bipartisan support; geopolitical dynamics favorable

Technology Roadmap

JEM's R&D (¥1.4-1.7B annually, 6-9% of revenue) focuses on:

• MEMS probe performance: finer pitch, longer needle life, higher pin count capability
• Advanced substrate development for next-gen probe cards
• Assembly and processing technology for HBM4+ requirements (expected to require even tighter pitch and higher channel counts)
• Potential logic testing expansion — qualifying for TSMC Kumamoto adjacency

R&D spend has been remarkably consistent through cycles — a positive signal that management doesn't cut technology investment in downturns. The 853-patent portfolio ($157M estimated value) provides a defensive IP moat, though notably JEM has no patent litigation history (unusual in the patent-intensive probe card industry).

Near-Term Event Risk

• FY2026 full-year results: Expected May 19, 2026. Q4 performance vs. raised guidance will be critical. If Q4 shows deceleration, it signals cycle peak.
• Memory maker capex announcements: Samsung and SK Hynix typically announce mid-year. Any cuts would impact JEM directly and immediately.
• New factory construction progress: Any delays or cost overruns would undermine the equity raise justification.
• Auditor situation: Monitor for any auditor change announcement — would be a significant governance positive.
• No known litigation, regulatory, or M&A event risk.

Bear Case Scenario

HBM demand plateaus in late 2027 as AI capex normalizes. Samsung and SK Hynix shift probe card orders to cheaper Korean suppliers. JEM's new factory comes online just as demand softens. Revenue falls 25-30% from peak, operating margins collapse to low single digits, and the stock — trading at ~28x peak earnings — de-rates to 15x trough earnings. Target: ¥2,500-3,500 — a 60-70% drawdown.

Historical precedent: In the FY2022-FY2024 downcycle, the stock fell from ~¥5,000+ to ¥1,252 (-75%). Operating profit declined 80%. This volatility cuts both ways — the stock can be a multi-bagger off the bottom but also a 75% drawdown from the peak.

Downside Scenario Math

If HBM capex normalizes and JEM reverts to mid-cycle earnings (OP margin ~15%, revenue ~¥22B), EPS would be approximately ¥200. At a 15x trough P/E, the downside target is ¥3,000 — approximately -65% from current levels. At a 20x normalized P/E, you get ¥4,000 — still -53% from here.

The key variable is whether the current cycle is "normal" or structurally different due to AI/HBM. If HBM demand truly represents a structural shift in test intensity (not just a cyclical bump), then mid-cycle earnings are higher than history suggests — maybe ¥22-25B revenue with 18-20% OP margins. That gives a more favorable normalized valuation. But betting on "this time is different" in a cyclical semiconductor business requires high conviction.

Portfolio Correlation

• High correlation to semiconductor capex cycle
• Correlated to other Japanese tech holdings (Santec, Dexerials if held)
• HBM-specific exposure provides some differentiation vs. broad semi names
• Concentration warning: If you hold other HBM-exposed names (SK Hynix, Micron, FormFactor), JEM adds to that cluster risk

---

Decision Log

2026-04-07 — Discord datapoint: HBM probe card record + Zhubei expansion

From Collyer Bridge in market chatter Discord (Apr 1, 2026), citing a Taiwan-based semi testing interface provider M-H Corporation report:

• JEM "is back in the HBM game" after the Feb retrace
• M-H Corporation report says probe cards have pushed probe card lead times to as long as 16 months, with visibility for some orders extending up to two years. Chairman Ko Chang-In expects the company's operations to grow quarter by quarter in 2H 2026, targeting double-digit annual revenue growth and a strong likelihood of reaching a new record high for the full year.
• To support future capacity expansion, MPI (6278) noted it has recently acquired two plots of land in Zhubei. Starting in the second half of 2026, the company will ramp up production of vertical probe cards (VPCs) and MEMS probe cards, with capacity expected to increase by up to 50%, in response to the growing scale of HBM demand and its strategy to increase in-house production of probe cards.

Why it matters: Independent confirmation of the HBM probe card thesis on JEM, with named lead-time data (16-24 months) and a competitor's capacity story. Collyer's view: "If you think about what the Feb results confirm, that they're totally in the HBM game, why is it not compelling at these levels." Worth a re-look at the Feb retrace as an entry rather than a danger signal.

Technical Buy Check (as of early March 2026)

Trend

Indicator	Status
Price vs. 50-day MA (~¥7,857)	Above (+8%) — Bullish
Price vs. 200-day MA (~¥5,395)	Above (+57%) — Bullish but extended
50d vs. 200d MA	Golden cross — Bullish
Trend structure	Higher highs and higher lows — Uptrend

Momentum

Indicator	Status
RSI (14-day)	69.3 — Borderline overbought (threshold: 70)
MACD	+157.4, positive histogram — Bullish
All technical indicators	12/12 MAs Buy, 9/9 oscillators Buy

Support & Resistance

Level	Price	Distance from ~¥8,490
All-time high / Resistance	¥10,270 (Feb 25, 2026)	+21% above
Fibonacci Pivot	¥8,350	Near current price
50-day MA Support	¥7,857	-7% below
Prior consolidation	¥6,000-6,500	-24% to -29% below
200-day MA Support	¥5,395	-36% below

Volume & Liquidity

• 50-day average volume: ~302,196 shares/day
• Volume elevated around Feb earnings revision but normalizing
• Liquidity caution: This is a small-cap on the Standard Market with relatively low volume. Larger positions may face slippage.

Technical Verdict

NEUTRAL — leaning favorable for a scaled entry. The uptrend is intact (golden cross, all MAs bullish), but RSI at 69 is right at the overbought line. The ~30% pullback from the ¥10,270 high provides a better entry than buying at all-time highs. Key risk: The +57% gap above the 200-day MA is extreme and historically unsustainable. Mean-reversion risk is elevated.

Pre-Buy Scorecard

Question	Answer
Can I state the thesis clearly?	Yes — HBM probe card demand growth
Do I understand the business?	Yes — probe cards for memory testing
Are the financials healthy?	Yes — net cash, 13-23% ROIC, 40% gross margins
Is the valuation reasonable?	Borderline — 28x post-dilution PE is "things go right" pricing
Am I falling into a behavioral trap?	Caution — FOMO risk after 578% run; narrative seduction ("HBM is the future")
Do the technicals support buying now?	Neutral — uptrend intact but extended above 200d MA
Have I sized the position appropriately?	Yes — 1-2% max with scaled entry
Do I have a clear exit plan?	Yes — sell on HBM capex decline or PE >35x without earnings support

Behavioral Traps Audit

Trap	Status	Notes
FOMO	CAUTION	Stock up 578% in 52 weeks. Extraordinary run. Risk of chasing.
Confirmation bias	Clear	Both positives and negatives fully analyzed.
Authority bias	Clear	Only 1 analyst covers. No famous investor thesis to follow.
Narrative seduction	CAUTION	"HBM is the future" is powerful. But probe cards are cyclical — last downturn saw revenue drop 30%.
Recency bias	CAUTION	Q3 was exceptional (+78% OP growth). One great quarter doesn't mean the cycle lasts forever.

Recommendation: WATCHLIST — Scale in on Pullback

Conviction: Medium-Low. The HBM thesis is compelling, but the execution risks are significant: the stock has already priced in substantial growth after a 578% rally, management has diluted shareholders meaningfully, the memory cycle is volatile, and Korean competitors are gaining traction.

Why not buy now? At ¥8,490 (~28x post-dilution FY2026E P/E), too much good news is priced in. The risk/reward is unfavorable — you're paying for continued HBM growth, successful factory ramp, and no competitive losses all at once. If any of those assumptions fail, the downside is severe (60-70% drawdown in the bear case).

Entry strategy: Wait for a pullback to ¥5,500-6,500 (20-22x post-dilution FY2026E earnings), which could come from a sector rotation, memory cycle concerns, or general market weakness. The stock fell from ¥10,270 to ¥8,490 in a few weeks — it's volatile enough to create opportunities.

Alternative (aggressive) entry: If Q4 FY2026 results (May 2026) beat guidance and FY2027 guidance exceeds expectations, a starter position at current levels becomes more defensible — but only at 1-2% portfolio weight.

Parameter	Recommendation
Target Position Size	1-3% of portfolio (small-cap, cyclical, Standard Market)
Entry Strategy	Scale in: 40% at ¥6,500, 30% at ¥5,500, 30% on confirmed FY2027 guidance beat
Maximum Loss Tolerance	25% from average cost (stop-loss around ¥4,500-5,500)
Add Trigger	FY2027 guidance >¥32B revenue with 25%+ OP margin; or pullback to ¥5,500-6,500
Trim Trigger	Stock approaches ¥10,000+ without earnings upgrades; Korean competitor share gains visible in quarterly results
Review Schedule	After each quarterly earnings report
Expected Holding Period	12-24 months, re-evaluate at each earnings cycle

Rationale for small size: This is a sub-$1B market cap stock on the TSE Standard Market with 1 analyst, 15.8% fresh dilution, high cyclicality, customer concentration in 3 memory makers, and a discredited auditor. These are real risks that warrant a small, scaled position — not a full conviction bet.

Exit Criteria

Sell triggers (full exit):

• HBM capex guidance from Samsung/SK Hynix/Micron declines materially (demand plateau)
• Operating margins fall below 15% for two consecutive quarters (pricing pressure or volume loss)
• A major customer publicly switches to Korean competitor for next-gen HBM probe cards
• Auditor situation worsens — any accounting irregularity surfaces

Trim triggers (reduce position):

• Stock exceeds 35x forward P/E without corresponding earnings acceleration
• Stock approaches ¥10,000+ without FY2027 guidance upgrade
• Korean competitor share gains become visible in JEM's quarterly revenue trajectory
• New factory construction delays beyond 6 months

Take partial profits:

• At ¥10,000+ (if entered at ¥6,500 level, that's +54% — lock in some gains on a cyclical name)

Analyst & Institutional Sentiment

• Coverage: Extremely thin — only 1 analyst covers the stock with a "Buy" rating
• Target price: ¥5,000 (appears stale — stock at ¥8,490)
• Short interest: Not available (limited for TSE small caps)
• Institutional ownership: Low (~25-33%) — typical for TSE Standard Market small-cap
• SmartKarma published coverage ("More Money for More Probe Cards") analyzing the equity offering — suggests some buy-side attention
• Simply Wall St featured JEM as one of "three Japanese growth companies with high insider ownership"

The thin coverage is both an opportunity (under-researched = potential information edge) and a risk (limited institutional sponsorship, lower liquidity). Any new analyst initiation would increase institutional awareness.

Key Open Questions

1. Has JEM announced an auditor change following Hibiki's second FSA order? (Critical governance item)
2. What is the exact customer revenue breakdown? (Download the 有価証券報告書 from EDINET)
3. How much of Micron Hiroshima's probe card orders go to JEM vs. MJC?
4. What is the construction timeline for the new factory? When does it reach nameplate capacity?
5. Is JEM qualifying products for TSMC Kumamoto logic testing, or is the plant purely for memory?
6. What are the Korean competitors' (Korea Instrument, TSE) specific wins with Samsung/SK Hynix?
7. JEMCO domestic subsidiary — what does it actually do? Confirm function and any related-party role.
8. Okubo Kogyosha stock-lending arrangements — trace through subsequent large-shareholding filings.

Recent Developments (as of March 2026)

• Q3 FY2026 (Dec 2025 quarter): Revenue ¥8.35B, beating estimated ¥6.70B — a significant beat demonstrating accelerating demand. Net income ¥1.77B, up from ¥710M in Q2.
• Feb 25, 2026: Equity offering announced — 1.74M new shares + 260K overallotment for ~¥14B. Stock hit limit-down on the announcement.
• Feb 6, 2026: Upward earnings revision — full-year OP raised from ¥4.8B to ¥6.5B (+35.4%). Dividend raised from ¥60 to ¥80/share.
• Stock price trajectory: Ran from ¥1,252 (April 2025 low) to ¥10,270 (Feb 25, 2026 ATH) — a roughly 8x move. Pulled back ~17-30% from highs to ¥7,100-8,500 range as of March.
• 52-week range: ¥1,252 — ¥10,270 (extraordinary width reflecting cyclical recovery + Japan semi narrative).
• Next earnings: May 19, 2026 (FY2026 full-year results).

Filing & Disclosure Review

See 6855-filings for detailed quarterly and annual filing analysis.

Recent key disclosures:

Date	Filing Type	Summary
Feb 25, 2026	Equity Offering	1.74M new shares + 260K overallotment; ~¥14B proceeds for new factory; 15.8% dilution
Feb 6, 2026	Earnings Revision + Q3 Results	Revenue ¥20.675B (+40.3% YoY), OP ¥5.028B (+77.5%), full-year upward revision: OP ¥4.8B → ¥6.5B (+35.4%)
Feb 6, 2026	Dividend Revision	Annual dividend raised from ¥60 to ¥80/share

Red flag check: No restatements, no auditor changes, no material weakness, no going concern, no insider selling, no activist involvement.

Corporate Structure

Japan Electronic Materials Corp. (6855.T)
├── JEMCO Co., Ltd. (Japan) — function undisclosed
├── JEM America Corp. (Fremont, CA)
├── JEM Europe S.A.R.L. (Grenoble, France)
├── JEM Taiwan Probe Corp. (Chupei, Taiwan)
├── JEM Shanghai Co., Ltd. (Shanghai)
├── JEM (Shenzhen) Co., Ltd. (Shenzhen)
├── JEM (Hong Kong) Co., Ltd. (Hong Kong)
├── JEMCO Co., Ltd. (Hwaseong-si, Korea)
├── JEM (Thailand) Co., Ltd. (Chonburi)
└── JEM SE Asia Pte. Ltd. (Singapore)

Related but separate:
  Ohkubo Kogyosha Co., Ltd. — family holding company (3.37% shareholder)
    └── Controlled/co-controlled by Ohkubo family members

The structure is straightforward — parent with 10 wholly-owned subs organized by geography. No layered holding structures, no SPVs, no off-balance-sheet vehicles.

2026-05-02 — Stage 3 watch flag from weekly Japan semi screen

The 2026-04-25 weekly screen (KB/brief/weekly-semi-review.md) flagged JEM as the weakest held position structurally, despite +33% annual EPS revisions. Screen takeaway: "price below MA50, 35% off highs — structure breaking despite EPS up. Watch for Stage 3 confirmation."

Live datapoints (2026-05-02):

Indicator	Value	Read
Current price	¥6,920	—
52w high	¥10,270	-32.6% off high
52w low	¥1,832	+278% off low
Price vs MA50 (50-session approx)	+1.4% above	Screen said below; live is just above — at the line
Price vs MA200 (6mo proxy)	+29.3% above	Still bullish on long trend
TEL Q4 print 2026-04-30	+14.6% EPS beat	Mild positive read-through to probe-card demand

Interpretation. The screen's "P<MA50" call is right on the line — JEM is testing the 50-day, not decisively below. The substantive concern is the -32.6% drawdown from the ¥10,270 ATH despite +33% annual EPS revisions, which is a classic divergence (numbers up, price down) that resolves either by (a) price catching up if the numbers hold, or (b) numbers eventually catching down to price.

What would confirm Stage 3: decisive close below MA50 with volume + lower-low under prior swing low + MACD rolling negative. What would invalidate: reclaim of ¥7,500+ on volume into the next print, or a clear bullish reaction to TEL's beat.

Action: monitor — no thesis change, but position size should reflect that this is the weakest held structurally. Re-check after next earnings print or on confirmed break of MA50.

---

Sources

Company:

• JEM Official IR
• JEM Products — Probe Cards
• JEM Global Network

Financial Data:

• StockAnalysis — 6855
• MarketScreener — JEM
• Yahoo Finance — 6855.T
• IRBank — Officer/Financial Data
• Kabutan — 6855
• Alpha Spread — Valuation
• Morningstar — 6855

Regulatory & Governance:

• Hibiki Audit — FSA Orders
• SESC Enforcement Database
• JEM Annual Securities Reports (EDINET)

Industry & Competitive:

• Mordor Intelligence — Probe Card Market
• Variant Perceptions — JEM HBM Analysis
• SmartKarma: "Japan Electronic Materials — More Money for More Probe Cards"
• FormFactor Q4 2024 and Q2 2025 earnings
• Technoprobe CMD 2025 presentation
• MJC Medium-Term Business Plan FV26

Employee:

• OpenWork — JEM Reviews

---

Topics

• japan-semi
• probe-cards
• hbm
• semi-equipment
• 6855-filings