Korean Memory ATE Stocks: Neosem, Exicon and OpenEdges Re-Ranked

Sat, 25 Apr 2026 18:00:00 +0900

SemiScope Part 1. Three Korean back-end test & IP names — Neosem (253590 KQ), Exicon (092870 KQ), OpenEdge Technology (394280 KQ) — re-ranked by timeframe rather than by a single CXL-purity prism. The conclusion that the consensus comparison reaches is internally consistent but reaches it through one lens; once you swap lenses, the rank order changes for the most actionable horizon.

TL;DR

Short-term momentum (the 2026 earnings inflection): Exicon > Neosem > OpenEdge. The consensus comp note that flags Exicon as an afterthought is reasoning purely through a CXL-purity prism. On a CLT-thesis prism — single-source DRAM low-frequency tester with Samsung’s final qualification cleared in late 2024 — Exicon is the cleanest J-curve entry of the three.
The 2024-2025 loss / revenue-decline reset is the most under-reflected fact in the consensus comp. Neosem revenue fell 39.3% YoY in 2025 with operating margin compressed to 6.4%. Exicon ran a KRW 15.9B operating loss in 2024. OpenEdge’s 3Q YTD operating loss of KRW 21.3B is widening, not shrinking. All three are in turnaround positioning — that means volatility before the inflection.
Long-term IP platform value (2027-2029): OpenEdge > Neosem > Exicon. Here the consensus is right. The caveat is that OpenEdge’s royalty mix at 0.4% of revenue means the J-curve inflection sits further out than the framing implies.

Bottom line — by timeframe, not by ranking:

Name	Position type	Entry timing	Key event
Exicon	Near-term J-curve bet (CLT volume-revenue recognition)	Now	1Q26 swing-to-profit stability, cumulative CLT install base
Neosem	Cycle turnaround + Gen6 / CXL 3.1 sole-source bet	After 1Q26 print	1Q26 revenue recovery + first CXL 3.1 mass-production order
OpenEdge	Long-dated IP-platform option (3-5 yr)	Scaled accumulation	First LPDDR6 license disclosure + royalty mix reaches 5%

The “Neosem = Core / OpenEdge = option / Exicon = afterthought” conclusion is coherent only under a single CXL-purity prism. Through a 2026 earnings momentum + sole-source moat prism, Exicon is the most attractive near-term name — the central re-evaluation of this report.

1. What the Consensus Comp Missed

The consensus comparison doc treats one-axis prism (CXL substitutability) as the binding constraint. Three under-weighted facts shift the picture.

1-1. Neosem (253590 KQ)

Item	Consensus comp	SemiScope addition
2025 results	Not stated explicitly	Revenue KRW 63.9B (-39.3% YoY), operating profit KRW 4.1B (-75.3% YoY), OPM 6.4% — a meaningful self-cycle slowdown
Advantest SSD ATE exit	Not mentioned	Advantest withdrew from SSD ATE in Jan 2025 → Gen5/Gen6 SSD Tester is effectively a sole-source structure for Neosem
Nov 2025 IR guidance	Not mentioned	Order momentum surged from late August. Sep-Nov bookings ran ~2x the Jan-Aug pace. Company guides to all-time-high 2026 results
HBM back-end entry	Not mentioned	HBM-dedicated BX burn-in tester is “under consideration” only — Techwing and Advantest have already moved. Late-mover risk

1-2. Exicon (092870 KQ)

Item	Consensus comp	SemiScope addition
2024 results	Not mentioned	Revenue KRW 31.6B (-61.6% YoY), operating loss KRW -15.9B — starting point is a deep cyclical trough
CLT sole supply	Not mentioned	Samsung cleared final CLT qualification in late 2024; single supplier of LF tester from 2025 — much more concrete than the comp’s vague “Samsung partnership” framing
Cumulative orders (Oct 2025-Mar 2026)	Not mentioned	KRW 48.8B (CLT/Burn-in) + KRW 30.2B (SSD/CLT integrated) ≈ KRW 79B — about 2.5x of 2024 revenue, in five months of bookings
2026 guidance	“Afterthought until Samsung CXL CapEx confirmed”	Company guides to first KRW 100B+ revenue year and full swing-to-profit in 2026
CLT technology edge	Not mentioned	11,520-parallel processing (vs ~500-parallel legacy), at least a 2-year technology lead vs competitors

If you fold these in, Exicon stops being a “Samsung-CXL-dependent afterthought” and becomes “single-source CLT supplier entering its J-curve.” The consensus comp’s conclusion needs partial correction.

1-3. OpenEdge Technology (394280 KQ)

Item	Consensus comp	SemiScope addition
2024-2025 results	“Deepest moat but low CXL revenue visibility”	2024 revenue -21.8%, operating loss KRW -24.3B (loss widened); 3Q YTD 2025 cumulative loss KRW -21.3B — the “long-dated option” framing should be read more conservatively
License ASP	Not mentioned	Avg ~$1.1M per deal, up sharply from prior quarter ~$0.7M — leading-node IP mix expanding
LPDDR6 + Samsung 4nm	Not mentioned	Core new IP under multi-customer engagement in 2025
UCIe IP	“Under development”	Controller IP development complete and in validation; PHY in design start — more specific stage than the comp implies
Royalty revenue	“Low CXL visibility”	1H 2025 royalty mix 0.4% (KRW 0.029B) — given the lag from license to volume, the J-curve inflection point likely sits in 2026-2027 or later
Break-even timing	Not mentioned	Korea Investment models BEP in 2026 vs Shinhan models +87.5% revenue growth in 2025 — the consensus dispersion is unusually wide

2. Critical Re-Evaluation of the Consensus Comp

2-1. “OpenEdge has the deepest technology moat”

Partial agreement. The design-in switching cost on hard IP and the Memory Subsystem integration capability inside Korea is effectively monopolistic, and that part is fine. But against Synopsys and Cadence on a global comparison the gap is meaningful, and silicon-proven track record at leading nodes (N3/N2) is well behind.

“Deepest moat” should be read as “deepest moat within Korea.” OpenEdge’s moat reads less as a global multi-bagger candidate and more as a quality compounder anchored to the Korean AI/ASIC ecosystem.

2-2. “Neosem is the most genuinely irreplaceable name in CXL”

Agreed, conditionally. First-mover on CXL 1.1/2.0 plus the device-validation history with Samsung is the strongest single asset right now. But the comp under-weights the probability that CXL 3.1 introduces a dual-sourcing policy.

Implication: in a CXL 3.1 50:50 split scenario between Exicon and Neosem, Neosem’s “irreplaceability premium” gets partially diluted.

2-3. “Exicon’s Samsung relationship is strong but CXL purity and lead are weak — afterthought”

Partial agreement → active re-interpretation. The CXL-purity statement is factually correct, but this report views CLT — single-source, 11,520-parallel capacity, Samsung final-qual cleared in late 2024 — as forming a J-curve thesis independent of CXL as a Samsung DRAM back-end efficiency tool.

Through the CXL prism, Exicon is an afterthought. Through the DRAM back-end ATE prism, Exicon has the strongest near-term momentum. Both prisms are operating simultaneously, so Exicon should re-rank one tier higher than the consensus places it.

2-4. “Asymmetry: Neosem most balanced, OpenEdge long option, Exicon afterthought”

The framing collapses time horizons. Disaggregated:

Short term (6-12 months) momentum: Exicon > Neosem > OpenEdge. Exicon’s KRW 79B cumulative bookings + KRW 100B revenue guidance is the strongest signal.
Mid term (1-2 years) CXL cycle: Neosem > Exicon > OpenEdge. Agree with the consensus.
Long term (3-5 years) IP platform: OpenEdge > Neosem > Exicon. Agree with the consensus.

3. Three-Name Comparison Table

3-1. Financials

Item	Neosem	Exicon	OpenEdge
2024 revenue (KRW B)	105.2	31.6	15.3
2025E revenue (KRW B, prelim)	63.9 (-39.3%)	~60-70 (rebounding)	Wide range (Shinhan KRW 40.4B vs KI conservative)
2024 OP (KRW B)	16.5	-15.9	-24.3
2025E OP	4.1 (prelim)	Swing-to-profit attempt (1H -8.6)	Sustained loss (3Q YTD -21.3)
2026E guidance	“All-time-high revenue”	“KRW 100B+ revenue, swing to profit”	BEP (Korea Investment model)
Loss exposure	Stays profitable, but OPM compressed to 6.4%	2024 deep loss → 2026 recovery	Chronic R&D losses → long-dated BEP

3-2. Business & technology

Item	Neosem	Exicon	OpenEdge
Business model	Equipment (ATE)	Equipment (ATE)	IP / EDA
Core sole-source position	Gen5 SSD Tester (Advantest exited)	CLT (11,520-parallel, Samsung qualified)	LPDDR5X / LPDDR6 IP (within Korea)
First-mover area	CXL 1.1 / 2.0 mass-production tester	CLT (LF memory tester)	NPU + Memory Subsystem integrated IP
Late-mover area	HBM inspection equipment	CXL segment	Leading-node (N3/N2) memory IP
Revenue-recognition lag	Short (6-9 months from order)	Short (6-9 months from order)	Long (license recognized immediately, royalty 2-4 years out)
Customer mix	Samsung, SK hynix, Micron	Samsung (dominant)	Diversified globally and in Korea

3-3. Trend exposure summary

Trend	Neosem	Exicon	OpenEdge
Gen6 SSD (PCIe 6.0)	Strong tailwind	Tailwind	Neutral
CXL 2.0 → 3.1	Strong tailwind	Tailwind	Tailwind (indirect)
HBM4 / HBM4E	Neutral	Tailwind (indirect)	Neutral
LPDDR6	Neutral	Neutral	Strong tailwind
AI-inference ASIC diversification	Tailwind	Tailwind	Strong tailwind
Chiplet (UCIe)	Neutral	Neutral	Tailwind → Strong
Automotive ADAS ASIC	Neutral	Neutral	Strong tailwind
Memory CapEx cycle	High beta	Highest beta	Low beta

4. Scenario Analysis (All Three, Same Lens)

Scenario A — Memory CapEx normalizes + CXL 3.1 ramps (Bull)

Korean memory CapEx normalizes in 2026 → back-end ATE order cycle recovers.
Samsung CXL 3.1 line goes live; dual-sourcing policy splits Neosem and Exicon.
Beneficiary intensity: Exicon ≥ Neosem > OpenEdge — equipment names track cycle directly; OpenEdge is lagged by IP-to-volume conversion.

Scenario B — AI inference ASIC + automotive ASIC explosion (Bull)

Design-house diversification accelerates across AI inference and ADAS.
LPDDR6 + UCIe adoption accelerates → OpenEdge license ASP compounds.
Beneficiary intensity: OpenEdge > Neosem > Exicon — OpenEdge is the direct IP-diversification beneficiary; equipment names benefit only indirectly.

Scenario C — Memory CapEx recovery delayed + R&D burn accelerates (Bear)

Samsung CapEx slows, memory pricing stays soft.
All three stay under R&D pressure with poor revenue visibility.
Damage intensity: Neosem > Exicon ≥ OpenEdge — Neosem most exposed to cycle recovery; OpenEdge already in loss so marginal incremental damage is bounded.

Scenario D — Global IP / ATE leaders accelerate Korea entry (Bear)

Synopsys / Cadence expand share in Korean leading-node IP; Advantest re-enters SSD ATE.
Damage intensity: OpenEdge > Neosem > Exicon — OpenEdge most exposed to global competition; Exicon’s CLT 11,520-parallel barrier holds the longest.

5. Action Guide for Allocators

5-1. Per-name action map

Name	Action	Buy trigger	Sell / trim trigger
Exicon	Near-term J-curve bet / Core candidate	1Q26 swing-to-profit stability confirmed	Signal that CLT sole-source ends (competitor entry) / Samsung DRAM CapEx slows
Neosem	Enter after 1Q26 print / cycle-turnaround bet	1Q26 revenue recovers above KRW 20B + first CXL 3.1 mass-production order	HBM inspection-tool entry fails + dual-sourcing signal in Gen6
OpenEdge	Scaled accumulation / option position	First LPDDR6 + Samsung 4nm license disclosed + royalty mix reaches 5%	1H 2026 loss widens further + R&D headcount cost steps from KRW 11B → KRW 13B+

5-2. Where SemiScope diverges from the consensus comp

Item	Consensus comp	SemiScope (re-rank)
Core #1	Neosem	Exicon (near-term momentum) — or Neosem after 1Q26 print
Long option	OpenEdge	OpenEdge (agree)
Afterthought	Exicon	Exicon is in fact the #1 candidate on near-term momentum (re-evaluated outside the CXL prism)

6. Final Note (Allocator’s Frame)

The consensus comp ranks the three names through a single prism — CXL substitutability — and through that prism the conclusion (Neosem = Core, OpenEdge = option, Exicon = afterthought) is internally coherent. But once you fold in the 2026 earnings data and the order-disclosure flow, the picture shifts.

Exicon, viewed outside the CXL prism, is the strongest near-term momentum name in this trio. Single-source CLT, Samsung final-qual cleared in late 2024, KRW 79B cumulative bookings, and KRW 100B+ revenue guidance form a clean J-curve thesis. Low CXL exposure is not a weakness here — it is the strength of an independent thesis built on Samsung DRAM back-end efficiency.
Neosem is a name where consensus has already reset once on the 2025 revenue collapse. 1Q26 is the first revenue-recovery confirmation print; before that, this remains a cycle bet. The first CXL 3.1 mass-production disclosure is the entry-confirmation signal.
OpenEdge — agree with the consensus framing, but the 2025 widened loss and 0.4% royalty mix imply the J-curve inflection sits further out than the framing suggests. The 2026 BEP-or-not print is the watershed for the long-term valuation.

Final recommendation: Rather than collapsing the three names into a single 1-2-3 ranking, run them as time-horizon-separated bets. Exicon = short term, Neosem = mid term, OpenEdge = long-dated option. Layered exposure by timeframe delivers the best risk-adjusted setup of the available framings.

Appendix — Exicon Deep Profile

Snapshot

Core products: CLT (Chambered Low-frequency memory Tester) at 11,520 parallel — next-gen DRAM stability tester; Gen5 SSD Tester (Samsung-supplied); Burn-in Tester; CXL 2.0 Tester; non-memory testers (SoC / CIS / DDI).
Margins / track record: 2024 revenue KRW 31.6B / operating loss KRW -15.9B (loss-making). 1H 2025 cumulative operating loss KRW -8.6B. Cumulative new orders Oct 2025 - Mar 2026 ≈ KRW 79B (CLT / Burn-in / SSD). Both company and sell-side guide to first KRW 100B+ revenue year + full swing-to-profit in 2026.
Technology moat: CLT 11,520-parallel processing (~20x prior-gen 500-parallel). Samsung final qualification cleared in late 2024, sole supplier of LF tester since 2025; estimated technology lead at least 2 years vs competitors.
Investment view: Event-driven turnaround cycle name. Exited the 2024 trough via single-source CLT and large-order J-curve entry. Highest exposure to Samsung DRAM CapEx among Korean back-end ATE names.

Customer landscape

Stage	Customer	Product / process	Status
Production	Samsung	CLT (DRAM LF tester, sole supplier)	In production. KRW 20.4B + KRW 19.6B + KRW 8.8B = KRW 48.8B total disclosed in Oct-Nov 2025
Production	Samsung	CLT + SSD Tester combined supply	Additional KRW 30.2B disclosed in Mar 2026 (~95.5% of latest reported revenue)
Production	Samsung	Gen5 SSD Tester	In volume
Production	Samsung	Burn-in Tester	In volume
In progress	Samsung	CXL 2.0 Tester (Neosem dual-source potential)	In development + adoption push
In progress	Samsung	Gen6 SSD Tester	In R&D
In progress	Samsung S.LSI	DDI Tester, AP Tester	In development off CIS-tester base
Potential	SK hynix	Memory Tester	Limited share, expansion under exploration

Quantum-jump triggers

Trigger #1 — CLT volume cycle entry + 20+ cumulative installs. Per company guidance, hitting ~20 cumulative CLT installs at Samsung DRAM lines in 2026 supports KRW 100B+ revenue. 2024 revenue KRW 31.6B → 2026 KRW 100B+ implies ~3.2x topline + swing-to-profit.
Trigger #2 — CXL 3.1 dual-vendor entry. If Samsung adopts a dual-sourcing policy on CXL 3.1, Exicon could capture ~50% of equipment supply alongside Neosem. CXL ASP runs ~2-3x SSD Tester levels, implying meaningful incremental revenue.
Trigger #3 — Gen6 SSD Tester R&D → volume. R&D-line Gen6 SSD Tester supply could materialize in 2026, with volume entry in 2027. Risk: Neosem retains pole position given Advantest exit.
Trigger #4 — Non-memory (SoC / AP / DDI) tester volume recognition. CIS-tester-derived DDI / AP testers winning a Samsung S.LSI or external fabless design-in would meaningfully de-risk the cycle dependency.

Risks & watchlist

Single-customer concentration. Revenue is overwhelmingly Samsung-dependent — 2024’s -61.6% revenue decline is the cleanest illustration. Any Samsung CapEx slowdown flows directly into a single quarter.
CLT competitor catch-up. No competitor is currently visible at 11,520-parallel chamber engineering, but a 2-year window is not infinite; monitor Dlight / UnTest Sci (and others) for entry signals.
CXL 3.1 first-mover not secured. Neosem retains the lead in CXL. If Exicon stays second-source through CXL 3.1, the CXL-cycle alpha is capped.

Five checkpoints for the next 1-2 quarters

1Q26 print — magnitude of the swing to profit. How much of the KRW 79B order book recognizes as revenue and operating profit. The cleanest stability check.
Cumulative CLT install count for 2026. Company guides to ~20 units. Quarterly tracking of whether installs follow the trajectory.
CXL 3.1 Tester Samsung qualification disclosure timing. Most likely 1H 2026. Pass = dual-source seat secured. Miss = no CXL-cycle participation.
HBM-line entry for inspection equipment. Whether HBM4E back-end-validation burden flows into CLT / Burn-in demand. A formal HBM-spec disclosure from the company would be the meaningful signal.
Non-memory (DDI / AP) tester revenue disclosure. A separate disclosure with the line item exceeding 10% of revenue is the quantitative diversification signal.

Disclaimer: For research and information purposes only. Not investment advice. Names cited are for analytical illustration; readers should perform their own due diligence and consult licensed advisors before any investment decision.

DRAM ATE on Korea Invest Insights