DEEP RESEARCH · ADTECHNOLOGY

ADTechnology (200710) Deep Dive: Through the Strategic Chasm Toward a 2026 Mass-Production Turnaround

Re-reading four years of losses as planned investment — the pivot from TSMC VCA to Samsung DSP and the R&D pipeline that powers the 2026 ramp

Published: 2025-11-15 · Design house / foundry ecosystem view · Original Naver Blog post

0. Bottom Line First

• Revenue collapsed from KRW 164.2B (2022) to KRW 100.2B (2023, −39%); operating losses of KRW 17.4B and 17.0B in 2023 and 2024, and another KRW 8.2B loss in H1 2025.
• 2026 consensus model: revenue KRW 289.3B (+57.6% YoY), operating profit KRW 4.2B (return to black) — conditional on the ramp pipeline executing.
• Korea's largest design-house engineering pool (384 staff) is both the source of recent losses and the operating leverage that turns the cycle.
• Valuation should be anchored to the NPV of post-2026 MP cash flows from the R&D pipeline, not to historical statements.

1. The 2020 Strategic Pivot: TSMC VCA → Samsung DSP

Official fact: ADT was previously a formal TSMC VCA partner with a stable revenue base. In 2020 it switched to Samsung Foundry's Design Solution Partner (DSP) program. CEO Park Joon-kyu publicly described the move as a response to "continuous pressure as a listed company to grow revenue and meet market demands."

Interpretation: Rather than remain "one of many" in TSMC's VCA ecosystem, ADT bet on becoming a flagship partner inside Samsung's — a high-risk/high-reward path to access Samsung's most advanced nodes (8nm, 5nm, 4nm, 3nm GAA) and to become Korea's #1 design house.

2. The Financial Chasm (2022–H1 2025)

The DSP transition came with immediate financial pain. The 2021 top line, which had exceeded KRW 300B, fell as legacy TSMC programs rolled off — and Samsung-based MP revenue had not yet started to fill the gap.

Interpretation: The 2023–2024 losses are best read as a "planned cost of internalization" — absorbing Samsung's PDK, design methodology and IP portfolio. The CEO's note that more than 20% of engineering resources were spent on Samsung Foundry platform infrastructure is consistent with that reading.

3. H2 2025 Inflection: NRE → MP

Design-house revenue typically splits into NRE (engineering services in the early phase) and MP (mass production). MP is "far" larger than NRE in dollars and is the core driver of profitability.

4. Growth Engine: the R&D Pipeline into 2026

A. Automotive — getting to 5nm first

• BOS Semiconductors 5nm NPU (Eagle-N): Samsung 5nm NPU targeting L4+ autonomy. MPW validation completed in 2024; "Eagle-N" samples shipped to multiple global automotive OEMs in September 2025; MP targeted for 2026. Tenstorrent-based chiplet architecture scales from 250 TOPS to 2,000+ TOPS.
• Ubitium (Germany) 5nm MPU (UB410): Joint development of a next-generation general-purpose MPU. ADT handles backend design and Samsung FinFET process tuning. Tape-out in H2 2025, MP in 2026 — targeted at on-device AI for robotics, autonomy and smart appliances.

B. AI / HPC — into the leading nodes

• Domestic S-co 4nm AI chip: MP targeted for 2026; one of the two engines of the 2026 swing to profit.
• U.S. V-co 4nm program: Award decision expected within 2025. A win would be the second-largest development award ever taken by a Korean design house — a litmus test for ADT competing head-to-head with TSMC's ecosystem (Alchip, GUC) in the U.S. AI market.
• 3nm GAA server project: Contract signed October 2023 with a "global enterprise." ADT designs not only the SoC but also the HBM interposer — turnkey delivery. Signals that ADT is becoming a key design partner for the Samsung SAINT / I-Cube advanced packaging ecosystem competing with TSMC CoWoS.

C. ADP (Advanced Design Platform) — proprietary platform

ADP500 on Samsung 5nm pre-integrates and pre-validates major IP (Arm Cortex-A53 subsystem, LPDDR4, PCIe) as a customer-ready architecture. It compresses time-to-market for fabless clients, and is evidence that the design-house industry is moving from a pure service business toward validated platform solutions.

Table 2 — Key R&D pipeline (2025–2026)

5. Competitive set: 384 engineers as design capacity

Skilled engineers are the design house's core asset. Based on the SemiFive IPO filing, ADT's headcount at the end of H1 2025 was 384 — larger than SemiFive (357), Gaonchips (270) and Asicland (244), the largest pool in Korea.

Interpretation: 384 engineers means ADT is the only Korean design house that can run BOS (5nm), Ubitium (5nm), S-co (4nm) and G-co (3nm GAA) in parallel. Gaonchips ran a small OP profit (KRW 3.5B) but its leading-node exposure looks smaller; Asicland's identity remains TSMC VCA. Inside Samsung's DSP camp, ADT is the partner whose "All-in" loyalty has already been tested.

6. Risks

• Liquidity: H1 2025 operating cash flow was KRW −12.6B. A ramp slippage could require new funding.
• Project concentration: The 2026 swing rests on a handful of programs (BOS, S-co 4nm). Automotive qualification delays or AI-chip competitive loss can rewrite the whole scenario.
• Samsung Foundry dependence: ADT's fate is tied to Samsung's 5/4/3nm competitiveness. If yield trails TSMC, fabless client products lose competitiveness — and MP volume directly with them.

7. 2025–2026 Catalysts to Watch

1. Short term: Q3 / Q4 2025 results — does first MP revenue land meaningfully in the P&L; does the NRE:MP mix start shifting?
2. Medium term: U.S. V-co 4nm award announcement — the strongest single signal that ADT has entered the U.S. AI market.
3. Long term: BOS Semiconductors commercialization — when global OEMs publish 2026 vehicle design wins.

8. Bottom Line: Reframing ADT as a "Keystone Partner"