AD Technology _ A Samsung Foundry Design Partner

 

AD Technology (KOSDAQ: 200710): A Samsung Foundry Design Partner Two-Thirds of the Way to Break-Even

Revenue direction turned for the first time in five years; the profit sign has not

DRAFT — NOT FOR PUBLICATION. This English edition was written under an agreed one-off procedural exception. The Korean edition has not cleared its publication gate, and the English edition is authorised for drafting only. It must not be posted, distributed, or shared outside the working file set.

Written August 27, 2026

Filings cut-off August 26, 2026

Currency Korean won (KRW), consolidated unless stated otherwise

KOSDAQ is Korea's secondary market for smaller and growth companies, comparable to the Nasdaq. DART is Korea's regulatory filing system, the equivalent of the SEC's EDGAR.

 

What Is Not Yet Settled — Stated Up Front

ⓐ What is unsettled

① In projecting 2028 revenue, the author applies the following two calculations the scheduled completion date (July 30, 2027) of one large contract that currently supports revenue. This piece carries one of them.

② There is a wide gap between the 2026 revenue figure put forward by Korean sell-side analysts (KRW 256–289bn) and my own estimate (KRW 187bn). Where that gap comes from, I could not establish within my source material.

③ No peer-comparison reference table exists yet for this industry — a table of historical figures for companies in the same business. Without it, I could not run the step of measuring this company against its peer group a second time.

ⓑ Why they are unsettled

For ①, the volume that follows the contract's completion has not yet appeared in filings. For ②, the basis for the analysts' estimate is not laid out in public material. For ③, the peer reference table has not been built.

ⓒ Effect on the conclusion (direction and size)

●       ① separates the 2028 revenue projection by KRW 112bn between the two calculations. Which one holds depends on the disclosed value of follow-on contracts.

●       ② is a case where my estimate may have been the smaller one. If the analysts' figure holds, the sign of 2026 operating profit turns from negative to positive.

●       ③ does not change the direction of my conclusion. This piece carries business facts only and does not carry a value for the company. It is, however, one of the reasons I decline to carry one.

ⓓ What would resolve them, and when

① Disclosure of a new or expanded supply contract of KRW 10bn or more — any time (via DART)

② Whether Q3 quarterly revenue exceeds KRW 64.5bn — November 14, 2026

③ Construction of the peer reference table — next quarterly review

 

0. The Company in Thirty Seconds

AD Technology designs semiconductors on behalf of companies that want to sell them, and — where the customer asks for it — delivers the finished chip as well. The market it sits in is growing as artificial intelligence (AI) datacentres expand, and the two companies it genuinely competes with at home are ASICLAND and Gaonchips.

Where the company is headed — It stands at a point where annual revenue of roughly KRW 258bn would begin to leave profit for shareholders. First-half 2026 revenue was KRW 82bn, about KRW 41bn per quarter. What would change the picture is specific: quarterly revenue above KRW 64.5bn, and disclosure of the volume that follows the two contracts ending in the first half of 2027.

Structural observations — First, the revenue direction turned for the first time in five years: organic growth (measured against the same period a year earlier, with no change in the consolidation perimeter) was +10.4% in Q1 2026 and +35.0% in Q2 2026, two consecutive positive quarters. Second, design-service exports came through as actual revenue — KRW 6.9bn in 2024, KRW 19.0bn in 2025, and KRW 40.2bn in the first half of 2026.

What is confirmed / what is not

●       Confirmed — its standing as an official Samsung Foundry design partner, the turn in revenue direction, and the realisation of export revenue.

●       Not yet confirmed — the sign of profit, and the volume that follows the two contracts (together KRW 75bn) ending in the first half of 2027. Over the past five years the company spent KRW 51.5bn more than it took in as cash; whether that money comes back is also still open.

 

1. The Company — When It Began and What It Has Done

AD Technology was founded in 2002 as a semiconductor design specialist. It started as a design partner (VCA) to Taiwan's TSMC and became an official design solution partner (DSP) to Samsung Foundry in 2020. In 2023 it was selected as an Arm Total Design partner.

The company is organised into three units: an SoC unit handling physical design, a platform unit that builds design frameworks, and an infrastructure unit for foundation design assets. For accounting purposes, however, reporting is consolidated into a single segment plus "other".

Item	As of June 30, 2026
Employees	877 consolidated (726 in R&D); 443 parent-only (365 in R&D)
Registered domestic patents	19
Consolidated subsidiaries	11, plus 2 equity-method affiliates
Shares outstanding	13,462,007 (treasury 486,181, or 3.61%)
Fiscal year end	December

 

Source — 2026 half-year report (filed with DART, August 2026), Section on shares and employees; 2025 annual report (filed with DART, March 2026).

 

2. The Core Technology — What It Makes and What It Does In Place Of Others

This company produces the design of an application-specific integrated circuit (ASIC). In doing so, it takes over work the customer struggles to do alone: turning a design concept into a form a fabrication plant can actually print.

A company that wants to sell a chip (a fabless firm) often knows what the chip should do but cannot turn that into a manufacturable layout on a leading-edge process. That translation is what a design house provides, and it is where this company sits. Where the customer wants it, the company also bundles wafer production, packaging, and test into a finished-product delivery (a turnkey arrangement).

Three things make this difficult to copy.

What	Detail
Experience with two foundries	It has worked with both Samsung Foundry and TSMC process lines
Owned design assets	It holds its own memory and foundation design assets, with 19 registered domestic patents
People	726 R&D engineers — though those engineers are also its fixed cost

 

2-2. Pipeline

The largest development programme confirmed in filings is a high-performance server chip built on a Samsung leading-edge process with an Arm server core design. It runs from November 2024 to July 2028.

Source — 2025 annual report (filed with DART, March 2026), business overview section.

 

3. The Business Model — How It Makes Money

3-1. What it sells

Two things: design services and production. The service share has climbed steadily — 77% in 2024, 85% in 2025, and 90% in the first half of 2026.

Revenue mix, H1 2026	What it is	Share	How revenue arrives
Services — HPC	Design of high-performance datacentre and server chips	36.2%	Project-based (percentage-of-completion)
Services — AI	Design of AI accelerators and neural processors	21.1%	Project-based (percentage-of-completion)
Services — other	Big data, mobile, and others	25.3%	Project-based (percentage-of-completion)
Services — automotive	Design of automotive semiconductors	7.3%	Project-based (percentage-of-completion)
Products	Finished-chip delivery	7.8%	On delivery (recurring)
Other income	—	2.4%	—

 

3-2. Who it sells to

Customers are businesses, and their names are withheld under non-disclosure agreements. That is something the company does not disclose. Supply-contract filings have, however, revealed several engagements.

Engagement	Contract value	Period
Automotive semiconductor customer	KRW 35.2bn	September 2023 – February 2027
Design software customer	Not disclosed	August 2023 – November 2026
Undisclosed counterparty	KRW 39.6bn	March 2026 – May 2027
Storage controller customer	KRW 12.1bn	Completed
Server-side customer	KRW 22.4bn	Completed

 

Four customers each accounting for 10% or more of revenue together made up 69.2% of first-half 2026 revenue. Concentration is high.

Source — supply-contract disclosures (filed with DART, April 2026 and June 2026); 2026 half-year report (filed with DART, August 2026).

3-3. How it gets paid

Service contracts are paid in instalments — on signing, at milestones, and on completion. Revenue is recognised in proportion to how far the work has progressed (percentage-of-completion). Progress is measured by dividing costs incurred to date by total expected costs (the input method).

One point deserves attention. Total expected cost is an estimate the company makes, and when that estimate changes, profit already recognised changes with it. In 2025 such changes added KRW 2.0bn to profit; in the first half of 2026 they added KRW 7.8bn. The company states in its filings that the same amount will be reversed with the opposite sign in future periods. The KRW 7.8bn added to current results will therefore be taken back out later.

Source — 2026 half-year report (filed with DART, August 2026), Note 32.

3-4. Cost structure

For the first half of 2026, the two largest components of cost of sales were as follows.

Item	Amount	As % of revenue
Outsourced processing	KRW 20.1bn	24.5%
Raw materials and design-asset purchases	KRW 21.2bn	25.9%

 

Together they account for roughly half of revenue leaving the company. Half of the value this business creates goes to suppliers and design-asset vendors.

3-5. Pricing and scale

This company sits in a position where its power to set its own price is limited. Customers sit in front of it and fabrication plants and design-asset vendors sit behind it, which constrains what it can set in the middle. The cost structure above shows that constraint in numbers.

What it gains from scale is a step change in profit. Fixed costs stay flat while revenue grows, so profit rises in steps rather than in line — operating leverage. The 726 R&D engineers and the selling and administrative expense base go out every year regardless of revenue. Where that step sits is set out in numbers in Section 7.

 

4. The Market — Is the Ground It Stands On Growing?

4-1. Size and current position

The size of the global market cannot be stated as a single number. The nine research houses I checked put this market at anywhere between $160m and $18.9bn for 2025–26, with compound annual growth rates ranging from 5.24% to 15.43%. What each includes and excludes differs. The figures the company cites — a $775bn global semiconductor market growing at 22% a year in non-memory — are a McKinsey projection, and appear here as company-stated material.

I therefore use the combined disclosed revenue of the three Korean companies in this business as the coordinate I can actually measure.

KRW bn	H1 2025	H1 2026	Year on year
Three-company total	121.5	237.6	+95.6%
AD Technology's share	54.9%	34.5%	−20.4pp

 

The three are AD Technology, ASICLAND, and Gaonchips. Others in adjacent businesses are not included.

Source — half-year reports of each company (filed with DART, August 2026).

The market nearly doubled in a year, and this company's share of it fell. That said, this is a single half-year observation; whether it continues is not yet known.

4-2. What is driving the market

Driver	Size	How far along	Basis
① Demand for AI and high-performance computing chip design	Large	Under way — AI and HPC rose from 31% of this company's revenue in 2024 to 38% in 2025 and 57% in H1 2026	Filed data
② Expansion of Samsung's leading-edge process	Large	Under way — the server programme runs to July 2028	Filings
③ Widening overseas customer base	Moderate	Under way — service exports rose from KRW 6.9bn (2024) to KRW 19.0bn (2025) to KRW 40.2bn (H1 2026)	Filed data

 

One further note on AI. AI has two faces for this company. One expands the market; the other could automate design work and lower the barrier to entry. On the filed evidence so far the first face dominates, and the second has not been observed. I have counted AI once, on the market side only.

4-3. Ceiling and headwinds

●       Ceiling — the production slots foundries allocate and the number of design engineers available set how much work the company can absorb in a year.

●       Currency — with exports at 49% of total revenue, the direction of currency exposure has reversed. On half-year report figures, a 10% rise in the US dollar reduces pre-tax profit by KRW 2.3bn. In the prior half, the same move added KRW 0.9bn.

 

5. Competition — Who It Faces and On What

5-1. Competitors by layer

This is an oligopoly of a small number of firms, and the barriers to entry are leading-edge process experience, foundry partnerships, and design headcount.

H1 2026, KRW bn	AD Technology	ASICLAND	Gaonchips
Revenue	82.0	113.7	41.9
Operating profit	−1.5	+5.0	−3.6
Gross margin	16.6%	I could not verify	I could not verify
SG&A ratio	18.5%	11.7%	I could not verify
Principal foundry	Samsung	TSMC	Samsung

 

On first-half 2026 figures, this company ranks second of the three domestically in both revenue and operating profit. Its gross margin, at 16.6%, is the highest of the three. The order reverses at the SG&A line — the margin it earns is the widest, but the cost of running the company is larger than its peer's.

Abroad there are two substantially larger design houses, GUC and Alchip. I did not reconcile their financials within the source material for this piece.

5-2. Substitutes

The largest substitute is the customer doing the design itself. As a customer's internal design organisation grows, the work it hands out shrinks. After that come the large overseas design houses.

The two domestic firms named in Section 0 sit on a different layer from these substitutes: with the domestic two, this company divides today's orders; a substitute can shrink the market itself.

5-3. Defences and open questions

●       Defences — its standing as an official Samsung Foundry design partner and an Arm Total Design partner, and a record of having actually brought chips through leading-edge processes.

●       Open — that standing has opened the door to orders but not yet the door to profit. Contracts continued through the first half of 2026, yet operating profit was −KRW 1.5bn, and its domestic share fell 20.4 percentage points while it held the same credentials.

 

6. The Moat — How Long Can It Hold This Position?

A "moat" here means something others cannot easily copy, which lets a position last. I checked six sources one by one.

Source	Present	Basis
Technology and patents	Yes	Leading-edge design work, owned memory and foundation design assets, 19 registered patents
Switching cost	Yes	Once a development programme starts, changing design houses mid-stream is difficult
Credentials and entry barriers	Yes	Official Samsung Foundry design partner; Arm Total Design partner
Economies of scale	Partial	Headcount and track record have accumulated, but the company is still mid-sized
Brand and reputation	Partial	A record on leading-edge programmes is still building
Network effects	No	—

 

In sum — I read the moat as present but thin. The credentials are a barrier but not exclusivity (three firms domestically, two large ones abroad), and the domestic share fell while those credentials were held.

My analysis begins by multiplying two things — the company's own strength and the change in its environment. If either is zero, the whole is zero. Neither was zero here, so the analysis continued. What I watch is that one of the two terms — the company's own strength — keeps thinning.

What would weaken the moat — a decline in Samsung Foundry's leading-edge competitiveness, or the departure of a customer that accounts for a large share of revenue.

 

7. Financials — The Body As the Numbers Show It

7-1. Summary financials (reported)

KRW bn	2023	2024	2025	H1 2026
Revenue	100.2	106.5	164.5	82.0
Gross margin	12.4%	11.0%	19.1%	16.6%
Operating profit	−17.4	−17.0	+2.9	−1.5
Net financial income	+2.2	+2.7	+1.5	−2.5
Profit attributable to shareholders	−16.0	−14.3	+3.6	−6.2

 

Source — 2025 annual report (filed with DART, March 2026) and 2026 half-year report (filed with DART, August 2026), consolidated statements.

Quarterly figures show the texture of the business more clearly.

KRW bn	Q1 25	Q2 25	Q3 25	Q4 25	Q1 26	Q2 26
Services — HPC	4.9	8.9	5.3	2.2	16.3	13.4
Services — AI	8.5	5.8	14.6	12.8	3.4	13.8
Services — automotive	1.7	2.4	2.7	25.2	1.7	4.3
Services — other	3.9	4.5	7.8	8.7	5.1	7.4
Products	5.0	5.8	4.2	5.4	3.5	2.8
Other income	1.2	1.2	2.0	6.3	1.0	0.9
Total revenue	32.6	34.0	36.8	61.0	36.0	45.9

 

Quarters swing widely, because revenue clusters according to when projects progress.

7-2. History of losses and their character

2023 and 2024 were loss-making at the operating line. In 2025 operating profit turned positive at KRW 2.9bn and KRW 3.6bn was left for shareholders. In the first half of 2026, however, operating profit was −KRW 1.5bn and the shareholders' share −KRW 6.2bn.

One more line changed sign. Net financial income went from +KRW 1.5bn in 2025 to −KRW 2.5bn in the first half of 2026, because of the convertible bond described below.

7-3. How long the cash on hand lasts, annual cash burn and its composition

Item	As of June 30, 2026
Cash and cash equivalents	KRW 70.5bn
Short-term financial instruments	KRW 62.6bn
Total	KRW 133.1bn (~$95m)
Quarterly cash burn	About KRW 18.0bn ((half-year operating cash −KRW 9.7bn + property and intangible purchases KRW 26.3bn) ÷ 2)
How long the cash on hand lasts	About 7.4 quarters
Borrowings + convertible bond (book)	KRW 84.8bn + KRW 91.9bn
Debt-to-equity ratio	133%

 

On burn rate alone there is no immediate funding pressure. Over the past five years, however, cash generated from operations less spending on facilities and design assets came to −KRW 51.5bn. Profit attributable to shareholders over the same five years sums to −KRW 5.6bn.

Cash leaves in three places.

Where	Amount	What is confirmed so far
Construction in progress	KRW 27.9bn (end-2025; KRW 2.8bn a year earlier)	No filing yet announcing completion or entry into service
Capitalised development costs	KRW 17.7bn	Amortisation has not begun
Contract assets	KRW 40.4bn	Work performed but not yet converted to cash

 

The account that this was money spent ahead of growth has not yet been confirmed in filed figures. It would be confirmed when recovery begins in the three places above.

7-4. Funding history and potential dilution

Item	Detail
Instrument	Fourth privately placed convertible bond
Amount	KRW 130bn (~$93m)
Issued / maturity	April 29, 2026 / April 29, 2031
Coupon	0%
Conversion price	KRW 54,603
Shares issuable on conversion	2,380,821 (+17.7% of shares outstanding)
Conversion window opens	April 29, 2027
Early redemption right exercisable	October 29, 2028

 

Source — material event report on convertible bond issuance (filed with DART, April 2026); securities issuance result (filed with DART, April 2026).

With a 0% coupon, no cash interest goes out. In accounting terms, however, KRW 6.7bn a year is recognised as expense. The bond ends one of two ways. If it converts, existing shareholders' share falls by 17.7% (dilution). If the conditions for the early redemption right are met, the company may have to return KRW 130bn in cash.

7-5. Projected revenue (my own estimate)

KRW bn	2024	2025	2026	2027	2028
Revenue	106.5 (reported)	164.5 (reported)	187	212	241
In USD	~$76m	~$118m	~$134m	~$152m	~$172m

 

The two years marked "reported" are figures confirmed in filings; the three to their right are my estimates.

These figures are my base-case estimate and may differ from what actually happens. If the scenario changes, so do these values. The 2028 revenue figures for the positive and conservative cases appear in the table in Section 9.

This table assumes revenue tracks the growth of the market it sits in (13.62% a year). As noted in "What Is Not Yet Settled ①", there is a second calculation that applies the scheduled completion date of the large contract ending in July 2027. Which one holds depends on the disclosed value of follow-on contracts.

Two lines at which profit begins

Line	Annual revenue required	Composition
Operating profit reaches zero	About KRW 216bn	(SG&A KRW 32.4bn + development cost amortisation KRW 3.5bn) ÷ gross margin 16.6%
Profit for shareholders reaches zero	About KRW 258bn	The above + convertible bond finance cost KRW 5.8bn + other KRW 1.1bn

 

On a quarterly basis that is KRW 64.5bn per quarter. The first-half 2026 quarterly average was about KRW 41bn — two-thirds of the way there.

 

8. Evidence — How Far the Case That This Business Works Has Come

8-1. The revenue direction turned

Confirmed — organic growth was +10.4% in Q1 2026 and +35.0% in Q2 2026, two consecutive positive quarters, measured with the consolidation perimeter unchanged at 11 entities. Over the same period, six new or expanded contract disclosures followed one another.

What it means for the judgment — organic growth on the same basis averaged −15.5% a year over the past five years. This is where that direction bent.

One instance — HPC revenue went from KRW 4.9bn in Q1 2025 to KRW 16.3bn in Q1 2026.

What it would look like ahead — if this holds for two more quarters, a path to the KRW 64.5bn quarterly line opens.

8-2. It actually sold to overseas customers

Confirmed — service export revenue was KRW 6.9bn in 2024, KRW 19.0bn in 2025, and KRW 40.2bn in the first half of 2026 — 49% of total revenue in that half.

What it means for the judgment — this is revenue realised, not a plan or an announcement.

One instance — a design contract of KRW 39.6bn disclosed in April 2026 runs from March 2026 to May 2027, and was 58.24% complete at the half-year mark.

What it would look like ahead — a wider overseas base creates revenue independent of domestic conditions. As noted in 4-3, the currency effect runs the other way.

8-3. The nature of the work shifted

Confirmed — AI and high-performance computing rose from 31% of total revenue in 2024 to 38% in 2025 and 57% in the first half of 2026.

What it means for the judgment — the weight of the business moved toward datacentre chips, where individual engagements are larger and development periods longer.

One instance — the Samsung leading-edge server programme runs to July 2028.

What it would look like ahead — if development programmes carry through to production, recurring revenue appears. So far the movement is the other way: product revenue fell 41.2% year on year in the first half of 2026.

Where confirmed outside evidence has not yet arrived

I set this out at the end of the section rather than as an evidence line of its own. Building a line where no confirmed fact exists makes the confirmed set look larger to the reader than it is. There are three.

What	Current state	What would confirm it
Construction in progress, KRW 27.9bn	No completion or entry-into-service filing	Transfer to the operating account, with revenue growth in that unit
Capitalised development costs, KRW 17.7bn	Amortisation not begun	Amortisation starting, or the development converting to revenue
Two contracts ending in H1 2027, KRW 75bn	Follow-on volume is two engagements of undisclosed value	Disclosure of a new or expanded contract of KRW 10bn or more

 

The five-year cumulative −KRW 51.5bn in Section 7 supports this.

 

9. Three Futures — Why the Upper and Lower Spreads Differ

I set out 2028 in three cases. The figures below are projections, not reported results.

2028, KRW bn	Positive case	Base case	Conservative case
Revenue	300	241	241
Operating profit	+13.9	+4.2	−15.6
Profit for shareholders	+6.9	−2.7	−22.2

 

The sign of profit has not yet turned.

●       What the positive case becomes — revenue reaches KRW 300bn, clears the break-even line, and profit is left for shareholders for the first time.

●       What the base case becomes — revenue comes close to the break-even line and the operating line turns positive, but convertible bond costs are not covered and the shareholders' share is still negative.

●       What the conservative case becomes — revenue is the same as the base case, but the cost ratio stays at the first-half 2026 level (92.9%, excluding the effect of the cost-estimate change), and the operating line returns to loss.

Why the spread is tilted — the spread to the downside (−KRW 19.5bn) is twice the spread to the upside (+KRW 9.6bn). The two rest on different materials. The positive case's KRW 300bn revenue rests on a target the company set out at an investor presentation in May 2026. I confirmed that target through press material dated May 2026 and could not reconcile it against a primary filing. The conservative case's 92.9% cost ratio, by contrast, was already observed in the first half of 2026. The larger figure is on the upside; the weight of confirmed fact is on the downside.

The downside variables where the author's estimates may differ from reality are as follows

There is one place where my estimate may read darker than reality. Two contracts (KRW 12.5bn and KRW 22.1bn) are not disclosed by business segment, so I could not incorporate the effect of their ending in 2027. Working the other way, I assumed no new orders at all in 2027 and 2028 — an assumption darker than reality. The two run in opposite directions, but which is larger cannot be settled until the contract values are disclosed.

 

10. What to Watch — Which Signals, and When

If only one thing, it is revenue scale. The technology and the order flow are confirmed; what has not been reached is the revenue that covers roughly KRW 30bn of annual fixed costs and KRW 5.8bn of finance costs.

#	What	Confirmed so far	What would change the picture	When
1	Volume beyond 2027	Two contracts of KRW 75bn end in H1 2027; follow-on is two undisclosed engagements	Disclosure of a new or expanded supply contract of KRW 10bn or more	Any time (DART)
2	Crossing break-even	H1 2026 quarterly average about KRW 41bn	Quarterly revenue above KRW 64.5bn	November 14, 2026
3	Whether the revenue turn holds	Two consecutive positive organic quarters	HPC revenue growing for two more quarters	November 14, 2026
4	Domestic share	54.9% → 34.5% (one half-year observation)	Recovery above 40%, or two consecutive readings below 30%	November 2026 (peer Q3 reports)
5	Recovery of money spent ahead	Construction in progress KRW 27.9bn; no completion filing	Transfer to the operating account plus revenue growth in that unit	March 2027 (annual report)
6	Length of runway	About 7.4 quarters	Falling below 6 quarters	Each quarterly report
7	How the convertible bond ends	0% coupon; conversion window opens April 29, 2027	Any filing on early redemption	Any time → October 29, 2028

 

 

11. Strengths, Weaknesses, Opportunities, Threats — One Page

Box	Content
Strengths (already confirmed)	Official Samsung Foundry design partner; Arm Total Design partner
	Gross margin of 16.6% — the highest of the three domestic firms
	Two consecutive positive organic quarters (+10.4%, +35.0%)
	Service exports of KRW 40.2bn — revenue realised, not planned
	726 R&D engineers; 19 registered patents
Weaknesses (not yet confirmed, and cash)	Over five years, KRW 51.5bn more cash went out than came in
	SG&A ratio of 18.5%, against 11.7% at the domestic peer
	Domestic share fell from 54.9% to 34.5% (one half-year observation)
	90% of revenue is recognised on the company's own cost estimates
	Recovery of construction in progress and development costs is unconfirmed
Opportunities	Combined half-year revenue of the three domestic firms rose 95.6% year on year
	Expansion of Samsung's leading-edge process; the server programme runs to July 2028
	Two follow-on engagements whose value is not yet disclosed
Threats	Two contracts of KRW 75bn end in the first half of 2027
	Customers designing in-house, and the large overseas design houses
	A 10% rise in the US dollar reduces pre-tax profit by KRW 2.3bn
	The KRW 130bn convertible bond — dilution or cash repayment

 

In one sentence — the technology and the order flow are confirmed and the revenue direction has turned, but the company stands in front of the threshold where that carries through into profit for shareholders.

 

12. Summing Up — The Prevailing View and What I Confirmed

The prevailing view of this company is that revenue is rising sharply, so profit should soon follow steeply, and that becoming an official Samsung Foundry partner changed the company's constitution. What I confirmed in filings is this. Revenue is rising — that much is fact. It rose for two consecutive quarters, and exports of KRW 40.2bn in a single half are measured figures. But profit here does not rise gradually alongside revenue. It changes in a step at the moment annual revenue crosses KRW 258bn. First-half 2026 revenue was KRW 82bn, about KRW 41bn a quarter — two-thirds of the way to that threshold.

Fact assessment

What	Where it stands	Basis and certainty
Direction of the business	Revenue rising; the sign of profit has not turned	Two consecutive positive organic quarters (+10.4%, +35.0%) and exports of KRW 40.2bn are measured. With only two quarters observed, certainty is low
Position of the evidence	Three confirmed items outside the financials (direction turn, export realisation, shift in the nature of the work)	The next step is quarterly revenue of KRW 64.5bn and disclosure of volume beyond 2027

 

What would change this picture

Direction	Confirmable signal	When
Upward	Quarterly revenue above KRW 64.5bn for two consecutive quarters, or two or more new or expanded contracts of KRW 10bn+ disclosed	November 14, 2026 / any time
Downward	Organic growth turning negative for two consecutive readings, or cash on hand falling below six quarters of runway	Each quarterly report

 

Closing — AD Technology is not a company that has already improved; it is a company standing in front of a threshold, and whether it crosses can be checked against a single number: quarterly revenue of KRW 64.5bn.

 

13. Limits — Please Read This With the Rest

This analysis contains the following unresolved limitations.

1. Three items were left open at the review stage. ⓐ In calculating 2028 revenue I stepped one pace away from my standard method, and I could not re-verify within this review's source material whether that step is consistent with my own rules. ⓑ I could not establish where the gap between the analysts' estimate and mine comes from. ⓒ I could not directly verify whether any unresolved counter-arguments remained from an earlier stage. When each will be checked is set out at the top of this piece.

2. I adapted part of my standard method to fit this company. Calculating 2028 revenue using the measured growth of the domestic peer group is that place. The difference between the two calculations is stated at the top.

3. One comparison step was not performed. No peer reference table exists for this industry, so I skipped measuring the company against its peer group.

4. An admission about the estimates — this company does not disclose volumes and unit prices by business unit. The projected revenue in 7-5 was therefore not built up from volume and price; it was built by adding and subtracting at the level of amounts. It is closer to a proportional estimate.

5. What I could not verify — four things: the financials of the overseas competitors (GUC, Alchip); the gross margins of the two domestic peers; which business unit two contracts (KRW 12.5bn and KRW 22.1bn) belong to; and which institution issued the analyst estimate cited at the top, and when. I did not use that estimate as a basis for judgment.

6. What the company does not disclose — customer names, the value of the two follow-on engagements, when development-cost amortisation begins, and how the KRW 7.8bn cost-estimate change is distributed across future years.

7. This English edition carries less detail than the Korean edition in two places. Named contract counterparties and certain technical specifications were removed under my own sensitive-information filter for English-language material. No figure, conclusion, or assessment differs between the two editions.

 

Appendix — Key Figures on One Page

Item	Value	As of
Shares outstanding	13,462,007	Jun 30, 2026
Treasury shares	486,181 (3.61%)	Jun 30, 2026
Shares in free float	12,975,826	Jun 30, 2026
Shares issuable on bond conversion	2,380,821 (+17.7%)	Jun 30, 2026
Convertible bond amount / coupon	KRW 130bn / 0%	Apr 29, 2026
Employees (consolidated / R&D)	877 / 726	Jun 30, 2026
Registered domestic patents	19	Jun 30, 2026
Revenue (2023 / 2024 / 2025 / H1 2026)	KRW 100.2 / 106.5 / 164.5 / 82.0bn	Each period
Operating profit (2023 / 2024 / 2025 / H1 2026)	KRW −17.4 / −17.0 / +2.9 / −1.5bn	Each period
Profit for shareholders (2023 / 2024 / 2025 / H1 2026)	KRW −16.0 / −14.3 / +3.6 / −6.2bn	Each period
Cash and equivalents + short-term instruments	KRW 133.1bn	Jun 30, 2026
Borrowings / convertible bond (book) / debt-to-equity	KRW 84.8bn / 91.9bn / 133%	Jun 30, 2026
How long the cash on hand lasts	About 7.4 quarters	Jun 30, 2026
Contract assets	KRW 40.4bn	Jun 30, 2026
Construction in progress	KRW 27.9bn	Dec 31, 2025
Capitalised development costs	KRW 17.7bn (amortisation not begun)	Jun 30, 2026
Projected revenue (my own estimate)	2026 KRW 187bn / 2027 KRW 212bn / 2028 KRW 241bn	—
Principal contracts	KRW 35.2bn (to Feb 2027); KRW 39.6bn (to May 2027); two of undisclosed value	As disclosed

 

Sources

Document	Filed
Half-year report (June 2026)	DART, August 2026
Annual report (December 2025)	DART, March 2026
Supply contract disclosure (KRW 39.6bn)	DART, April 2026
Supply contract disclosure, amended (contract increase)	DART, June 2026
Material event report — convertible bond issuance	DART, April 2026
Securities issuance result	DART, April 2026
Disclosure of change in revenue or profit structure	DART, February 2026
Supply contract disclosure (KRW 12.3bn)	DART, November 2025
Supply contract disclosure (design software customer)	DART, December 2025
ASICLAND half-year report (June 2026)	DART, August 2026
Gaonchips half-year report (June 2026)	DART, August 2026

 

 

This piece reflects my personal analysis of publicly disclosed filings. It is not investment advice, nor an offer or solicitation to buy or sell any security.

I am not a licensed investment adviser or broker-dealer, and I do not provide individual consultations.

This piece does not provide a fair value, price target, or valuation opinion.

All data and figures are drawn from sources I believe to be reliable, including DART, Korea's regulatory filing system; I cannot guarantee their accuracy or completeness. Forward-looking figures are my own estimates, may change, and may prove incorrect. All figures are as of the stated date. KRW amounts are converted at a fixed rate of KRW/USD 1,400, which is not the spot rate on any given date.

Investment decisions and their outcomes rest solely with the reader, and loss of principal is possible.

This piece may not be used as evidence in determining legal liability.

As of August 26, 2026; filings and results published after that date are not reflected.

This piece is written in the public interest — to share information and support investors' right to know. It is not intended to damage the reputation of, or interfere with the business of, any company. Facts and my own interpretation are separated in the text; where interpretation appears, it is my view.

I look at the company's core competitiveness and the direction of its results first, rather than starting from the market's short-term assessment.

I do not hold a position in the security discussed (as of August 27, 2026). From the time this piece is published, I may buy or sell it without prior notice.