KOSDAQIT & Software355390

CrowdWorks

₩1,422▲ 0.21%2026-10-02 close

Market Cap

₩22.4B

Turnover

₩200M

Volume

150,000 shares

Shares out.

15.9M

PER

—

PBR

1.0×

EPS

-₩1,330

Dividend Yield

0.00%

PER, EPS, PBR and BPS are calculated in-house from the last 4 quarters (2025Q3–2026Q2) · Dividend yield is based on ₩0 per share · Prices as of the 2026-10-02 close

Report overview

CrowdWorks: AI Data Labeler at an Ownership Crossroads

CrowdWorks faces the twin tasks of reviving revenue and repairing profitability just as its controlling shareholder and CEO have changed and repeated capital raises have diluted the share base.

1
2025 consolidated revenue fell to KRW 10.48 billion from KRW 12.01 billion in 2024, while the operating loss widened to KRW 13.37 billion.
2
Between December 2025 and February 2026 the controlling shareholder and CEO were both replaced, putting the company in a governance transition.
3
The company has been expanding industry-specific data contracts, including participation in a UNIST-led shipbuilding AI project worth KRW 40.3 billion through 2030.
4
Three separate capital raises in August 2024, December 2025 and April 2026 bolstered equity but also raised dilution and funding-transparency concerns.
5
The global data-labeling market remains in a high-growth phase, but domestic competition from AIMMO and AIWORKX (formerly Testworks) is intensifying.

Business structure

Founded in 2017, CrowdWorks listed on KOSDAQ in August 2023 via a SPAC merger and operates as an AI data and solutions company.

The firm organizes its business around three pillars—Data, AI, and Education—expanding from data labeling into AI-ready data construction, custom model development and operation, and model evaluation and verification.

Its core solutions include Alpy, an all-in-one platform for agentic AI development and operation, the Alpy Knowledge Compiler, which preprocesses unstructured documents into RAG-suitable formats, and Alpy Evaluation, a model assessment tool.

Its client base spans roughly 430 companies including Naver, Samsung Electronics, and Krafton, and the company states that more than 70% of KOSPI's top 30 technology firms have chosen it for model development.

Its labeler pool exceeds 600,000 workers, operating under a hybrid model in which AI handles first-pass processing while human reviewers—described by the company as domain experts with experience in fields such as finance and construction—perform detailed quality checks.

Gartner has reportedly rated CrowdWorks alongside OpenAI, Google, and Microsoft as a leading generative AI vendor for the financial sector. Domestically, key competitors include AIMMO, which specializes in autonomous-driving data, and AIWORKX, the rebranded former Testworks.

Between December 2025 and February 2026 the controlling shareholder changed from founder Park Min-woo to XRP1 Association led by Kim Kwang-il, and the board was reorganized to six members under new CEO Kim Kwang-il, raising questions about the company's future strategic direction.

Earnings trend

Quarterly revenue · Operating profit

₩2B-₩3B25Q2₩2.4B-₩3B25Q3₩4.2B-₩4.6B25Q4₩2.8B-₩2.2B26Q1₩2.4B-₩3.6B26Q2

RevenueOperating profit

Annual revenue · Operating profit

₩11.9B-₩6.3B2022₩24B-₩1.8B2023₩12B-₩11.7B2024₩10.5B-₩13.4B2025

RevenueOperating profit

Quarterly results · Last 5 quarters
Quarter	Revenue	Operating profit	Op. margin
2025Q2	₩2B	-₩3B	−150.9%
2025Q3	₩2.4B	-₩3B	−127.9%
2025Q4	₩4.2B	-₩4.6B	−108.0%
2026Q1	₩2.8B	-₩2.2B	−75.8%
2026Q2	₩2.4B	-₩3.6B	−152.5%

Annual results
Year	Revenue	Operating profit	Net income (parent)	Op. margin	ROE	Debt-to-equity
2022	₩11.9B	-₩6.3B	-₩13.5B	−53.0%	−312.3%	313.3%
2023	₩24B	-₩1.8B	-₩9.6B	−7.4%	−67.0%	191.3%
2024	₩12B	-₩11.7B	-₩8.8B	−97.6%	−86.5%	218.6%
2025	₩10.5B	-₩13.4B	-₩21.2B	−127.5%	−123.3%	141.7%

Consolidated basis (financial statements including subsidiaries) · Final figures from DART filings · Net income attributable to owners of the parent · Data as of 2026-09-05

Earnings analysis

Consolidated revenue fell to KRW 10.484 billion in 2025 from KRW 12.014 billion in 2024, continuing a sharp two-year decline from the 2023 peak of KRW 23.978 billion.

The operating loss widened for three straight years, from KRW 1.786 billion in 2023 to KRW 11.726 billion in 2024 and KRW 13.366 billion in 2025, pushing the operating margin from -7.4% in 2023 to -97.6% in 2024 and -127.5% in 2025.

Net loss attributable to owners jumped from KRW 8.822 billion in 2024 to KRW 21.198 billion in 2025, with the fourth quarter of 2025 (revenue of KRW 4.238 billion) standing out for a particularly large net loss of KRW 8.891 billion that drove much of the annual deterioration.

Losses then narrowed somewhat in the first quarter of 2026, with revenue of KRW 2.849 billion, an operating loss of KRW 2.161 billion, and a net loss of KRW 2.700 billion, before widening again in the second quarter of 2026 as revenue slipped to KRW 2.351 billion, the operating loss expanded to KRW 3.586 billion, and the net loss reached KRW 3.052 billion.

Over the trailing four quarters from the third quarter of 2025 through the second quarter of 2026, cumulative net loss attributable to owners totaled KRW 17.772 billion.

On the balance sheet, equity attributable to owners fluctuated from KRW 4.337 billion in 2022 to KRW 14.293 billion in 2023, KRW 10.197 billion in 2024, and KRW 17.197 billion in 2025, a pattern consistent with repeated capital injections offsetting ongoing losses.

The debt ratio eased from 313.3% in 2022 to 191.3% in 2023, rose again to 218.6% in 2024, then improved to 141.7% in 2025. Operating cash flow remained negative throughout 2022–2025, with the 2025 outflow of KRW 10.963 billion exceeding the KRW 10.206 billion outflow recorded in 2024.

Industry analysis

The global data-labeling market is projected to grow from USD 2.61 billion in 2026 to USD 7.02 billion by 2031, a 21.94% compound annual growth rate, driven by foundation-model developers and autonomous-driving and industrial firms seeking high-quality training data.

Meta's USD 15 billion acquisition of Scale AI in June 2025 elevated labeling infrastructure from a commoditized outsourcing service to a strategic competitive asset, a shift credited with prompting OpenAI and Google to diversify their annotation vendors.

Hybrid sourcing—keeping sensitive data in-house while outsourcing bulk work—grew at a 22.48% rate in 2026, and while outsourcing still accounted for 63.43% of the market in 2025, budgets are shifting toward hybrid models amid intellectual-property concerns and a shortage of domain experts.

In Korea, CrowdWorks positions itself as the largest platform with a labeler pool exceeding 600,000, while competitors such as AIMMO, which specializes in autonomous-driving data, and AIWORKX (the rebranded former Testworks) compete in their own niches.

A broader industry trend is the shift away from simple repetitive labeling toward higher-value work such as reviewing AI-generated first-pass labels, specialized domains like medicine and law, and evaluating AI response quality.

Against this backdrop, CrowdWorks presents its end-to-end capability—spanning data construction, model development, operation, and evaluation—as its point of differentiation.

Outlook

In April 2026 the company was selected, as part of a UNIST-led consortium, to participate in a shipbuilding-focused 'hyperscale industrial AI' research support project backed by a total of KRW 40.3 billion in government funding through 2030, working alongside HD Hyundai Heavy Industries and HD Korea Shipbuilding & Offshore Engineering to build shipbuilding-specific datasets.

That May, at AI EXPO KOREA 2026 in Seoul, the company demonstrated its patented Alpy Knowledge Compiler, aimed at the performance degradation problem in RAG systems, along with its Workstage data-construction platform, showcasing its enterprise data preprocessing and review capabilities.

On governance, an extraordinary shareholders' meeting in February 2026 saw new directors installed alongside a simultaneous change in controlling shareholder and CEO, with the new management's top priority widely seen as reversing years of operating losses.

In a December 2025 shareholder letter the company said funds raised through its capital increase would be directed toward acquiring a profitable business with synergies to its existing AI operations, though no specific acquisition target has been confirmed or disclosed as of this report.

However, the February 2026 extraordinary meeting also added business purposes unrelated to AI data—including ammunition and guided-weapon components, communications equipment manufacturing, and entertainment and performance management—to the company's articles of incorporation, leaving the new management's actual strategic direction uncertain.

Cash available for M&A is estimated at roughly KRW 9 billion, while the separate KRW 23 billion raised through the earlier rights offering has already been earmarked entirely for operating expenses.

Valuation

PER

—

PBR

1.0×

ROE

-130.5%

EPS

-₩1,330

BPS

₩1,358

Dividend per share

₩0

CrowdWorks has posted net losses for years, putting it in a range where a conventional price-to-earnings ratio cannot be meaningfully calculated, so valuation discussion naturally centers on the relationship between share price and net assets.

The price-to-book ratio appears to sit at a level that is neither deeply discounted nor richly premium to net asset value, though it should be considered alongside the fact that book value per share itself has shifted repeatedly due to successive capital raises.

The company currently pays no dividend, so dividend-related comparisons do not apply.

Having moved past an early post-listing period of elevated valuation, market attention appears to have shifted from growth expectations toward financial stability and governance transparency amid weak earnings and ownership uncertainty.

Where the valuation heads from here is likely to hinge on the pace of revenue recovery, whether the operating loss ratio improves, and whether the previously announced M&A plans materialize.

PER, EPS, PBR and BPS are all calculated in-house (the same method as Naver and Toss) · Dividend yield = cash dividend per share from DART filings (supplemented by KRX) ÷ current price · As of 2026-09-05

Bull factors

Expanding industry-specific data contracts

Participation in the UNIST-led shipbuilding AI project, backed by KRW 40.3 billion in government funding through 2030, gives the company a stable, government-supported revenue stream.

Its collaboration with major shipbuilders such as HD Hyundai Heavy Industries and HD Korea Shipbuilding & Offshore Engineering illustrates the potential for further industry-specific project wins.

Proprietary technology such as the Alpy Knowledge Compiler for unstructured data preprocessing could serve as a differentiator in security-sensitive sectors such as manufacturing and shipbuilding.

Large client base and brand recognition

The company counts roughly 430 clients including Naver, Samsung Electronics, and Krafton, and states that more than 70% of KOSPI's top 30 technology firms have chosen it for model development.

Having been named alongside OpenAI, Google, and Microsoft as a leading generative AI vendor for the financial sector by Gartner may help bolster external credibility. Its labeler pool of over 600,000 is the largest domestically, offering scalability advantages when taking on new projects.

Improved balance sheet from repeated capital raises

The debt ratio has fallen markedly from 313.3% in 2022 to 141.7% in 2025, reflecting repeated equity capital injections.

The new controlling shareholder has disclosed plans to use available cash for acquiring a business with proven profitability and synergy with the existing AI operations, opening a potential path to improved scale and earnings through M&A in addition to organic growth.

Bear factors

Sustained revenue decline since the 2023 peak

Revenue rose to KRW 23.978 billion in 2023 before falling sharply for two consecutive years to KRW 12.014 billion in 2024 and KRW 10.484 billion in 2025. Over the same period the operating margin deteriorated from -7.4% to -127.5%.

The improvement seen in the first quarter of 2026 also reversed in the second quarter, with revenue declining again and losses widening, showing that any recovery has not yet been consistent.

Governance uncertainty

During the shareholder and CEO transition from December 2025 to February 2026, questions were raised about the lack of clear disclosure regarding the funding sources of the new controlling entity, XRP1 Association.

The acquisition process itself was not smooth, with the capital-raise payment schedule revised multiple times and the price of the founder's stake sale lowered. There have also been assessments that many of the newly appointed directors lack direct experience relevant to the core AI data and solutions business.

Expansion into business areas unrelated to the core

At the February 2026 extraordinary shareholders' meeting, business purposes including ammunition and guided-weapon components, communications equipment manufacturing, and entertainment and performance management were added to the articles of incorporation.

Because these lack a clear connection to the existing AI data business, they have raised concerns that new management might pursue unrelated diversification rather than focusing on the core business.

Risk factors

Capital raising and dilution risk

Rights and third-party capital increases in August 2024, December 2025, and April 2026 have continuously expanded the share count. Operating cash flow has been negative every year from 2022 through 2025, so the possibility of further need for external funding cannot be ruled out. Additional capital raises going forward could further dilute existing shareholders.

Intensifying competition and pricing pressure

As AI auto-labeling spreads, pricing for simple repetitive tasks has trended lower, and domestic competition from specialized rivals such as AIMMO and AIWORKX is intensifying. If large clients increasingly process sensitive data in-house as hybrid sourcing spreads, external outsourcing demand could decline.

Governance and strategy execution risk

Controversy over the funding transparency of the new controlling shareholder and concerns about the limited relevance of newly appointed directors to the core business persist.

With multiple business purposes unrelated to AI data now added to the articles of incorporation, it remains undetermined whether actual M&A or new-business initiatives will strengthen the core business or instead lead to strategic dispersion.

What to watch next

Mid-November 2026
The third-quarter 2026 report is expected around this time, offering a check on whether revenue recovery continues and how the operating loss ratio trends.
Fourth quarter of 2026
Watch for whether the previously flagged acquisition target with synergy and proven profitability is finalized and disclosed, and how the deal terms and use of funds are specified.
From the second half of 2026 onward
Monitor whether the initial data-construction milestones of the UNIST consortium's shipbuilding AI project are met and whether follow-on contracts materialize.
Fourth quarter of 2026
Check whether new management actually moves to pursue the newly added business purposes—such as ammunition/guided-weapon components or entertainment—or instead reaffirms a focus on the core AI data business.
Ongoing monitoring
Given years of negative operating cash flow, continue to watch for disclosures of further capital raises or convertible bond issuances and the resulting degree of shareholder dilution.

Overall view

Since peaking in 2023, CrowdWorks has seen two consecutive years of contracting revenue and a deteriorating operating margin that reached -127.5% in 2025, all while navigating a transition in which both its controlling shareholder and CEO changed between late 2025 and early 2026.

Its participation in a government-backed shipbuilding AI project and expansion of industry-specific data services through proprietary technology such as the Alpy Knowledge Compiler are positive developments, but controversy over the new management's funding transparency and the addition of business purposes unrelated to its core operations leave questions about strategic consistency.

Repeated capital raises have improved the debt ratio but have also continued to increase the share count and dilute existing holders, while operating cash flow has remained negative for four consecutive years.

The global data-labeling market continues to grow at double-digit rates, keeping the broader industry opportunity open, but how the company differentiates itself from domestic rivals and responds to the spread of automated labeling remains to be seen.

Key things to watch going forward include whether the previously announced M&A materializes, the direction of revenue and profitability in the upcoming third-quarter results, and whether new management ultimately chooses to focus on the core business or pursue unrelated diversification.

Sources

Korea Exchange (KRX) — Prices · Market Cap · Volume
FSS electronic disclosure system (DART) — Financial statements · Dividend filings

Show 18 more articles and sources

Report written 2026-09-05 · Data as of 2026-09-04

This content is AI analysis of market data and web search results, provided for information only. It is not a solicitation or recommendation to invest. Investment decisions and their consequences are the investor's own responsibility. Data may be delayed or contain errors.