Flitto

Founded in 2012, Flitto listed on KOSDAQ in 2019 under the technology growth company special listing track and operates an AI language data and solutions business.

The company runs three segments—data sales, platform service, and AI solutions—with 1Q26 revenue mix of 45.6% data sales, 31.6% platform service, and 13.7% solutions, while exports accounted for 70.7% of sales.

The data sales segment supplies high-quality language data for AI model training to US-based global Big Tech companies, with a single customer accounting for more than 10% of total revenue, reflecting a stable base built on large-scale contracts.

In solutions, the B2B real-time interpretation product Live Translation and the B2C Chat Translation service each grew roughly 200% year over year.

More recently, Flitto has been expanding from text-centric language data into physical AI data for robotics and industrial automation training, developing a dedicated crowdsourcing platform for this purpose.

On the public sector side, Flitto serves as the data-lead partner within the Upstage-led consortium for the government's sovereign AI foundation model project, handling training and evaluation dataset construction, and has also won data-building contracts from the National Institute of Korean Language, the Institute for Information and Communications Technology Planning and Evaluation, and the Korea Federation of Micro Enterprise.

The company operates subsidiaries in China and Japan, with Flitto Japan conducting a business similar to the parent locally.

Peer AI-data companies include Crowdworks, Select Star, AI Works, and Alchera, but the company states it faces limited direct competition since none of these peers specialize in language data as Flitto does.

Flitto's annual results show a clear turnaround trajectory.

After an operating loss of KRW 6.60bn (-48.4% margin) on revenue of KRW 13.64bn in 2022 and a loss of KRW 5.09bn (-28.7%) on revenue of KRW 17.76bn in 2023, the company narrowed its operating loss sharply to KRW 396mn (-2.0%) on revenue of KRW 20.30bn in 2024, before achieving its first full-year operating profit in 2025 with revenue of KRW 35.98bn, operating profit of KRW 6.15bn (17.1% margin), and owner net income of KRW 6.45bn.

On a quarterly basis, revenue and operating profit peaked in 3Q25 at KRW 11.78bn and KRW 2.83bn (roughly 24% operating margin), following 2Q25's KRW 9.38bn revenue and KRW 2.43bn operating profit, before margins compressed sharply in 4Q25 to KRW 10.21bn revenue and KRW 720mn operating profit.

In 1Q26, revenue fell to KRW 5.21bn and the company swung back into an operating loss of KRW 64mn, a result the company attributed to R&D spending expanding to roughly 19% of standalone revenue for solutions and platform upgrades.

By 2Q26, the company returned to profit with revenue of KRW 8.11bn, operating profit of KRW 1.40bn, and owner net income of KRW 1.56bn, exiting the prior quarter's operating loss.

Looking at the trailing four quarters (3Q25 through 2Q26) together, operating margins swung widely between roughly -1% and 24% depending on the timing of large data-supply contract recognition versus R&D expenditure cycles.

The large net loss of KRW 6.79bn in 2023 reflected both the operating loss and additional non-operating factors, after which the balance sheet gradually improved through capital raises and business restructuring.

The AI training-data industry is in a phase of rising demand for high-quality language data, driven by intensifying LLM quality competition among global Big Tech and the spread of AI agent and multimodal technologies.

In addition, physical AI data used for training robots and industrial automation systems is emerging as a new growth axis, commanding higher unit prices than text data because it covers elements such as action trajectories, failure cases, and recovery processes.

Within Korea's AI data industry, companies such as Crowdworks (a general-purpose AI data marketplace), Saltlux (AI agents and public-sector search), and Selvas AI (medical specialization) are each expanding in their own specialized areas, while data copyright verification and supply-chain issues have become industry-wide topics.

The government is also fostering the domestic AI ecosystem and data supply chain through national projects such as the sovereign AI foundation model initiative.

Flitto's differentiation lies in its language-data specialization, built on a user base of more than 14 million people across 173 countries and 42 languages that continuously generates real-usage data.

However, because industry-wide demand is tied to the budget-execution cycles of Big Tech model development, individual companies' results can swing significantly depending on whether large contracts are signed and recognized within a given quarter.

For 2026, the company has set goals of diversifying revenue and strengthening synergy between its data and solutions businesses.

As a new growth area, Flitto is developing a dedicated physical AI data platform leveraging its language-data crowdsourcing know-how, and has begun collecting high-quality Arabic voice data to target the Middle East market.

The AI image translation service currently operates only in B2B form, but a company representative said it plans to expand into B2C to grow that revenue stream.

On the national-project front, Flitto handled data quality management and multilingual/multimodal data collection and refinement as part of the Upstage-led consortium's second-phase Global Frontier Foundation Model Development project, receiving related government subsidies.

The company's CFO characterized the 1Q operating loss as strategic spending to secure AI solution competitiveness and stated that, as large global data-supply contracts get recognized in revenue during the second half, the company aims to pursue both top-line growth and profitability simultaneously.

In June, the company disclosed that it had signed data-supply contracts with global IT companies worth KRW 22.66bn and KRW 8.89bn respectively for AI language model research and development, and the timing of revenue recognition from these contracts is a key variable for future results.

On the public sector side, ongoing contract wins from the National Institute of Korean Language, the Institute for Information and Communications Technology Planning and Evaluation, and the Korea Federation of Micro Enterprise indicate the company's continued policy of expanding participation in government-led AI projects.