From Factory Data to On-Chain Trading, Web3 Firms Seek Opportunities in South Korea

South Korea's manufacturing data seen as an AI training asset

Lighter targets the local market with a simpler user experience

Global blockchain companies are identifying South Korea's manufacturing data and crypto trading market as new business opportunities. They say manufacturing and robotics data could become a new source of revenue if it is used to train AI models. They also argue that decentralized exchanges, or DEXs, need not only trade-verification technology but also the ease of use of mainstream financial apps to attract more users in South Korea.

Andrea Muttoni, chief executive officer of the DATA Foundation, and Vladimir Novakovski, chief executive officer of Lighter, outlined that view at the a16z Crypto Korea Summit held on October 1 at Josun Palace in Seoul's Gangnam district. As AI adoption grows, they said, systems that verify the origin of training data and the way transactions are processed will become increasingly important.

"South Korea's manufacturing data can be a new revenue stream in the AI era"

Muttoni said South Korea's strengths in manufacturing and robotics give it an advantage in the market for AI training data. Robots need large volumes of real-world data showing how people and machines move on factory floors to learn physical tasks. Data gathered online alone cannot easily meet that demand.

"US AI companies can keep scraping the internet, but that will not give them LG factory production processes or Hyundai Motor manufacturing data," he said. "If Korean companies organize that data into a usable format and provide usage rights, it can become a new revenue source." He added that the same data could also be used to develop in-house AI models and robots at domestic companies.

The key issue, he said, is building a system that can verify where data came from and how it was used. AI companies obtain data through brokers and other channels, but it remains unclear what data came from where and how it was used in training.

The DATA Foundation is building infrastructure to track that process. Muttoni said the Data Network has recorded more than 250 million "data receipts." Rather than storing the underlying data on the blockchain, the system logs when it was created, how it was collected and how it was used. Large datasets remain on conventional storage services, while records needed for verification are managed on the blockchain.

He also stressed that data providers should ultimately be compensated. "When AI produces a specific output, we need to be able to determine which data contributed and by how much, and reward those contributors," he said. "That is how the data economy can become sustainable."

Muttoni also pushed back on the idea that synthetic data will fully replace real-world data. AI-generated data is useful, but weather, market conditions, language and human behavior keep changing. AI will need a continuous flow of up-to-date real-world data, he said, adding that distinguishing collected data from synthetic data will become increasingly important.

Lighter says DEXs must be easy to use to win more users

Novakovski said user experience will determine whether decentralized exchanges win broader adoption. South Korean users accustomed to centralized exchanges will not choose a new service if it adds friction to the trading process.

"Users should not feel like they are using a DEX," he said. "It should feel like using a modern trading app or a neobank app." Complex technology, he said, should operate in the background while users benefit from lower costs and faster execution.

Lighter is an Ethereum-based DEX that uses zero-knowledge proofs to verify whether orders were processed according to predetermined rules. Novakovski said Lighter handles about 500 million orders a day while proving the outcome of each order. The platform is designed so users can verify that the system worked properly rather than simply trust the exchange's explanation.

He said that kind of verification will become more important as AI-driven trading expands. If something goes wrong when AI executes an order, users need a way to review the trading process and how the system operated. He added that it is also possible to prove whether preset risk limits were followed without disclosing the investment strategy itself.

Still, Novakovski was skeptical that AI will fully replace human investment decisions. Asset allocation and trading based on fixed rules can be automated, but even the top specialist trading firms already use advanced AI and massive computing power. AI is more likely to lift trading capabilities by one or two levels than eliminate traders altogether, he said.

In South Korea, Lighter is broadening its reach with local users. Novakovski said the platform added Korean-language support about a week after launch. The number of Korean-language users rose from about 30 initially to the thousands afterward. He added that the company has formed an Asia team and is meeting Korean traders directly.

Over the longer term, he said, computing resources, data and financial products will be connected through blockchain-based financial systems. Before a future in which AI makes every decision for people, a shift toward helping users make better judgments through better access to information and products is more likely to come first.