Quenching AI’s Thirst for Data: Our Investment in CUBIG

It’s accepted wisdom that AI models require access to extensive quantities of data for model training so that they can identify patterns, understand context and reduce bias, helping the models to generate accurate predictions.

But what if the data needed for this training isn’t available? The data might simply not exist in the required quantity. Or it might be private, owned by an entity that does not want to make it publicly available, or which will require compliance with onerous financial terms to license it. In such scenarios, synthetic data represents a credible way forward that many model builders are adopting.

Synthetic data is generated using techniques that typically require access to samples of real data in order to ensure that the synthesised data is realistic in terms of patterns, correlations, and structure. As a result, synthetic data can sometimes be too realistic - i.e. it can inadvertently expose attributes of the real data that cause the identity of the owners of it to be revealed. This impacts data privacy at an operational level and can also have legal ramifications, including the potential nullification of exemptions from data privacy laws such as the GDPR in the EU.

To mitigate against such exposure of real data and the identities of its owners, cutting-edge privacy-enhancing technologies (PETs) can be leveraged. Iona Star has been tracking developments in AI privacy for some time, and today we are pleased to announce that, alongside our colleagues at Friend Investment Partners, we co-led the Series A investment round in CUBIG, a leader in the space.

Founded in Seoul, Korea in 2021, CUBIG has perfected the use of differential privacy, which adds mathematical ‘noise’ to synthetic datasets during their generation to mask the contribution of real data from any single entity. This makes it impossible to infer specific information about a contributing entity while preserving the characteristics of the data provided. Uniquely, CUBIG is the only multimodal and domainless synthetic data company and platform to be powered by differential privacy.

With initial seed funding from the venture arm of Korea’s Naver cloud provider and from the Korea Development Bank, CUBIG has already secured early customers for its platform, including Kyobo Life Insurance, IBK Industrial Bank of Korea, and Gangnam District Office. Moreover, the company was recently listed as a next-generation core vendor in Gartner’s "Emerging Tech: Hyper-Synthetic Data" report, which represents important validation to underpin future enterprise sales growth.

 “At CUBIG, we believe that the future of AI will not be defined solely by the intelligence of models, but by the integrity and accessibility of the data that fuels them,” said Bae Ho, CEO of CUBIG. “Our ambition is to go beyond synthetic data generation and build a trusted data infrastructure that empowers industries to innovate without compromise. By working closely with partners like Iona Star, we are laying the foundation for a global ecosystem where enterprises, governments, and communities can securely share, verify, and utilise data at scale. This is not just about solving today’s challenges in AI privacy; it is about shaping a future where data can be both a growth driver and a public good.”

 “Data is the lifeblood of AI, but real-world data often comes with constraints around privacy and scale,” said Kevin McGivern, Partner, Iona Star. “It is also the next bottleneck for AI, and the future will be more synthetic data combined with comprehensive, integrated solutions and not only stand-alone data generation. CUBIG is redefining how businesses can responsibly harness synthetic data to accelerate innovation, all while maintaining privacy, trust and compliance.” 

 Building on its synthetic datasets, CUBIG plans to develop customised AI agent models for specific industries, such as finance, healthcare, and public sector services. It is also building an integrated data ecosystem that connects data verification, monetisation, integration, privacy, and AI utilisation functions with a mission to become the All-in-One Data Company.

CUBIG will deploy its recent funding and work with Iona Star to accelerate a global expansion, establishing operations in the UK and North America. The UK presence will allow CUBIG to address opportunities across Europe, where stringent data privacy regulations represent hurdles to AI adoption.