Physical AI Data Breakthrough, Naver Bets Big on 15 Billion

Physical AI data collection startup NdotLight has closed a 15 billion won (approximately $11 million) funding round, with Naver D2SF making a third consecutive follow-on investment as part of the deal. Korea Development Bank led the round, published on August 11.

Key Takeaways

Korea Development Bank led a 15 billion won funding round in NdotLight, published on August 11
Naver D2SF first invested in NdotLight during the startup’s seed round, then returned for the Series A
NdotLight collects and processes real-world sensor data, including camera, lidar, and motion capture feeds
Google recently surpassed Naver as South Korea’s largest mobile platform by monthly active users

The raise marks a notable vote of confidence in physical AI data infrastructure at a time when the scarcest resource in robotics and autonomous systems is not compute but clean, real-world training data.

Naver D2SF Bets On Physical AI Data Three Times Running

Naver D2SF, the corporate venture capital arm of South Korean internet giant Naver, made its first investment in NdotLight during the startup’s seed round. It returned for a second time during a subsequent Series A raise.

This third investment, as part of a round led by Korea Development Bank, makes NdotLight one of the most consistently backed physical AI bets in Naver’s portfolio, according to a PR Newswire release published on August 11.

NdotLight’s CEO is Jinyoung Park. The company collects and processes real-world sensor data, including camera, lidar, and motion capture feeds, specifically to train physical AI systems.

Physical AI, sometimes called embodied AI, refers to machine-learning systems designed to operate in and interact with the physical world.

This is distinct from language models or image generators, which work with digital inputs only. Physical AI includes robots, autonomous vehicles, and industrial automation systems that must perceive, plan, and act in three-dimensional space.

Why Physical AI Data Is The New Scarcest Resource

The training data physical AI systems require cannot be fabricated cheaply at scale the way text or image data can.

A language model can ingest terabytes of scraped web content. A robot learning to pick an oddly shaped object off a shelf, or an autonomous vehicle navigating an unmarked intersection, needs thousands of hours of real sensor footage from the physical environment it will inhabit.

This creates a structural bottleneck.

Compute for training AI models has grown faster than the supply of physical-world data to train them on. Companies like Tesla, Figure, and 1X Technologies have invested heavily in proprietary data collection fleets precisely because that data cannot be easily purchased or replicated.

NdotLight is positioning itself in the gap: a specialized data services company for physical AI builders who lack the resources to run their own collection infrastructure.

The Korea Development Bank’s decision to lead this round is also significant. KDB is a state-owned policy bank that frequently backs companies aligned with South Korea’s industrial strategy.

Its presence in the round signals that South Korean policymakers view physical AI data infrastructure as a strategic national asset, not merely a venture bet.

Also Read: Ripple Backs AI Compliance Push as XRP Ledger Hits 10M Agent Transactions

From Naver’s Search Engine To Its Robotics Data Play

Naver built its reputation as South Korea’s dominant search engine and has since expanded into cloud computing, mapping, and AI research. Naver CLOVA, the company’s AI division, has produced large language models and multimodal systems that compete across East Asia.

Naver D2SF, the startup arm that backed NdotLight, focuses specifically on deep technology bets that are too early for conventional venture capital.

Its portfolio includes companies in robotics, semiconductors, and AI infrastructure.

Google recently surpassed Naver as South Korea’s largest mobile platform by monthly active users, according to reporting in the Chosun Ilbo this week, a development that has intensified Naver’s urgency around AI capability. Backing physical AI data infrastructure through D2SF is part of a broader defensive and offensive posture: build the picks-and-shovels layer before competitors capture it.

NdotLight’s trajectory fits a recognizable pattern in Asian AI investment.

Rather than building a frontier model, the company focuses on a specific input layer that every frontier physical AI system eventually needs. This infrastructure-layer approach mirrors how companies like Scale AI in the United States built early dominance in AI data labeling before expanding into evaluation and model development services.

What The Korea Development Bank Round Signals For The Sector

The involvement of Korea Development Bank as lead investor carries implications beyond NdotLight’s balance sheet.

KDB has historically co-led rounds in sectors the South Korean government wants to accelerate, from semiconductor equipment to battery technology. Adding physical AI data to that list places it alongside strategic hardware sectors in the government’s industrial calculus.

For the broader physical AI data market, NdotLight’s raise arrives as several competing approaches are being tested.

Some startups use synthetic data generated in simulation to supplement real-world feeds. Others use crowdsourced collection via contracted human operators.

NdotLight’s approach, detailed in the PR Newswire release, centers on systematic real-world collection pipelines tailored to specific robot and autonomous system use cases.

The 15 billion won round is modest by the standards of frontier AI funding in 2026, when rounds of $500 million or more have become common for model developers. That modesty reflects the sector’s current stage.

Physical AI data companies are building infrastructure that will matter most in three to five years, as humanoid robots and autonomous systems move from laboratory demos toward commercial deployment.

Naver D2SF’s three consecutive investments suggest it believes NdotLight will be positioned to capture significant revenue when that deployment wave arrives.