Korean startups raise ₩72.16B in late-April week — physical AI leads

₩72.16B

Through 2025, the Korean startup funding headline was LLM apps and generative AI. ChatGPT Korean wrappers, content SaaS, chatbot tools — that was the capital concentration.

The pattern is shifting in 2026.

In the week of April 27 to May 1, 29 Korean startups closed funding. The 10 with disclosed amounts totaled ₩72.16B. Over half of that was robotics and physical AI. Roundup courtesy of Startup Recipe.

The two biggest rounds: Lobros (robot developer) at roughly ₩10B Series A and Loai (physical-AI startup) at ₩13B Series A. The two combined for ₩23B — about 32% of the week.

SMBA Minister Oh Young-joo: "Robotics is the new battlefield for Korean startups." Policy, capital, and tech are moving in sync.

Who's involved — government, capital, startups

Government dropped two major policies the same week.

First, ₩60T five-year strategic-technology program. AI, robotics, semiconductors, bio, quantum — 55 priority categories spanning government and private capital. AI/robotics share estimated around 35%.

Second, SMBA 2026 AI budget locked at ₩800B. Direct AI startup support, infrastructure, and talent training — up roughly 60% YoY.

For capital, government matching plus VC dry powder bumps round sizes. Series A averaged ₩5B last year; ₩8-10B is the new normal.

For startups, capital availability improves but competition intensifies. LLM-app saturation pushes pivots into robotics/physical AI.

The numbers — late-April rounds

Company	Round	Amount	Category
Loai	Series A	₩13B	Physical AI
Lobros	Series A	₩10B	Robot developer
Momocall	Seed	undisclosed	AI call assistant
ThePict	Seed	undisclosed	AI EdTech
25 others	varied	₩49.16B total	varied
Total (disclosed)	—	₩72.16B	—

Loai builds an integrated industrial-robot + physical-AI platform. ₩13B is the largest physical-AI Series A in Korea.

Lobros focuses on quadrupeds and industrial manipulators. ₩10B Series A funds production line buildout.

Momocall is an AI call assistant — auto-answer/booking for Korean SMB owners. SMBA Deep-Tech Youth Founders Academy cohort 1.

ThePict launched its EdTech Center: AI integration for career and employment training, riding the Korean education-AI adoption wave.

Wins and losses

Korean robotics/physical-AI startups get materially better fundraising — Series A sizes up, runway extending from 12-18 to 24-30 months.

The Korean economy gains a long-term structural shift from LLM apps to physical AI, dovetailing with manufacturing, semiconductor, and robotics strengths against US/China competition.

US/Chinese competitors — Boston Dynamics, Figure AI, Unitree — face a longer-term challenger emerging from Korea. Short-term direct threat is limited.

Korean VCs face pressure to find new categories as LLM apps saturate. Physical AI commands higher valuations but proven product-market fit is rarer.

Past cycles — Korean startup capital

Mobile app era, 2010-2014. Kakao, Baedalmin, Coupang. VC capital surged, many companies struggled with revenue-vs-valuation gap.

Fintech/blockchain era, 2017-2021. Toss, Bank Salad, Dunamu. Government policy plus capital concentration. 2022 valuation correction.

Content/metaverse era, 2020-2022. Hybe, SM, NCSOFT entering metaverse. Capital surge, demand shortfall, correction.

LLM-apps era, 2023-2025. Korean LLMs and wrapper SaaS. Saturation since 2025, partial correction.

Pattern: Korean capital trails global trends by 6-12 months. Physical AI started globally in 2024-2025; 2026 is the Korean ignition window.

Counter-moves

US — Figure AI, Apptronik, Sanctuary AI — own humanoid R&D. Capital advantage, but production/manufacturing scaling tilts to Korea/China.

China — Unitree, AgiBot, XPeng Robotics — government plus capital, already at production scale. Korea trails on capital and headcount.

Japan — Toyota, Honda, Sony — incumbent robotics strength but weaker startup vitality.

EU — 1X Technologies, Optimus camp — capital available, manufacturing depth missing.

Korea's edge: semi/display supply chain plus government policy. Cost efficiency and production speed are the real advantages.

Skeptics, by name

Sungju Kang (former NIPA director) acknowledges the LLM-to-physical-AI shift but cautions Korean players can't win global humanoid competition outright due to capital gap.

Kihun Kim (Partner, DSC Investment) — Series A pricing is rational, production-stage capital remains a separate challenge mid-term.

Both grant policy support's effect. Doubts focus on global market competitiveness.

Stakes

• Wins: Korean robotics/physical-AI startups — capital availability up, production financing in reach. Government — policy effect visible. Korean semis/display — robotics integration diversifies revenue.
• Loses: LLM wrapper SaaS — saturation pressure. Some US/Chinese humanoid players — Korean rise diversifies mid-term share.
• Watching: MOTIE — robot production infra support. Global VCs — Korean physical-AI Series B+ participation. Chinese government — countermoves against Korean robotics.

What changes

Devs: Korean robot startups bid up SW/AI/simulation engineer comp. Korean robotics engineering market reactivates vs US/China.

Founders: capital signal in physical AI. Pivots from LLM apps make sense for some.

Investors: Korean physical-AI category becomes a mid-term priority. Bridgewater, Tiger Global Korea entries possible.

Consumers: minimal short-term effect. Two to three years out, robot adoption accelerates in Korean manufacturing and services — restaurants, logistics, cleaning.

3-Line Summary

• Korean startups raised ₩72.16B in 29 deals in late-April week.
• Loai ₩13B + Lobros ₩10B — physical AI took 32% of the week.
• Capital center of gravity shifts from LLMs to physical AI.

Sources

• Startup Recipe — late-April funding roundup
• Startup Recipe — Momocall, ThePict, etc.
• ZDNet Korea — ₩60T strategic-tech program
• Korea.kr — ₩800B SMBA AI budget
• TechCrunch — Korean robotics rise