Key Takeaways
- • Jeff Herbst, co-founder of GFT Ventures and former NVIDIA vice president of business development, says AI investing has become "a market where speed determines victory," according to an interview with Chosun Ilbo.
- • GFT Ventures, founded in 2021 by Herbst and CEO Eum Jae-hoon, manages approximately $160 million (214 billion won) and has invested in more than 20 companies, with backing from Mirae Asset and SK Hynix.
- • Herbst argues AI bubble fears are misplaced, claiming demand for high-performance computing exceeds supply by 10 to 100 times, with data center, power, and memory bottlenecks possibly taking up to a decade to resolve.
- • Herbst names agentic AI and physical AI (robotics) as next-generation investment themes, and sees application-layer AI companies in South Korea as more promising than proprietary model development.
- • Herbst says founder resilience is a key evaluation criterion, shaped by NVIDIA CEO Jensen Huang's view that failing to learn from failure is a bigger problem than failure itself.
Summary
The pace of startup development is accelerating as AI shortens product and growth timelines, and investor patience has shortened accordingly, according to Chosun Ilbo's interview with Jeff Herbst, co-founder of GFT Ventures. Herbst spent 20 years in AI ecosystems and venture investing, including roles at NVIDIA where he was involved in more than 40 venture investments and more than 20 mergers and acquisitions as vice president of business development. He said that choosing a company that fails to scale quickly is likely a wrong investment, noting that current generative AI companies are generating revenue faster than he has seen in his career.
GFT Ventures was founded in 2021 by Herbst and CEO Eum Jae-hoon, who previously worked at Translink Capital. The firm was launched with Korean capital from Mirae Asset and SK Hynix, manages roughly $160 million (214 billion won), and has invested in over 20 companies. Herbst said valuation expectations have risen: where a $1 billion exit was once considered a success, investors now expect early-stage AI companies to grow into multi-billion-dollar businesses. He argued that the current structure allows entire funds to be returned even at higher prices and larger check sizes.
Herbst rejected AI bubble concerns, arguing that supply constraints are the larger issue. He said demand for high-performance computing exceeds supply by at least 10 to 100 times, and that resolving bottlenecks in data center sites, power supply, and memory could take up to a decade. He also addressed fears that investment would flow primarily into GPUs and data centers.
His investment themes include agentic AI, where AI autonomously handles tasks, and physical AI such as robotics. In South Korea, he sees more promise in applications built on existing models in video, healthcare, and security than in developing proprietary foundation models. Looking a decade ahead, he identified personalized AI as a pivotal shift and said the key question will be whether platforms or individuals hold sovereignty over personalized data.
Herbst credited NVIDIA CEO Jensen Huang, whom he described as seeing "10, 20, or 30 steps ahead of anyone else," with shaping his approach to problem-solving and failure. He said that at NVIDIA, failing to learn from failures was considered a bigger problem than failure itself, which led him to weigh founder resilience heavily in investment decisions.
GFT Ventures was founded in 2021 by Herbst and CEO Eum Jae-hoon, who previously worked at Translink Capital. The firm was launched with Korean capital from Mirae Asset and SK Hynix, manages roughly $160 million (214 billion won), and has invested in over 20 companies. Herbst said valuation expectations have risen: where a $1 billion exit was once considered a success, investors now expect early-stage AI companies to grow into multi-billion-dollar businesses. He argued that the current structure allows entire funds to be returned even at higher prices and larger check sizes.
Herbst rejected AI bubble concerns, arguing that supply constraints are the larger issue. He said demand for high-performance computing exceeds supply by at least 10 to 100 times, and that resolving bottlenecks in data center sites, power supply, and memory could take up to a decade. He also addressed fears that investment would flow primarily into GPUs and data centers.
His investment themes include agentic AI, where AI autonomously handles tasks, and physical AI such as robotics. In South Korea, he sees more promise in applications built on existing models in video, healthcare, and security than in developing proprietary foundation models. Looking a decade ahead, he identified personalized AI as a pivotal shift and said the key question will be whether platforms or individuals hold sovereignty over personalized data.
Herbst credited NVIDIA CEO Jensen Huang, whom he described as seeing "10, 20, or 30 steps ahead of anyone else," with shaping his approach to problem-solving and failure. He said that at NVIDIA, failing to learn from failures was considered a bigger problem than failure itself, which led him to weigh founder resilience heavily in investment decisions.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.