Zeroth.ai

Zeroth.ai

Tak Lo founded Zeroth.ai in 2017 after observing that Asia's emerging AI founders lacked the specialized technical mentorship and pre-seed capital that Silicon Valley routinely provided. The accelerator established dual headquarters in Hong Kong and San Francisco, reflecting a thesis that Asian AI talent required direct exposure to US go-to-market discipline while retaining regional R&D advantages. Zeroth positioned itself as a bridge between Asia's deep engineering benches — spanning India, South Korea, Japan, and Southeast Asia — and the global venture ecosystem. Zeroth backs pre-seed AI and deep-tech companies with initial checks typically ranging from $20,000 to $120,000, combined with a three-month intensive program focused on product-market fit, technical mentorship, and investor readiness. The portfolio spans enterprise software, robotics, fintech, and digital health, with confirmed investments including Fano Labs, a Hong Kong-based speech recognition company that later raised from Horizons Ventures, and BizSeer, a Chinese AIOps platform. Co-investors across the portfolio include Sequoia Capital India, Horizons Ventures, and various Asian sovereign-linked funds — Zeroth often serves as the first institutional investor, de-risking deals for later-stage GPs. Zeroth operates with a lean central team led by Tak Lo, who previously built Techstars' presence in Asia. In 2022, Zeroth expanded its footprint by deepening ties with Japanese corporate venture arms and accepting a growing share of founders from South Korea and India. The accelerator has no disclosed philanthropic foundation or adjacent family-office structure. As of early 2024, Zeroth began publicly emphasizing dual-use AI startups — companies with both commercial and defense applications — reflecting Japan's and South Korea's increasing appetite for sovereign AI capabilities. Zeroth's structural differentiator is its chronological and geographic arbitrage on pre-seed AI: it was the first accelerator in Asia to demand that every cohort company have a machine-learning or robotics core, not merely an AI-adjacent label. This forces a technical filtering mechanism that generalist accelerators cannot replicate without specialist partners. The firm's enduring constraint — undisclosed fund size and opaque deployment totals — reflects the broader Asian pre-seed market, where many accelerators avoid public AUM disclosure to maintain pricing flexibility with early-stage founders.