Key Takeaways
- • Impa Ventures, a newly formed AI-focused fund founded in 2024 by veterans of crypto investing (Jingwen, Shiran, and James/Liang Jie), operates with $50 million in capital and employs a deliberately conservative "Problem First" investment thesis that contrasts sharply with current AI industry exuberance.
- • The fund's founders draw direct parallels between AI market dynamics and the peak of the crypto bubble, with Jingwen noting that today's AI sector appears more aggressive and speculative than even the most frenzied periods of crypto investing she experienced at FBG Capital.
- • Impa Ventures prioritizes B2B early-stage investments with eight of nine portfolio companies operating in the B2B space, explicitly rejecting what the founders characterize as inflated narratives and unsustainable valuation multiples prevalent in AI venture capital.
- • The fund's composition reflects experienced deal-making across both crypto and traditional venture ecosystems, with co-founder James previously at Huachuang Capital and team members carrying institutional discipline from earlier market cycles and high-profile misses (e.g., Pinduoduo).
Summary
Impa Ventures represents a notable contrarian positioning within the current AI investment landscape. Founded in 2024 by Jingwen (a 1993-born investor with deep crypto experience), Shiran (former FBG Capital colleague), and James Liang Jie (previously at Huachuang Capital), the fund operates with $50 million in committed capital and explicitly markets itself as a disciplined alternative to what founders view as irrational exuberance in AI venture funding.
The fund's founding thesis emerges directly from experience in the crypto sector during its peak. Jingwen, who graduated from the Schwarzman Scholars program in 2017 with a thesis focused on cryptocurrency, spent years within FBG Capital's accelerated ecosystem where major investment decisions could be made with minimal due diligence and story-driven valuations frequently produced explosive returns. Rather than applying this model to AI, Impa's founders have intentionally adopted opposing constraints: a "Problem First" framework that begins with identifying genuine business challenges before evaluating technological solutions, an approach designed to filter out narrative-driven valuations.
The fund's portfolio composition reflects this philosophy. Of nine investments to date, eight are B2B focused enterprises, indicating a deliberate selection toward projects with identifiable customer problems and revenue-generating potential rather than speculative consumer-facing or infrastructure plays. This selectivity stands in marked contrast to broader AI venture capital, where recent funding activity for major players like Anthropic and OpenAI has driven valuations to $800 billion and above, according to reporting from industry sources.
Jingwen's candid assessment—that AI has become "too crazy" compared even to crypto's most exuberant periods—signals an emerging divergence among experienced investors regarding sustainable valuation foundations in the AI sector. The fund's structure, combining veterans who navigated both the crypto boom and institutional capital markets, suggests a deliberate bet that disciplined, problem-driven investment theses will outperform momentum-driven allocations as the AI cycle matures.
The fund's founding thesis emerges directly from experience in the crypto sector during its peak. Jingwen, who graduated from the Schwarzman Scholars program in 2017 with a thesis focused on cryptocurrency, spent years within FBG Capital's accelerated ecosystem where major investment decisions could be made with minimal due diligence and story-driven valuations frequently produced explosive returns. Rather than applying this model to AI, Impa's founders have intentionally adopted opposing constraints: a "Problem First" framework that begins with identifying genuine business challenges before evaluating technological solutions, an approach designed to filter out narrative-driven valuations.
The fund's portfolio composition reflects this philosophy. Of nine investments to date, eight are B2B focused enterprises, indicating a deliberate selection toward projects with identifiable customer problems and revenue-generating potential rather than speculative consumer-facing or infrastructure plays. This selectivity stands in marked contrast to broader AI venture capital, where recent funding activity for major players like Anthropic and OpenAI has driven valuations to $800 billion and above, according to reporting from industry sources.
Jingwen's candid assessment—that AI has become "too crazy" compared even to crypto's most exuberant periods—signals an emerging divergence among experienced investors regarding sustainable valuation foundations in the AI sector. The fund's structure, combining veterans who navigated both the crypto boom and institutional capital markets, suggests a deliberate bet that disciplined, problem-driven investment theses will outperform momentum-driven allocations as the AI cycle matures.
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.