A VC firm from the crypto world says AI is too crazy, and they're staying conservative.

@Chen Zhiyan

A lot of people are saying that in certain ways, the AI industry is starting to look like Crypto.

Jingwen of Impa Ventures — someone who actually worked in Crypto — says that compared to today's AI, they even look conservative.

Jingwen describes herself as a '93-born woman who aspired to VC, "accidentally" ended up in Crypto, and stumbled into her first bucket of gold.

At the peak of FBG Capital, a crypto fund, she lived inside that accelerated world of encryption. Millions of dollars could be deployed in the span of a conversation. A single story could grow wildly and explode into gold.

In 2024, she founded Impa Ventures with Shiran and James (Liang Jie). Shiran was Jingwen's colleague at FBG; James was Shiran's former colleague at China Growth Capital.

Jingwen says they are a Problem First fund — look at the problem first, then the solution.

Impa Ventures currently has $50 million under management, focused on early stage. So far, 8 of the 9 projects they've invested in are B2B.

Three months ago, Jing Liu shared part of this story in his "China Investment Chronicles" series. At the time, he spoke with James about the inner world of an investor who missed Pinduoduo. We've placed that section at the end of this article; it helps you understand the temperament of this new fund's team. Recently, we spoke with Jingwen and James again.

This is the fifth new fund story presented by elsewhere, following Nebulon Ventures, Source Code Rhythm, Creek Stone, and Tiny Fund.

AI Seems to Be Playing Crypto's Game

elsewhere: Before deciding to launch an AI investment fund, you spent a long time in crypto investing. Can you tell us that story?

Jingwen: I graduated from Schwarzman College in 2017. Coincidentally, my thesis was crypto-related, so I talked to a lot of people. Someone thought I was job-hunting and introduced me to Shuoji Zhou, the founder of FBG Capital.

At the time, the crypto circle had trouble recruiting people with proper backgrounds. I was from NUS computer science and Schwarzman College, so I went for the interview.

The process was absurd: the boss was over an hour late, sat down and asked what salary my existing offer was. I said 7,000 SGD, and he directly said "I'll double it." My feeling was, this place is simply full of dumb money.

elsewhere: But the salary still tempted you.

Jingwen: My original offer didn't start until September. I figured I graduated in June with nothing to do, so I'd try this for three months and see what these people were all about.

It turned out to be the absolute peak of wild growth. FBG was a top Asian blockchain fund at the time, with LPs including Hongshan and Ribbit Capital. The boss himself was very wealthy, so they invested quite casually. My job was flying around the world to various events — Berlin, San Francisco, Tokyo, Singapore. In the office, sometimes I'd run into the boss in the hallway, say I liked a project — a few hundred thousand, a million dollars, two or three sentences, and the money was deployed.

elsewhere: How long did this peak last?

Jingwen: Only about half a year. By 2018 the bull market was ending; by 2019 it had fully turned into a bear market. My work changed 180 degrees overnight: from giving out money to asking for it back. Projects we'd invested a million in, we'd try to get 500,000 back. That was it. After that, basically no more investing.

elsewhere: Looking back now, what kind of world was Crypto?

Jingwen: Crypto is a world with the accelerator pressed down. It's essentially the development process of a new technology from emergence to落地. There are many similarities to what's happening in AI now.

Another point that's somewhat similar now: stack the team, hype the narrative, pump the data, then exit. In today's AI game, the bag-holders at the end have been replaced by institutions and LPs, but the game logic hasn't changed.

Back to Crypto — I don't think that world has passed. It's moving in a more solid direction: real applications, on-chain fundamentals, especially the combination with AI. There are many directions I think are genuinely promising.

We just don't want to replicate that game in AI. Impa started with our own money — we treat our own money and external capital with no difference. Only this way can we more purely search for true Alpha.

We Are a Conservative AI Fund

elsewhere: Tell us about your fund.

Jingwen: Starting from 2024, Shiran, James (Liang Jie), and I built this AI-focused early-stage fund together. Shiran and James were colleagues at China Growth Capital; Shiran and I were colleagues at FBG. We've already invested in 9 projects.

"Chinese founders + global market" is the major opportunity our era has given us, and it's still early. Day one global is our fund's biggest characteristic: Shiran and I are normally based in Singapore, James is generally based in Shanghai. Of our 9 investments, 3 are in Shenzhen, 2 in Singapore, 1 in the US, 1 in Sydney, 1 in Shanghai.

elsewhere: Why Impa Ventures?

Jingwen: Impa is an NPC in The Legend of Zelda. When the protagonist begins the main quest, Impa gives the map and guidance. She's a critical link in the protagonist accomplishing their mission, but she's definitely not the protagonist. What we can do is, at their earliest stage, become one link in helping them get things done.

This understanding is actually the starting point of our entire investment methodology — since founders are the protagonists, what we need to do isn't bet on narratives, but find the person who is genuinely solving a problem.

elsewhere: As a new fund, you must get asked a lot: what's your differentiation?

Jingwen: Our starting point for looking at projects is somewhat different from most funds — we don't start with sectors, we start with problems. So naturally, what we end up investing in is also different.

We're true AI Believers. We believe AI is reshaping the underlying infrastructure of many industries. This isn't narrative; it's really happening. But we've also been through that Crypto cycle — seen how bubbles grow, and how they pop. So we know the difference between spray and ocean.

Believe in AI, but don't believe in many narratives waving the AI flag.

elsewhere: Specifically, how do you distinguish spray from ocean?

Jingwen: AI Believer, but Skeptical. Long-term optimism about the industry, skepticism about specific projects.

The judgment manifests in several things:

• Problem First, not Narrative First.

Start from the problem, not from the sector. We don't first set an "AI+X" theme then go find projects; instead, we first see a real, underestimated pain point, then ask: who is solving this problem? Does the solution hold up? Founders who can answer these two questions well matter more than any glossy background.

• Global by default, not Global by design.

In college, I did a half-year VC internship in Israel. That was the first time I truly understood how venture capital operates, and also where I saw a batch of founders who squeezed global ambition out of a tiny domestic market. Israelis do business naturally day one global. That experience influenced me deeply; it's also when I decided to do VC.

Later doing crypto investing, many projects I invested in weren't Chinese teams but the true global developer ecosystem — those years flying to Berlin, San Francisco, Sweden, meeting many excellent developers, many of whom have also moved into AI.

• Young but experienced.

On our team, James is the only post-'80s; the rest are post-'90s/'95s/'00s. No historical baggage, but all three partners have been through complete cycles and have successful exit experience. People who've seen bear markets judge "real demand" and "bubble narratives" differently.

elsewhere: In AI, what game do you want to play?

Jingwen: Our methodology in four words — "respect common sense," especially business common sense. Some might think we're too conservative.

Many people now look at founders through "colored bubbles," thinking as long as they're from a big company or a prestigious lab they'll succeed. But what value have you actually created for customers? Does your productivity improvement really cover costs?

Impa cares more about value creation.

elsewhere: What is an unsexy but commonsensical project?

Jingwen: We invested in a project called "Lightyear Reach." The founder met with over 100 investors; everyone thought B2B wasn't sexy. But after talking with him, I discovered he solved a real need for Chinese supply chain going global: many small factories have no overseas marketing capability, and he uses AI to help them automatically find customers and run campaigns. This is a typical "distribution intelligence" process.

Model companies created intelligence, but how to sell intelligence into specific scenarios — that's the opportunity for startups. These kinds of projects aren't fancy, but their business is solid.

Additionally in Australia, we invested in a medical Admin AI project. It doesn't touch diagnosis and treatment, only solves the most tedious administrative workflows: scheduling, triage, summarization. Overseas, these processes were entirely handled by humans answering phones; now AI can achieve full online workflow.

We prefer projects that first solve existing, certain "pain points," rather than fantasizing about ethereal needs.

elsewhere: VC is a business of betting on excess returns. Can this approach achieve that?

Jingwen: For example, we invested in an on-device digital human company, betting on the 3D route for real-time digital humans. The digital human theme isn't unfamiliar to capital markets, but our logic for liking this company is simple: they hand rendering over to users' own devices, with the cloud only transmitting drive data, reducing costs by 99% compared to mainstream cloud solutions.

We estimate that serving just a few hundred thousand DAU overseas users can run to $10 million ARR. The team has top scientists in the 3D digital human field, aiming to achieve Cai Haoyu's Anuttacon LPM video model effects, but running on mobile phones and PC browsers. The prospects for multimodal AI and social entertainment scenarios are huge.

Our core is to find the extremes in the Power Law. Businesses that can only grow to a few hundred million dollars in scale — we simply won't make a move.

elsewhere: Your dreams are quite big...

Jingwen: In college I had a pretty funny dream: I wanted to make the Midas List.

elsewhere: That's many VC investors' dream.

Jingwen: I learned about the Midas List in college. It's not that I thought it would be so amazing after getting on. My logic was: primary market investing is what I like doing, and I hope to reach a milestone in what I like doing.

Now that so-called "dream" seems pretty silly, but often people actually have no goals, so having a "funny" goal is still better than having none.

Conversation with James from Three Months Ago

@Jing Liu

When I first started updating the "China Investment Chronicles" series, I wrote about someone. The title at the time was: The Forgotten Person in Pinduoduo's Capital Myth.

The story was roughly: a Hongshan investment manager who early on knew Colin Huang, then repeatedly brought Pinduoduo (then still called Pinhaohuo) to IC. But due to various circumstances, it never got approved. Until half a year after he left, Hongshan invested. This critical round in Pinduoduo's history later became a legendary investment for Hongshan.

A hundred-billion-dollar return, related or unrelated to you — that's the difference of this half year.

This person was Liang Jie. Between 2012 and 2016, he worked at Hongshan, and it was during this period that he saw Colin Huang and Pinduoduo.

To be fair, in the investment industry, this kind of story isn't rare. As long as you haven't made partner — or even managing partner (GP), 9 out of 10 investors probably have a basketful of grievances.

Yet Liang Jie is one of the most impressive people in my memory. This even became a long-term narrative about him.

A few years ago, he came from Shanghai to Beijing on business, and we met in the lobby of the Westin at Liangmaqiao. It was already late at night when we met. Originally I wanted to chat about industry topics with him, but somehow we quickly got to this story, and he casually pulled out the Pinhaohuo-era BP from his phone. His memory could be precise down to the layout and wording of every page.

After leaving Hongshan, he went to two more funds, and later even ran a fund focused on出海. But they could only be described as unremarkable years.

Perhaps because when I met Liang Jie I was also new to the investment industry, and among a pile of legends, this inverted story occasionally emerged, making it hard not to remember more.

Around 2020, the entire China VC industry went all-in on B2B. At the time Liang Jie was still insisting on looking at B2C and platforms. I remember him saying: if investing in China was left with only B2B opportunities, then he wouldn't do it. That wasn't his conviction.

elsewhere: I didn't tell you in advance that I was going to write about you, and this isn't exactly a "good story." What was your reaction when you saw it?

Liang Jie: A bit surprised. But I don't want to be labeled as "unlucky."

First, I don't think luck explains everything. It must be that my own accumulation wasn't enough — such as ability to handle things and relationships, such as conviction not being strong enough. Second, luck is important, but it also needs accumulation. For those at the table, as long as you don't leave, there's always a chance.

elsewhere: If the present you went back ten years, would you be more confident pushing Pinduoduo (Pinhaohuo) through IC?

Liang Jie: Definitely. Compared to 10 years ago, whether in business understanding or how to push things forward, there's been some improvement.

elsewhere: Someone commented: since you believed so much, why didn't you later buy Pinduoduo stock on the secondary market?

Liang Jie: In 2015, as an early-stage company investment, and buying stock on the secondary market after the 2018 IPO, are two time periods and two completely different things.

Perhaps the better question is: if I was so bullish, why didn't I join Pinduoduo, or invest a bit myself at the time?

Frankly, I really didn't think about it then. One, the $600 million valuation compared to the cash I could deploy at the time — it didn't seem to matter much. But this also conversely shows I really didn't understand that well — I didn't think it could become a hundred-billion-dollar company. After leaving Hongshan, I also talked with Colin (Huang) about other possibilities; he introduced me to LPs, joining his fund, and so on.

elsewhere: Have you regretted leaving Hongshan? For example, if you'd stayed another half year or year, maybe it would have been a story related to you.

Liang Jie: In mid-2015, a relatively major turning point happened, and it was also one reason I left.

That August, Hongshan had an offsite, and the conclusion was: venture's focus would shift from B2C to B2B, and verticals like education and healthcare. I was reluctant in my heart.

To be fair, during the period I was pushing it, Neil Shen's reaction was still positive. But honestly speaking, did I have conviction that Pinduoduo could succeed at the time? Probably not really. I just felt the story was big, growth was fast, the person was exceptionally strong.

elsewhere: How long did it take you to calm down?

Liang Jie: After going to the next fund, it was actually fine. Every stage has more important questions.

Luo Xiang once told a story: after he became famous he got a lot of abuse, and he felt terrible. A friend asked him: after becoming famous you got a lot of unmerited praise, did you feel ashamed and terrible? He said no. Then why accept unmerited praise gladly but not unmerited abuse?

I think the standouts in this industry, or any industry, are the few who are both smart and hardworking and lucky; if we think we're not stupid and can still do things, yet compare or demand of ourselves against that extremely lucky minority, we're being too greedy.

elsewhere: Actually, stories of pushing to IC but not getting through are common in the VC industry. Why is your story so remembered (aside from because I wrote about it!)?

Liang Jie: Probably because of the magnitude it (Pinduoduo) later reached, making it impossible to avoid. Like you dated a girlfriend and it didn't work out, and later she became a huge star...

elsewhere: I remember you studied engineering. How did you enter the VC industry back then?

Liang Jie: I studied materials for undergrad, microelectronics for master's. During my master's I read The Great Game, and this book opened a window for me as an engineering student: so capital markets are so magnificent, and so important. A seed to do VC began to sprout.

Graduated master's in 2006, tried every possible way to get into VC. But there was never an opportunity. In the financial crisis year of 2008, there was a Swiss fund of funds Adveq that interviewed me for 5 rounds and almost gave me an offer. Their founder came to Shanghai once, and our chat at the Grand Hyatt Jin Mao remains vivid in my memory.

Later I spent $499 on a financial modeling training, because many people questioned my lack of finance background and not knowing how to build models. And I sent over 100 resumes to every China VC I could find. Finally in 2011, I got my only offer — Walden International.

elsewhere: Then Hongshan.

Liang Jie: Right. Of course only after leaving did I more realize that Hongshan at the time was already the center of the universe.

elsewhere: How have these years been?

Liang Jie: Running my own fund "Horizon Ventures," the scale didn't get built (didn't raise institutional money) — unwilling but had to accept — AI came but felt hard to participate — met like-minded partners and returned to the table — discovered the opportunity is huge and participants are few — grateful and expectant.

elsewhere: Do you sometimes feel you were just a bit short on luck?

Liang Jie: Many people smarter and more impressive than me have left this industry. I'm still participating in this tech wave, very fortunate.

Early-stage investing has long feedback cycles, and one success covers a multitude of sins. Everyone is waiting for that small-probability good luck, which is irrational and unhealthy.

Last year I went to Hangzhou's Big Lotus to watch China vs. Australia in football, lost 0:2. Coming out I ran into a TV station interview. The reporter asked why I was still smiling after China lost. I said, none of the Chinese players on the field today played particularly badly, they all performed to their level, especially Wang Yudong who played very well. Australia was clearly stronger than us, losing 0:2 I have nothing I can't accept.

Like a child taking an exam, normally a 70-level student, he scored 75, what do you have to be unhappy about?

elsewhere: Once he scores 75, you'd still want him to try for 90 though.

Liang Jie: Of course if he gets a 90 I'd be happier. But to expect that way is inappropriate.

elsewhere: As an investor, can you try to evaluate yourself?

Liang Jie: Can use football as analogy. Football is one of very few things where I enter flow state as soon as I play. Even when playing badly, I've never stopped. I've been playing pickup games in a small county town since middle school, playing until now.

On post-graduation teams I was a starter (basically a starter as long as I showed up consistently), but never someone who could decide the game. However, over time, I realized starting a few years ago, I gradually became someone who could decide the game.

Why is that? One, I have a will to win; two, while others declined very noticeably, my body could basically maintain or even improve; and my understanding of football deepened.

Analogizing to investing, I think I can keep doing this. I missed the opportunity for one-battle fame, but I can persist to the end, and can influence the game.

elsewhere: Finally want to ask you: this time, do you want 75, or 90?

Liang Jie: The vast majority of things in our lives are beyond our control: our birth, our IQ, our opportunities. If we truly achieve something, what we should thank is things outside ourselves. So-called talent is heaven-given, opportunity is era-given, truly cannot be forced, only gratitude and being oneself.

Cover image: Friedrich, $2, 1808–1810, Alte Nationalgalerie