Data, Dreams, and Drive: Startup Advice from Early Backer of 17+ Unicorns
Q&A with Larry Li, Founder and Managing Partner of Amino Capital
This story is part of the Entrepreneurship of Life series, a collection of interviews with immigrant startup founders, venture capitalists, and tech business leaders in the US.
INTRODUCTION
Larry Li, or “Master Li” as many came to call him admiringly, is the guiding force behind global venture powerhouse Amino Capital. From his immigrant beginnings to a key player in Silicon Valley, Larry's journey is all about pursuing dreams, demonstrating drive, and – in terms of tactics – harnessing data.
In a checkered shirt, khakis and sneakers, Larry is as down-to-earth as they come, yet his achievements speak volumes. In barely over a decade since its Day 1, Amino has championed 17 unicorns – often from their earliest days – among other success stories.
A few months ago, Larry and I had a long conversation. We covered topics such as:
From engineer to founder and investor: How Larry’s early experiences shaped his approach to investing.
Data as a moat: What this core philosophy means for Amino, and how businesses can build a moat around data.
Secrets of the startup game: Insights on pivoting, founder-market fit, timing, and fundraising.
Two sides of generative AI: Larry's take on its challenges and opportunities.
The power of founding teams: Co-founder dynamics that make or break a business, and how entrepreneurs can find the right partner.
The immigrant edge: The connection between immigrant founders and startup success, as evidenced by Amino’s portfolio.
… and so on.
Let's dive in!
Larry Li is a Founder and Managing Partner at AMINO Capital, a global VC firm based in Palo Alto, focusing on data-driven SaaS and generative AI. Larry has steered investments in hundreds of startups, including 17 unicorns, 25 successful exits, and over 30 companies valued over $100M, such as Chime, Webflow, Rippling, Grail, Weee!, Replit, Turing, Dfinity, OmiseGo, Wyze, and Beacons.ai. Prior to founding Amino, Larry had initiated a fund that invested in ZOOM’s initial round in 2011.
Larry has been featured on the Midas Seed List, the TechCrunch List for First Check VCs, and Forbes’ Most Notable Chinese American Businessmen, among other recognitions. He is a sought-after speaker on global innovation, startup mentor for the StartX incubator program at Stanford, thought leader with over 600K followers on social media, and the author of best-selling books “Logic of Investment” and “VC, Demystified”.
Larry received bachelor and graduate degrees in engineering and business from Tsinghua University and the University of Florida.
(Note: Bolded questions below were from me, and the rest were summarized responses from Larry.)
I. FROM FOUNDER TO FUNDER
“Our innovative ideas and great engineering weren't enough; understanding what customers really needed (and what they didn’t) was what mattered.”
Before Amino Capital, you had quite the journey. Could you share key learnings from your background, especially your own founding experience, that shape your investment approach today?
Absolutely. I was blessed with a solid technical and business education from Tsinghua University, studying engineering, economics, and management. I later came to the US and explored the business world by working for companies of various sizes.
Co-founding a mobile security startup with a few Tsinghua alumni marked a key transition in my career. We took on a slew of challenges—from breaking into enterprise markets as new immigrants (very tough, as we concluded) to building for fickle consumers. These experiences provided invaluable lessons.
At one point in the early 2000s, we developed an online video sharing platform but didn’t seriously pursue it. At the time, it was generally expensive to store and share videos online, and we figured consumers wouldn’t pay for the service. We weren’t “wrong”, but we missed the point. In less than a year, YouTube went viral with very short and low-resolution videos. This was eye-opening for me: how did we miss the shot? Our innovative ideas and great engineering weren't enough; understanding what customers really needed (and what they didn’t) was what mattered.
YouTube correctly assumed that limited video length and quality would not deter its initial users. Granted, its networking and storage costs were too high for a standalone platform, as we figured. However, we all saw what happened soon after - Google acquired YouTube for $1 billion and subsidized them for years, before they started “printing money”. YouTube took a gamble by focusing on user engagement over foreseeable economics, and it paid off big time.
This taught me that successful startups often defy conventional business logic, and that investors can't rely solely on analytical thinking. Conclusions from a “logical, business-school style” analysis can be trumped by an intense (or even a bit “insane”) focus on the customers.
I have later applied this epiphany in my investing. Zoom is a case in point: we backed their seed round in 2011 out of an angel fund I had created with a few Tsinghua alums before Amino.
At the time, video conferencing was generally considered an enterprise service only (because “why would personal users need more than FaceTime and Skype?”), and the enterprise market was dominated by the likes of Cisco. Therefore, betting on Zoom’s shot at the mass market went against established views.
However, we recognized the potential for mass market penetration given increasing internet speeds, and we believed Zoom could fill that void by being dead set on user experience. They did conquer the mass market – and went on to become a major enterprise player also. We were right to set aside conventional wisdom.
II. DATA AS A MOAT
“Fill real gaps at unglamorous and overlooked places; constantly grind, ship, and iterate for customers. That’s how you get the data flywheel spinning.”
"Data as a moat" is a core investment philosphy of Amino. How did this belief develop?
From our respective backgrounds, our entire founding team share a deep appreciation for data's power. My first job in the U.S. was in data management at the country's then fourth-largest bank. My co-founder, Dr. Huican Zhu, was the brains behind Google’s image search and laid research foundations for their Asian language searches. Another co-founder, Dr. Xiaoliang (David) Wei, was VP Engineering at Facebook. Their work all relied heavily on advanced data analytics and modeling.
While some startups and VCs bet on frontier technologies, we prefer companies that build new applications of an existing technology and develop an edge from amassing and leveraging data. You could sometimes replicate a credentialed team of “deep tech” experts relatively quickly, but you can never easily copy proprietary data accumulated over years.
Google’s long-time dominance in search is a classic example of a data flywheel. Facebook and LinkedIn got to their size owing heavily to their login-gated communities. This way, they’ve managed to assemble valuable proprietary data that Google can't index.
What should founders do to create a data moat around their businesses?
It comes down to two things: relentless execution, and a laser-focus on customers. Sounds straightforward, right? But few pull it off.
Let’s start with execution. To build a data moat, you must continuously out-execute and out-innovate your competitors. Data perishes, and not all kinds of data are born equal. To make sure you keep collecting the freshest, most relevant customer intel, you must consistently provide the best customer experience and never stop pushing your product forward.
This is particularly true today with generative AI. Most companies are building on top of a third-party foundational model, so think about why customers should engage with your product and give you data, as opposed to using a competitor or even GPT-4 itself?
Now, about customer focus. A common mistake is to be obsessed with concepts in one’s head and over-build (or worse, build cool stuff that nobody really needs). If you start by listening to the users, what they need might not be fancy at all. Like I mentioned, the early YouTube wasn't a technical masterpiece. Many techies wrote it off, yet users couldn't get enough of it.
By the same token, founders must not avoid nitty-gritty, 'dirty work'. User needs can often seem trivial , but they're goldmines for those willing to look. Consider how Airbnb's founders started — they photographed host properties themselves.
At Amino, we've put money behind this philosophy. Chime grabbed our attention when they set out to serve low-income young adults overlooked by traditional banks, with aspirations to build a digital-native, full-service bank and eventually a data-powered FinTech conglomerate like Ant Financial. Chime is a $25 billion “decacorn” today.
Then there's Replit, a buzzing site for developers with 25 million users and a $1.6 billion valuation. They initially lured in young coders with free services and handy collaboration tools for their open-source hobby projects. Traditional developer platforms did not provide or specialize in these. By zeroing in on this community’s needs, Replit has gathered billions of lines of code and stacks of documentation. This wealth of data puts them in the driver's seat for AI code generation – one of the most promising LLM (large language model) applications right now.
These guys realized it's not about having the flashiest tech; it's about filling real gaps at unglamourous and overlooked places, and constantly grinding, shipping, and iterating for their customers. That’s how you get the data flywheel spinning.
Generative AI has caused tremendous excitement, FOMO, and increasing concerns over privacy and data security. Some enterprises forbid employees from using GenAI tools; many are wary of AI labs using their sensitive data for model training or leaking such data by accident.
What do all of these mean for AI startups trying to build a data moat?
These challenges are real. While not necessarily new, they are amplified by the accelerating pace of data accumulation with GenAI. But so are opportunities. Every Google search collects data from you, but like it or not, how many of us can quit using Google? When your service is valuable and irreplaceable, and assuming you handle data responsibly, users come despite those concerns.
Another way to view this is that every concern people raise for a new technology presents new business opportunities. For example, to address fears of AI labs possibly mishandling user data, many vertical AI applications use models like GPT-4 but protect customers by anonymizing their interactions with these models. This way, the applications serve as a “buffer zone” to guard sensitive client information, while building data moats themselves.
Our portfolio company Reality Defender is another such example. They specialize in identifying deepfakes across image, audio, and video. The more they’re used, the better they get. Many emerging businesses like this are mitigating GenAI’s side effects and building data moats at the same time.
I heard that you guys at Amino use data heavily for your own operations as well? How does that work?
Correct, we must practice what we preach!
We built an in-house database initially to keep up with the sheer volume of opportunities we came across. Imagine potentially great investments slip by just because there are too many to remember—we aren’t okay with that.
So, we ditched the individual notes and combined everything into what we call Amino Resource Management (ARM)—it's our take on a CRM system. It records the nuts and bolts of every meeting, both among our team and with founders—key takeaways, intel, and impressions.
Ever since the pandemic, ARM has turbo-charged our virtual meetings. Picture this: sitting in different time zones, we’re chatting with a founder on one side of our screens, while reviewing data points and sharing thoughts in real-time on the other, via our system.
Whenever we wrap a meeting, we're clear on the next steps, or we set reminders to circle back on specific issues—sometimes 6 months or a year later. It’s our organized way of nurturing founder relationships over time. Sometimes playing the long game is the smart move, and our system ensures nothing falls through the cracks.
This tool isn’t just for meeting notes; we're scooping up public intel on our targets too—tweets, LinkedIn updates, news—you name it. Before we jump back into a conversation with a startup, we're already clued in on their latest.
Better still, as our database grew, it started serving up patterns pointing to startup success. Do founders who dropped out of college do better or worse on average than founders who finished it? We ask questions like this to our database with tens of thousands of startups and founders. Interesting answers emerge.
ARM began as manual labor (the kind of “dirty work” well worth doing), but it’s increasingly automated. Our homegrown solution is now an essential part of our investment decision-making.
III. THE DOUBLE TOOLKIT ADVANTAGE
“Success requires not just inspiration, but also the humbleness to learn and the flexibility to adapt.”
Amino has backed a number of unicorns that have immigrant founders. You even have a running joke to pick founding teams where someone speaks English with an accent. How much do you think an immigrant background contributes to a startup's success?
A lot. Out of the 200-plus investments we've made, 17 have become unicorns, and 80% of those were founded / co-founded by immigrants, mostly first-generation. It’s not just a coincidence.
Statistics is part of the story – over half of Silicon Valley are immigrants after all. Then there's the fact that those who got here usually mean business – they’re gutsy, resourceful, often packing some serious education. But here’s the key: immigrant founders operate with a double toolkit. They blend insights from back home with what they picked up here, and that combination is killer for problem-solving.
People used to think being an immigrant founder meant more hurdles, but what we’re seeing suggests that immigrant founders are not just keeping pace, they’re often racing ahead.
Everyone knows about Elon Musk of South Africa, Satya Nadella from India, or Sergey Brin of Russia. In Amino’s own mix, we have Weee! – America’s largest online Asian supermarket, launched by Chinese immigrant Larry Liu; Replit – helmed by co-founder and CEO Amjad Masad from Jordan; Webflow – whose co-founder and CEO Vlad Magdalin emigrated from the former USSR as a refugee at the age of 9; and then there’s OmiseGo – co-founded by Jun Hasegawa, a Japanese immigrant in Thailand, it became the first billion-dollar ICO after Bitcoin and Ethereum.
These are entrepreneurs from all over, but they share this common trait: they bring a richness of perspective that sparks creativity. That’s really the edge, what gives immigrant-led startups the inside track to success.
Many immigrant-founded companies are inspired by successful businesses in the founder’s home country. For example, Weee! initially built the American version of “community crowd-sourced shopping”, like Pinduoduo had done in China. But imported ideas do not always make it.
My question is: what’s the trick to successfully localizing foreign business models?
Indeed, immigrant entrepreneurs often bring ideas from their motherlands, but success ultimately hinges on continuous learning and adapting.
Like you said, Weee! got off the ground taking inspirations from early Pinduoduo – it relied on so called “community leaders” to bulk order groceries at discounts and distribute them among friends and neighbors. As Weee! expanded, however, it constantly evolved its operations to better serve the US market.
For instance, the US population density is generally lower than China’s. When people started buying more from Weee! including perishables, the “community leader” model could only go so far. Weee! began to hire truck drivers, and over time they also built warehouses, but they had to take each step gradually and deliberately, making sure the economics worked.
Today, with over 11 million users, Weee! looks very different: no more community leaders, but the platform retains a social element at its core – people post pictures, comments, and even recipes for their grocery purchases, and they share coupons with friends. Weee! has localized – with a unique character that keeps its customers coming back.
Every market has its subtleties, and transplanting ideas without considering the varied contexts is a recipe for failure. Success requires not just inspiration, but also the humbleness to learn and the flexibility to adapt.
IV. THE HARD THING ABOUT HARD THINGS
“[An early-stage founder should] constantly ask yourself two questions: why me? And why now? Unless you have a good answer for both, you’re aiming wrong and need to pivot.”
When a startup is struggling, as its investor, how do you decide whether you should continue your support, or make the tough call to cut losses?
Larry: 99% of the time, we wouldn’t write off a company that we had trusted enough to back, as long as it's still trying. Entrepreneurship is brutal. Many of our unicorns have hit rough patches at some point, where cash was running out – this is way too common for any startup. But until the last dollar is spent, there's always a chance for a turnaround.
Several times, we have put more money in when other investors have thrown in the towel. One such example that worked out really well for us was Chime Bank – at one point they needed cash and literally no one else would write a check, but we decided to double down with a follow-on investment. They survived, and then became a breakout. Our conviction paid off.
That said, there are times when we must step back. We’ve seen founders, talented as they might be, suffer health and family crises under the intense stress of running their business. In such cases, I've advised them to prioritize their own and their family’s well-being, even if this means we as investors would take a loss. The entrepreneurial lifestyle isn’t for everyone. When necessary, it's more humane to call for an early stop, before severe and permanent damages are done to both the entrepreneur’s personal life and the investors’ capital.
What advice do you have for entrepreneurs on when they must pivot and how?
Calling a pivot is tough. We have seen many companies pivoting too late and exhausting their runway. My suggestion to founders is to constantly ask yourself two questions: why me? And why now? Unless you have a good answer for both, you’re aiming wrong and need to pivot.
“Why me” is about “founder-market fit”: what unique skills, strengths, or resources do you have to solve the problem you’re targeting, better than the next person could?
“Why now” is about timing. If there was an opportunity for this business, people would have done it last year, or even five years ago. If not, there’s a reason for that. Has the barrier now been lifted—due to technological progress, regulatory changes, or something else?
For example, iPhone only became viable with 2G mobile internet; Instagram with 3G allowing mobile image transfers; TikTok with 4G and 5G making mobile video uploads feasible. Seizing these windows of opportunity has real promises.
At Amino, we prefer to support companies that sail with technological tailwinds, rather than those mired in years of R&D. An entrepreneur needs continuous progress to secure funding, inspire the team, and attract customers. An ambitious long-term plan without tangible near-term milestones won't cut it when resources are scarce. Many academics try to start companies by asking VCs for $20 million and five years to work on their vision, but most often that's just not realistic.
At the end of the day, whether starting out or looking to pivot, entrepreneurs should scrutinize potential ideas by asking “why me” and “why now”.
Let’s talk about startup fundraising. You’ve said that entrepreneurs should opportunistically raise money whenever the market supports it. Others have argued that fundraising “too much” could lead to excessive dilution and lax financial discipline. What’s the sweet spot?
Whether a startup is flush in cash or down to the last bucks, capital discipline is non-negotiable. Unless you watch it closely, money disappears as fast as water slips through your fingers—every team member loves a free lunch! This is a common pitfall, especially for first-time leaders. So, financial discipline shouldn’t affect fundraising plans; you must always be disciplined.
Dilution is important, but think of it this way: it is least dilutive to raise funds when your startup is far from needing the cash. It’s the same as Zuckerberg being offered negative interest rates on his home mortgage – banks are fighting for his business. Similarly, companies with traction and a decent runway get far better terms with less equity surrendered. Conversely, when you’re desperate for cash, painful dilution might be inevitable, or worse, it might be too late.
Fundraising early also gives you more options, because it allows you time to get to know more potential investors and build those relationships. You never want to have to take whatever money you can get, as bad investors can do as much damage as good investors can help.
So, the strategy is to be proactive and raise in good times. There's no such thing as too much money in the coffer.
Switching topics to co-founders. I know you like founding teams with a history of working together. What’s your take on “co-founder dating”, where founders seek partners outside their immediate network? What’s the key to finding the right partner?
Ideally, pair up with someone you've got history with, like a former classmate or work buddy—someone whose strengths and flaws are known to you, and who you trust and have rapport with.
Finding partners through “co-founder dating” can work out after some period of bonding and collaboration. Yet, it's much riskier given the complexity of co-founder relationships. The issue of equity splits alone has split up countless founding teams. Why should it be 60/40, or 51/49? It’s a fraught art and can only be navigated with trust, goodwill, and recognition for each other’s distinctive contributions.
The other vital thing for a founding team is complementary skills and temperament. Too much overlap in either aspect tends to result in conflict. The frequent failure of co-CEO arrangements (BlackBerry being an example) illustrates this pitfall.
In addition to complementing each other, co-founder roles need to fit individual strengths and personalities. For example, a CEO should generally be a good spokesperson for the company. If the CEO is quiet, while the CTO does speaking engagements all the time, it could lead to confusion, role overlap, and a falling-out eventually.
V. BUILDING BLOCKS FOR STARTUP SUCCESS
A light one for my final question: how did you all come up with the name "Amino Capital”?
Actually, that wasn’t our original name. We started out as “zPark Capital”, but quickly realized its drawback—we were always placed last in any list of VCs sorted alphabetically. So we wanted a different name that starts with an A, sounds credible, and is a real word.
Among investment funds, most good names are long gone. "Amino" stood out because it was distinctive in finance, less familiar to people and therefore still available. Funny enough, some friends did assume we had moved into biotech investing because of the name change!
But above these considerations, 'amino' resonates on a deeper level—it represents the essential building blocks in biology. Just as no advanced life form could survive without amino acids, no startup could succeed without a few robust foundational elements. That is our mantra at Amino Capital.
Keyi (Author): If you enjoyed this story, subscribe (for free) here to my interviews of immigrant founders, VCs, and operators, and you’ll get new stories in your inbox. Also — why not share it? :)
Find me on LinkedIn or Twitter, and leave any comment below!