Why Do the World’s Top Investors Bet on Zilliz
By Zilliz on Nov 15, 2022
An interview between 5Y Capital and Charles Xie, founder and CEO at Zilliz
An interview between 5Y Capital and Charles Xie, founder and CEO at Zilliz
5Y Capital is a private equity and venture capital firm with a mission to make the world better by supporting entrepreneurs to build great companies. Its founding partner Richard Liu, one of the world’s best venture capital investors for the past decade, was named #4 on the Forbes’ Midas List this year.
5Y Capital was one of the early investors of Zilliz and led the company’s series A round in 2018. What did the venture capital firm see in Zilliz to give them enough confidence to invest in the then early-stage startup? What is so special about Zilliz? What does the founder see that others do not?? How should entrepreneurs _deal with _uncertainty when stepping into an untapped new world?
Charles Xie, founder and CEO at Zilliz, recently sat down with 5Y Capital and shared his journey of entrepreneurship.
01 To infinity and beyond
5Y Capital: You go by “Star-Lord” as your alias, right? Why did you pick Star-Lord?
Charles: It’s company tradition for everyone to choose their own hero, whether it’s a true historical figure or a fictional one. It’s actually the first thing we “task” our new employees to complete and has been ingrained in our culture.
I chose Star-Lord as my hero because the Guardians of the Galaxy, which he led, are one of Marvel’s few superhero teams. When you look at each member of his squad separately, none of them are as powerful as those popular superheroes, yet together they form an impenetrable team. And, more significantly, they become stronger as a team over time. Their team attitude is something I admire - working together to achieve the impossible.
And, unlike other Marvel heroes who have a home planet to go back to, Star-Lord and the Guardians always show up in their spaceship, ready to explore the enormous galaxy. They never stopped when one task was completed, but instead kept their gaze fixed on the infinite. I respect their curiosity in the unknown, and I aim to be as curious as they are.
5Y Capital: How did this culture come about? Is there someone else’s hero that impresses you?
Charles: It started as a fun thing to do. I had my hero, so I shared this idea with the team and they all loved it. “This is fun,” we thought. “Why don’t we make it our thing?” We also ask folks to explain why they chose their hero. One thing I find fascinating is that most individuals choose a character who closely reflects themselves. Groot, for example, was our first engineer’s hero. Groot is a selfless figure who consistently prioritizes the interests of the team over his own. And I notice the same traits in that engineer at work.
5Y Capital: Has your understanding of Star-Lord evolved throughout the years? How do you see this superhero now?
Charles: Pretty much the same, actually, but I’m not saying he’s perfect. I believe that each of us is endowed with unique talents, personality traits, and abilities, as well as flaws in various ways. Superheroes, too. As we progress, we become more aware of our limitations. Limitations we make every effort to accept and overcome.
Star-Lord is a superhero, yes. But he’s also someone that you can trust and build a long-term friendship with. You don’t think of him as high and mighty, or as cold and unapproachable; rather, he’s extremely down-to-earth. Over the last few years, I’ve come to believe that having that kind of mindset is also essential for starting a business. In the end, a few ephemeral moments of glory do not define you or your company. It’s the work you put in every single day, every decision you make, and every detail you shape that builds the company.
5Y Capital: You mentioned before about buying 100 identical black T-shirts at once to save time. Is that something you still do?
Charles: Yes, but now I humor myself with more colors. Even so, when I try to check out with 30+ t-shirts in my basket that are the exact same style, I still get this amused look from the cashier as if I’m not thinking straight. Some would tease me, “You know they’re not on sale, right?”
I’ve been practicing minimalist clothing ever since I started my company. It simply makes my life much easier. Instead of standing in front of my closet for 15 minutes every morning trying to figure out what to wear, I’d rather spend that time on more important things.
Being an entrepreneur isn’t easy. There are so many things to do and always a bigger challenge ahead of you. However, building a business is a marathon, and I believe it’s particularly important to set aside some time to rest and ponder. People tend to make poor judgments when they are overwhelmed. So, no matter how busy I am, I make time to read, meditate, listen to music, watch a movie, or play sports.
02 History reveals a pattern from a macro perspective
5Y Capital: What was the feedback when you first proposed the concept of vector databases? Were there questions and challenges? What was it like?
Charles: Nobody understood where I was coming from when I first established Zilliz in 2017 with the goal of creating a new database system for the AI era. Most investors I dealt with at the time didn’t understand why we needed another database, and the best I ever got was “Interesting idea.” It was rough. I was exhausted.
On a bizarrely warm Thursday in November 2017, I went to a meetup in Shenzhen, where I met Richard and other partners at 5Y Capital. We were talking about trends in AI and Big Data and the technology stack I envisioned for data processing in the AI era. What should have been a 60-min session ended up being a near 3-hour heated discussion. We started at 11 a.m., and by the time we were done, we realized it was already well past lunch. And a week later, I was looking at the term sheet on my desk extended by 5Y to lead our series A. Now that I think back, it all kind of makes sense. As 5Y’s slogan goes, “Where others see crazy idealists, we see visionaries.” They understood and believed in me, because they themselves are often seen as “crazy idealists”.
I soon realized, though, that fundraising was actually the easier part and that the real challenge awaited. When we started, even our own engineering team had reservations. We observed new technological breakthroughs in the AI era, as well as unprecedented data processing capabilities offered by distributed and heterogeneous computing. But it was unclear then what kind of products would emerge.
When we embraced innovation, we also opened up to a great deal of uncertainty and volatility. We had to navigate through diverse user needs in order to define the core features and functionalities called for in a database for modern AI.
We ultimately zeroed in on a new category in 2019 after nearly two years of testing our concepts with the market and users. Milvus was created as the world’s first open-source vector database that year. We were thrilled to see its user base quickly grow to dozens of enterprises just through word of mouth. It gave us more confidence that we were on the right track. Since then, we’ve been focused on vector databases, and we’re excited to see more people join us on this adventure. That’s fantastic, and we’re proud to be the trailblazer.
5Y Capital: Do you get anxious in the face of such vast uncertainty? How do you deal with it?
Charles: Looking at how technology has evolved through time from a macroscopic perspective, I am confident about the future of vector databases. Long-term patterns can only be unveiled in the study of many interconnected histories of science and technology over the course of human civilization. Certainly, it’s hard, if not impossible, to recreate specific instances of achievement in history. However, I believe that history reveals a pattern at the macro level, even repeating itself to some extent.
Before starting Zilliz, I had been working in the database industry for almost 15 years. The database market has seen the emergence of purpose-built databases during the last decade, breaking the monotony of a few general-purpose relational incumbents. MongoDB, for example, is used for documents, Neo4j for graph data, and InfluxDB for time series data.
I believe that the database sector, like many other industries throughout history, will witness more niche markets defined by specific user needs. In the AI era, we’ll see new forms of databases come under the spotlight, such as vector databases, and perhaps even quantum databases when we enter the quantum computing era.
What also helped with anxiousness was user feedback. Our vector database has been open sourced from its inception, and we have received a great deal of user feedback over the years. This has allowed us to swiftly validate our work in each product iteration and new feature release. Those that have proven to be spot-on, we keep and optimize; whereas for those that might have diverged from user demand, we would ditch and move on.
5Y Capital: Were there any difficult decisions along the way?
Charles: Milvus 1.0, which began with a standalone architecture, had taken off by 2020 and had approximately 300-400 users globally. The nimble design enabled us to achieve speedier roll-outs but wider adoptions on ten million-scale use cases. However, as the volume of vector data continues to grow rapidly, we predict a tenfold or possibly a thousandfold surge in the years to come. It soon became apparent that the standalone architecture’s limited scalability would make it difficult, if not impossible, for it to continue to compete on data at billion and even ten-billion scale.
So we were at a crossroads here. Do we continue to improve and enhance the old architecture, or do we start all over and redesign the whole thing in a cloud-native way? Given the growing popularity of Milvus 1.0, that was a particularly difficult decision. Switching to a different path would pose a great risk on the company as it was difficult to predict the time and money it would cost us. Looking back, I still think it was one of the most difficult decisions I’ve ever had to make. It was a matter of life and death for the company.
5Y Capital: What’s the decision in the end? Is the decision well-received?
Charles: It took us about three months to feel confident enough to make the decision. We went through countless experiments and prototypes to have each of our design tested before we were finally able to align everyone on the same page.
So we kicked off the rebuild in the second half of 2020. It gave us a head start in offering a highly scalable, cloud-native solution, and it propelled us to become the world’s first vector database startup with 10-billion scale search capacity.
In hindsight, I think what truly drove the decision we made is straight-up engineering thinking. We are a group of engineers who set out to create a better – more advanced and more user-friendly – solution for our users. It was as simple as that.
5Y Capital: Why did you adopt the open source model from the start? What was your first encounter with open source?
Charles: I came across the Apache web server when I was in college and later had the opportunity to contribute to the development of the Globus Toolkit, an open-source toolkit for grid computing led by the University of Chicago and the Argonne National Laboratory. Because of open source, a C.S. student in Wuhan, China and engineers in Illinois, United States could collaborate to advance technology. Isn’t that exciting? Furthermore, thanks to open source, we have access to some of the most cutting-edge technology from around the world and can even be part of the innovation.
Open-source has since always held a special place in my heart and later became a vital component to Zilliz’s ethos. “Talk is cheap. Show me the code.” We hope that with open-source we can make our technology accessible to all and create tangible business values for our users. To me, that’s really how we gauge ourselves as a company.
03 The hardest part is to acknowledge your own ignorance
5Y Capital: What do you find the most challenging in building a startup? Talents? Organization? Or navigating the market?
Charles: No, the hardest part for me is to acknowledge my own ignorance every single day but push forth nonetheless. Every day I wake up to yesterday’s mistakes and bitterness. Today I might be a tiny bit wiser, still not enough for tomorrow’s challenges, but I won’t be stopped.
This is also the biggest challenge the team has to confront along the way. Running a company is never smooth sailing. Depending on the stage and size of the company, how things are done varies: R&D, evaluation of product-market fit, organizational development, team management, culture cultivation. This is especially true when we are stepping into an untapped new world of technological innovation. We must constantly face our ignorance and improve our perception.
5Y Capital: So what’s your trick? How do you cope with the frustration of “facing your own ignorance every day”?
Charles: The first thing is to acknowledge it with full transparency and open ourself to the possibility of learning. I’ve built a regimen that works for me throughout the years. I start by talking to as many people as possible who are more experienced and knowledgable. Put things into perspective. So I know exactly where I fall short and where I need to grow.
Once that’s established, my extensive learning phase starts. I’m a book person – some people might prefer videos these days – so I find the best books on the subject and read them all.
The last stage is particularly important to me, meditation. I personally find meditating super effective in helping me fully absorb the information I acquired and make it part of my own.
These three stages form a full cycle of what I like to call “cognitive psychology iteration”. I try to complete one iteration every year.
5Y Capital: How have these iterations you completed in the past few years affected the way you look at the world?
Charles: Yes, one of my biggest revelations is probably the difference between low-dimensional and high-dimensional thinking and how it can define our success or failure. A quest unachievable in a low-dimensional universe would likely be easy-peasy once we arrive at a higher level. For example, once we’re on AP Calculus course books, those high school math problems would likely seem trivial. Similarly, in the software engineering world, a lot of code mistakes could be avoided once we pick up appropriate data structures and advanced design patterns.
In fact, whether it’s achieving our life goal or our business success, the key lies in improving our perceptual capability and moving beyond low-dimensional thinking. That’s why when in times of struggle, I always tell myself, “You only find it difficult now because you’re thinking in a low-dimensional realm.” Think about Steve Jobs. That man “put a ding in the universe”, while I’m struggling with the day-to-day operations of a startup.
With that in mind, instead of rolling up my sleeves and diving right into a challenge, I pause and reflect. I would push myself out of my comfort zone and think outside the box in search of the best solution. In fact, this is my process of progressing beyond low-dimensional thinking.
5Y Capital: But this process can be excruciatingly unpleasant, I assume?
Charles: You know, the process of developing higher dimensional thinking actually goes against human nature. The knowledge, successes, and experience we harvested over the years together form a thought system that defines who we are today. And that same system could at the same time be the largest roadblock on our path forward because we humans can be so easily trapped in our old thinking patterns. So to reach a higher level, we have to smash that old system into pieces, break out of the fort we’ve built around us, and give way to a brand new self.
So, to answer your question, absolutely, it is excruciatingly hard and definitely unnatural to acknowledge your own ignorance and brace up to overcome it. But it has to be done.
5Y Capital: But you don’t know for sure when or if you’ll reach the next level, do you? How do you deal with this uncertainty?
Charles: To me uncertainty is the ultimate driver for self-fulfillment because it often comes hand in hand with great opportunies. We are driven because we believe in “More pain, more gain.” Imagine, what would you do if everything in your life was already set in stone? Would you still have a goal at all?
Uncertainty about life, about what we can possibly achieve with our life, fascinates me. In fact, this is most likely why many people regard self-fulfillment as their life goal and strive so hard to accomplish it. “Life is like a box of chocolates. You never know what you’re gonna get.” It is uncertainty that continuously motivates us.
5Y Capital: What is your ultimate goal of self-fulfillment?
Charles: Stay curious. I know that people say it all the time, and it’s easier said than done. People tend to get more bigoted as we age. Even many prominent figures in history fell prey to dogma in their latter years. I hope that until the day I die, I am still brave enough to stay curious about the world and true to myself.
This article is a translation of the original source published by 5Y Capital. Source: www.5ycap.com
Official Zilliz family logos and assets for download.
For media and analyst inquiries.