When Alexandr Wang co-founded Scale AI in 2016 at age 19, his company’s name was something of a joke. After all, with only three employees, Scale had yet to scale up.
Now, the company has around 300 employees and is worth $3.5 billion, making Wang the CEO of one of the biggest AI companies in Silicon Valley. After dropping out of MIT and joining Y Combinator’s spring 2016 batch, Wang didn’t know what kind of startup he was going to create. He then had a breakthrough moment: companies wanted to integrate machine learning, but most didn’t have the tools or resources that Big Tech did for their in-house AI projects. That’s when Wang hit upon the idea for Scale, which makes AI infrastructure for developers to build from.
The company grew to unicorn status within three years, and has raised nearly $300 million to date. Clients include companies from a range of industries: OpenAI, Lyft, Nuro, Airbnb, Samsung, and Pinterest, among many more. According to Wang, every industry could be using AI in just a few years’ time.
The Business of Business spoke with Wang about Scale’s beginnings out of Y Combinator, his challenges as a founder, and what the future of AI will look like.
The Business of Business: I’d like to hear a little more about you and why you wanted to found Scale AI in the first place.
Alexandr Wang: Before starting Scale, I was a student at MIT. And in school, machine learning and AI were very exciting technologies. There was a lot of excitement, especially in this college atmosphere around this. I saw a lot of potential around artificial intelligence and machine learning, and really was excited about how it could change the world. But it wasn't yet making a real impact.
“One of the really cool things about AI is that data is the new code.”
MIT was a very engineering oriented school. Mechanical engineering majors are building catapults on the lawn, and electrical engineering majors are building robots. And computer science majors are building apps for their friends to use. But there was nobody building anything with AI, despite the fact that there were hundreds of students at MIT, all brilliant, very hardworking people. We're all studying AI. And when I dug into it, I realized that the data was the big bottleneck for a lot of these people to build meaningful AI. It took a lot of time and resources to add intelligence to data, to make it usable for machine learning. There were no standardized tools or infrastructure, there was no AWS, or Stripe, or Twilio to solve this problem.
I even discovered it firsthand, because I wanted to build a camera inside my fridge, so it could tell me when to refill my groceries and what I need to buy. Even for that I just didn't have any of the data to make it work. We saw at that point, there was this really big hole or problem that needed to be solved for AI to actually reach its full potential. That was the inspiration behind starting the company. Since then, obviously AI has only become more important and more impressive. A more top of mind technology, self driving cars have been this huge industry that has popped up, along with more recent advancements from OpenAI with GPT-3, and protein folding AI. It just continued to explode in this massive way, which has been really incredible to see.
So what exactly does Scale do?
Scale has built the data foundation for the future of AI. So what we've done is build this infrastructure that other organizations can build on top of to support their ambitious AI efforts. We're partners with all of our customers, which go everywhere from self driving car companies to companies like States Title that are looking to innovate old industries like mortgages and home owning, to research organizations like OpenAI. So we basically power other organizations to build their AI by empowering them with good data.
The problem we saw is that you couldn't build incredible AI if you treated data as an afterthought, because the issue with how AI works is that it's very much a garbage in, garbage out kind of paradigm. So if you have bad data in, you're going to get bad AI out. Let's say we're building a chat bot, for example, on bad data, then you're not going to get a performance system at the end of it. And what we solve for long these organizations.
What makes Scale different from competitors?
There really aren't that many companies tackling the full scope of building a data foundation for these customers in the same way. It's really pivotal to the success of any AI effort to ensure that they have a number of pieces to this puzzle, they need what's called data annotation, which is basically training AI with a very high quality graph, what's called ground truth data. It's this process of adding intelligence to that data. That's one big problem. There's another around infrastructure, and just being able to manage all of your data with an organization, so we built a product called Nucleus to solve that. We also built models as a service product where we allow customers the ability to easily build AI on top of this data foundation that we've laid for them.
“We need more standardization and more platforms that other people can build on top of. Otherwise, [AI is] still going to be a very, very specialized technology.”
What we've noticed is that we don't see a lot of other companies out there trying to tackle the full scope of the data foundation for AI. On the one hand, there's the big tech companies, like the Facebooks, and the Googles, and the Amazons of the world, and they've been incredibly successful at AI. But they've chosen to actually build everything in-house. They have lots of teams and lots of resources and deep expertise. But they're not actually bringing that to the rest of the world as explicitly. Then there's a lot of companies that solve individual pieces of the puzzle — maybe there's an organization that just solves data annotation as a problem, or just solves the infrastructure, like dataset management. But by doing that, they throw a lot of people at the problem and they lack the expertise for being able to solve these problems at a really high level and providing an integrated data foundation for these customers. So all we do ultimately at Scale is bring the deep AI expertise that has been built up at these big tech companies and make it accessible to everyone else in the community.
It makes sense that Big Tech would just do their own thing. They have thousands of people and the resources to do it.
Yeah, but it's so inefficient, right? We need more standardization and more platforms that other people can build on top of. Otherwise, it's still going to be a very, very specialized technology.
If you were to rewind back to 2015, before Scale was started, it really was just AI, it was just a game of the behemoth. If you fast forward to today, there's a lot of other companies and a lot of other organizations that are able to do AI and actually improve their products and build incredible things with it.
You recently said that Scale broke even for the first time since being founded in 2016. What did that path to profitability look like?
Historically, there have been a lot of organizations in AI that haven't built necessarily sound businesses. A lot of that has been building technology itself that is super, super exciting. You can build great technology, But I think because of our role, where we really focused on partnering with other companies in building their own AI, and being a foundation for the whole ecosystem, that's really allowed us to build Scale in a much more sustainable way relative to the last generation of AI companies. That's been a big part of why we've been able to be really sustainable in our growth and ultimately grow in a much more even-keeled way.
Scale just raised another round of funding. How are you planning to use that money?
We announced another round of funding last December, I believe, and it's super exciting for the company, a great sign of the momentum and the excitement around AI in general. Our goal is to be able to further our investments into our products. And by continuing to build, like I mentioned before. When you look at the world of AI, there's a lot of piecemeal solutions, and a lot of disconnected solutions. Our goal is to build as holistic of a data foundation and data infrastructure platform as possible for our customers. And so we really want to be investing into our product to be able to support that for a new generation of customers.
“I think the story of technology is one of exponential curves, of everything getting faster and happening faster.”
On Scale’s site, there’s a section about metaprogramming — that machine learning is a form of metaprogramming where the developer writes the code that writes the program. The site jokingly called it “demoralizing.”
I think one of the really cool things about AI is that data is the new code. With AI, it's actually data that builds the program, tells the program what to do. Let's take one of our customers, States Title, they're building an AI that's able to automatically recognize or automate a lot of the real estate process and bring that old school industry into the 21st century. A lot of what we've done with them is build AI that's able to automate the document workflows, and a lot of the paper processes that happen in home buying. To be able to do that, you don't have programmers who go in and for every single type of document that may come through, program where exactly the algorithm should be looking for the title number, and all these different things. It's actually more about the data, providing a lot of data to the algorithms, so that the algorithm can learn like a child would and learn from the patterns inside the data. There's a lot of power in being able to use data as the new code rather than to write code.
Where do you see machine learning by 2030? If it's done so much in the last five years, I imagine 10 more years would just be astronomical.
I think the story of technology is one of exponential curves, of everything getting faster and happening faster, and the technology improving faster than anyone could imagine. We're really excited about 2030, we think that AI is a technology that is going to change our daily lives just as much as the internet did, or even potentially more than the internet did. And just like how, in the early days of the internet, you couldn't really tell how it was going to change your life, you just knew it was going to change your life.
Now, obviously, we look back, and we have Facebook and Instagram and TikTok, and Uber and Instacart. All these incredible products are really enabled by the internet. We're going to see something similar with AI 10 years from now. In 10 years, we may not know exactly, but I think we have some hints on the things that are going to happen.
I think that healthcare is going to become dramatically safer and more accurate, and more democratized because we're going to use AI to revolutionize a lot of the old school processes. Agriculture is going to be revolutionized as well — there's a lot that we can do to increase crop yields and decrease the amount of waste in the overall food and agriculture system through use of AI and machine learning. I think that we're really excited about how transportation can be revolutionized with technologies like self driving cars, and truck delivery robots. So there's a lot of transformational changes that will happen because of AI that we're excited to enable over the course of the next decade. Looking back, we won't have known all the ways in which it will happen, but I'll be really excited. I think it'll impact retail, it’ll change how we buy things. We'll build AR, VR, and more robotics. There's just so much stuff that's going to happen in our lifetime.
Do you have an area that you're more interested in, that you’re monitoring?
We actually do work in all these industries that I mentioned. We work with financial services providers, like Brex or PayPal, or automotive players like Toyota, or players in AR and VR, or players in robotics, so we work with a huge variety of different organizations at Scale, which is super exciting. One of the reasons why a lot of us are excited about our mission is that we get to play a role in enabling AI across the board. In terms of our favorites, I mean, health care is something I'm really excited about. Because it's been an industry that has not faced a lot of transformational improvements from the internet yet, but I think AI can potentially make such a big impact on the healthcare industry.
So how has the pandemic affected Scale?
Like every company, we've tried to figure out what's the best way for us to adapt to working from home. How do we take care of our employees and make sure that everyone in the Scale community is taken care of? And then how do we then extend that and aid our customers and make sure that we're still supporting them in the exact same way that we did before? It's been a lot of reconfiguring, in turn figuring out exactly how we can work in this fully remote world, just like everyone has. But we've come out of that with a lot of learnings, and have learned to operate a lot more efficiently than before.
As a founder, what have your biggest challenges been, whether that's pre-pandemic, or now?
The biggest challenges for myself, personally, I've just been figuring out how we — it's really boring stuff — but how we can continue growing our company and making sure that we build a great company for everybody who's a part of it, and making sure that we're working for everyone who’s in our ecosystem. How do we continue serving our customers? How do we scale everything up, to be able to serve our employees, serve our customers, serve the members of our community? And I think that that process is scaling. You know, there's just so many, I'm sure you talk to other founders, and they probably tell you the same thing, each step along the way, there's a new challenge. There's different things that you don't expect, that aren't problems early on that become problems later on, or problems that you thought you solved early on, that become problems later. So that's been that's been a big part of it. We're in the world of AI and AI has just moved so fast. Making sure that we're always delivering products and delivering value to our customers in cutting edge ways to keep up with how the industry has developed, that's been a really big focus of mine as well.
Do you feel like it's more challenging to be a founder now than it was in those early days?
Definitely different. I would say one thing that we're very blessed to have is customers who have really stuck with us and grown with us and have helped us. And so today, I think we just have a much broader platform to be able to do all the incredible things that we want to be enabling for AI practitioners and our partners. Having a greater platform is exciting for us, because it allows us to have more impact, and be able to actually make a bigger dent in the world of AI and how it gets deployed. It's hard to rate things exactly. I would always trade the ability to have more impact and be able to enable this transformative technology over other things. The journey has been wonderful.
What advice would you give to other founders? Any lessons you've learned?
There's a number of things. The first is really for if you're a founder, and you're early on, and you're trying to figure out what ideas are exciting: looking for bottlenecks and solving them for the whole industry. I think that's a tried and true method. In any new industry, or any even old school problems, there's always going to be bottlenecks that can be solved with technology or a new approach. Identifying those bottlenecks is a big piece of advice I would give.
Always surround yourself with the best people you know. I think this is super, super simple. And we've all heard it a thousand times, but it's so true, like 400 people, not only people who are smart, who are ambitious, but also those who reinforce your best qualities and call you out on your worst. When building a company, it's important that you're always doing things that are authentic to your own values and authentic to how you want to be building a great company. And I think there's a lot of times when you'll talk to investors or people and they'll contribute advice that maybe goes against your instincts, but I think it's important to always be authentic. And focus, because you really can't be everything to everyone, you need to figure out what you focus on. What can you do better than anyone else? And what's the special thing that you can offer?
I think that's another big thing. This is an early stage to late stage finding, that in the early stages, you are in it, and you're in the thick of everything that's happening. And you need to find ways over time to build a power team around you to solve those problems. And I think that's one of these incredible journeys many companies go through.
I do think that businesses in every industry need to be thinking about their AI strategy today, or the economy risks being left behind. It's important for every business to be thinking about AI and how that impacts their business. And that's an exciting personal journey for us as well at Scale.
About the Data:
Thinknum tracks companies using the information they post online, jobs, social and web traffic, product sales, and app ratings, and creates data sets that measure factors like hiring, revenue, and foot traffic. Data sets may not be fully comprehensive (they only account for what is available on the web), but they can be used to gauge performance factors like staffing and sales.