Scale AI Inc., a three-year-old startup run by a 22-year-old, is teaching machines how to see. For that, it just joined Silicon Valley’s list of unicorns with a fresh $100 million investment that puts its valuation above the coveted $1 billion mark, and its artificial intelligence (AI) technology has already attracted big-name customers in the field for autonomous vehicles, according to Bloomberg.
Alphabet Inc.’s (GOOGL) Waymo, General Motor Co.’s (GM) Cruise, and Uber Technologies Inc. (UBER) are all buying what Scale has to offer, because well, self-driving cars are machines that need to be able to see. Scale stands out because it has built a set of software tools that are significantly reducing the time it takes to train a machine how to process and interpret visual imagery. And less time means lower costs.
“There is a really big gap between the handful of giant companies that can afford to do all this training and the many that can’t,” said the company’s co-founder and chief executive officer (CEO), Alexandr Wang. Using Scale’s technology, “tasks that used to take hours end up taking just a couple of minutes.”
What It Means for Investors
In order to accurately identify and label an object, computers need to be able to match the image of that object with a recognizable pattern. That’s the AI aspect of machine learning. But the pattern itself doesn’t just fall from the sky. The pattern is the result of a process whereby a human must first identify and label the given object. That generally means tracing the outline of the object using the cursor of a mouse, and not just for one photo, but millions, if not more.
“It takes billions or tens of billions of examples to get AI systems to human-level performance,” said Wang, who has an army of 30,000 contractors scattered across the globe aiding in the object-identification and labelling process. “The humans are pretty critical to what we’re doing because they’re there to make sure that all the data we provide is really high quality,” Wang told TechCrunch.
Scale’s software reduces the overall time this process takes by scanning the images, identifying and applying a label to the interpreted object, and then prompting a human to verify that the object was correctly labeled. Intervention is only needed if the software incorrectly identifies the object. In such cases, rather than retracing the entire object, the human worker need only click once on the object and the sophisticated software will create the outline.
Some of Scale’s investors, including Accel and Peter Thiel’s Founders Fund, say the company’s software is more advanced and is able to label data faster and cheaper than the current alternatives. Founders Fund led the latest Series C round of financing, which also included investments from Coatue Management, Index Ventures, Spark Capital, Thrive Capital, Instagram founders Kevin Systrom and Mike Krieger, as well as Quora CEO Adam d’Angelo.
“In general, AI and machine learning is just growing so quickly as a field, that it’s appropriate to raise this amount that will allow us to capitalize on our ambitions,” Wang said of the $100 million his company raised, based on a valuation fit only for unicorns. “We don’t want to be in the business of constantly needing to raise capital, so ideally this is the last fundraise for us.”
While Scale has attracted big customers in autonomous driving, a number of non-automotive companies from Airbnb to Pinterest are also interested in the company’s technology. Wang understands the wide applicability of Scale’s AI software. “We’re honing in on AI broadly,” he told TechCrunch. “Our goal is to be a pick axe in the AI goldrush.”