He was a child prodigy who landed a high-paying engineering job in Silicon Valley at the age of 17. He entered MIT on a scholarship to study artificial intelligence and machine learning, but dropped out before graduating. He started his own company and, before he turned 24, became the youngest billionaire on the planet.
Ladies and gentlemen, the multi-talented Alexander Wang, founder of Scale AI.
When his San Francisco-based startup was valued at $7.3 billion in 2021, its co-founder and CEO Alexander Wang was named “the world’s youngest self-made billionaire.” It was a short tenure, Forbes said, with Wang “quickly falling out of the billionaires club as valuations of private tech companies plummeted the following year.”
However, in 2024, the now 27-year-old “reclaimed the title” after announcing that Scale AI had secured a significant capital injection, increasing his own fortune (he owns about 14% of the company) to $2 billion.
So who is he and what makes his company so special?
“Privileged education”
Wang’s “privileged upbringing” was crucial in giving him “a solid foundation in science and technology,” as Mashable reports.
He was born in New Mexico to Chinese immigrant parents who worked as project physicists for the US Air Force and the military. It’s clear that he inherited a brilliant mind. He was also a “math genius” as a child and “loved participating in coding competitions,” ZeeBiz writes.
He said he was taught advanced physics when he was still in kindergarten, so it’s no surprise that he started making plans for a company while he was still a junior in high school. As a teenager, Wang joined debate teams, played the violin and traveled the country to math and physics competitions. He dropped out of high school a year early to work in Silicon Valley, where he became an engineer at the question-and-answer site Quora.
The first contact with the wonderful world of AI
He began to see the groundbreaking possibilities of artificial intelligence after attending a summer camp in San Francisco called SPARC, the Summer Program for Applied Rationality and Cognition. It was designed to bring together students with talent in math and science. There, he met early AI researchers, including OpenAI co-founder Greg Brockman and Dario Amodei, CEO of Anthropic.
In 2015, Wang went to MIT. As a freshman, he took five undergraduate computer science courses and spent his spring break working on an iPhone app called Ava. He dropped out.
Y and Altman combinator
Wang made the decision to start his own company after a conversation with Eric Wu, CEO of Opendoor. Wu tried to convince Wang to take a job at the then-startup company, encouraging him to take risks while he was young. “I knew I would regret it if I never took the risk of becoming an entrepreneur at the perfect time,” he said.
In the summer of 2016, Wang enrolled Ava in a startup support program run by the famed venture capital firm Y Combinator, then led by Sam Altman, the founder and CEO of OpenAI. Soon after, he and Lucy Gao expanded the concept to Scale AI. Altman ended up indirectly owning a stake in Scale through his deal with Y Combinator, which holds shares in the startups he nurtures.
What Scale AI does
According to the Wall Street Journal, Wang has built a vast army of more than 100,000 contractors who take on the “dirty work” that fuels the modern AI boom.
Sitting behind computers in cities around the world, Scale AI employees type stories, tag images and construct sentences that provide chatbots with the text they need to better understand human speech patterns. Their tasks range from composing haikus and summarizing news articles to writing stories in languages like Xhosa and Urdu.
It’s a labor-intensive business that has become so coveted by giants looking to get into the AI race that Scale’s revenue tripled last year, boosting its valuation to $14 billion. The 27-year-old founder compares his company’s importance in the AI revolution to Nvidia’s processors.
The first contracts
Scale AI launched amid a boom in funding for fully automated cars. As the WSJ explains, developers of the technology needed labeled images to help autonomous vehicles recognize objects like stop signs or pedestrians, and they turned to the startup for help.
Within months of its founding, Scale had signed contracts with Cruise and Tesla and began building its network of contractors. In 2017, Wang created Retasks, a subsidiary focused on hiring cheap labor overseas. At one point, Scale set up facilities in Africa and Asia to train data labelers.
Soon he was hiring hundreds of contractors through online chat groups. Many came from the Philippines. They were young people who worked in Internet cafes, playing video games while completing tasks on Retasks.
As the boom in fully automated cars slowed, Wang sniffed out other revenue streams. In 2019, he signed Scale’s first genetic AI contract, with OpenAI, to label data for an early version of the language model behind ChatGPT. The following year, he signed a deal with the U.S. military to help it build datasets for its efforts in AI applications.
The “soldiers” do not know the “general”
At Scale’s San Francisco offices, employees are recruiting, posting ads on sites like Reddit and LinkedIn touting the benefits of flexible remote work.
The startup’s contractors often don’t know they’re working for Scale. They complete tasks through two websites, called Retasks and Outlier, neither of which disclose their affiliation with the startup.
Projects are given codenames so that employees don’t know which clients they’re writing for. The scale typically uses animals, like the ostrich for OpenAI and the bee for Apple, though recently it’s gotten more creative. Google’s codename, for example, is Bulba, a reference to Pokémon.
Managing this empire is difficult. Some workers have quit, saying they are frustrated with late payments and wages that can be as low as $8 an hour. Others have found ways to cheat on the job to increase their productivity and earn more money. Sometimes the data they produce is of such poor quality that Scale’s own employees — even top executives and engineers — have been forced to redo it themselves.
When contractors turn to… ChatGPT
Meta’s codename is Flamingo. When Scale AI shut down a project for the tech giant last year, Wang declared a state of emergency.
Early last year, Meta Platforms asked the startup to generate 27,000 question-answer pairs to help train AI chatbots on Instagram and Facebook.
When Meta researchers received the data, they noticed something strange. Many of the responses looked the same or began with the phrase “as a multilingual AI model…” The contractors had used ChatGPT to write their responses, in complete violation of Scale’s purpose.
The researchers reported the disappointing results to Scale. The data had to be submitted from the beginning or the contract would be lost. Wang asked employees to drop everything and create new writing samples to submit to Meta. An internal leaderboard showed who had completed the most tasks. The winner’s prize: a vacation.
Scale later discovered that much of the bad data sent to Meta came from Kenyans who had become experts at making quick money on the Retasks platform. Now he has learned to spot the “lazy” ones in time.
