Alexandr Wang Jr.

Alexandr Wang grew up in New Mexico, landed full time engineering jobs in Silicon Valley as a teenager, then dropped out of MIT after his freshman year to chase machine learning full time. Instead of building yet another model, he spotted a more boring bottleneck that everyone else was ignoring: labeled data and the infrastructure to feed it into production systems at scale.

In 2016, at 19, he cofounded Scale AI as what he described as an API for human intelligence that could handle the tedious labeling work for self driving teams. That wedge looked small to early investors who thought self driving was too narrow, but Wang correctly bet that if you owned the data firehose for one frontier AI use case you could expand into every other high value domain that needed training data and evaluation.

The first proof of distribution power came from going straight at the hungriest customers in the world, self driving car companies that were drowning in sensor data and had money to spend. Within months of launch, Scale AI was serving names like Uber and Alphabet, and inbound from a Cruise Automation engineer via YC connections turned into a flagship contract that validated the product with the entire autonomy segment.

From there Wang kept climbing up market, selling the same core capability to OpenAI, Google, and eventually the Pentagon as defense programs realized they needed industrial scale labeled data and red teaming for AI systems. By 24 he was the youngest self made billionaire, and by 2024 Scale AI was reportedly doing around $870 million in revenue as customers treated it as critical infrastructure rather than a replaceable vendor.

That positioning paid off when the foundation model race accelerated and hyperscalers wanted to lock down their supply chains. In 2025 Meta bought 49 percent of Scale in a deal that valued the company around $29 billion, and then hired Wang as chief AI officer to run its Superintelligence Labs, effectively turning a former infrastructure supplier into the person setting Meta's long term AI strategy.

1. Start at the pain point every frontier team feels but no founder wants to own. Wang ignored the glamorous model layer and instead built a boring but indispensable data pipeline that solved labeling, quality control, and routing to cheap or expert labor, which made Scale AI impossible to rip out once it was wired into a customer's workflows.

2. Land with one desperate vertical, then use their logos as your outbound sales deck for everyone else. Self driving teams were the first wedge, but Wang systematically turned early wins with Uber, Alphabet, and Cruise into social proof that opened doors at other autonomy companies, then at big tech labs, and finally at government buyers that do not move without seeing that kind of reference stack.

3. Turn infrastructure into leverage over the ecosystem, then sell that leverage to the highest bidder. By owning the data and evaluation rails used by OpenAI, Google, and the Pentagon, Wang made Scale AI strategically dangerous for any foundation model competitor to ignore. Meta did not just write a big check, it effectively neutralized a risk and acquired a founder whose distribution now runs through one of the largest consumer platforms on earth.