Scale AI's push for more educated workers

Good morning,

Today, we’ll discuss a $14B AI startup’s push for more educated workers and what this reveals about the AI industry. Plus other tech news you need to know from this week. Let’s go 👇

This week’s insight

Context

Scale AI, a hot data labeling startup for AI, recently raised a massive $1B Series F round led by Accel (an existing investor) that values it now at nearly $14B.

It helps organizations develop AI applications by turning their raw data into high-quality training data. It does this by employing cheap overseas labor. However, with more money being poured into AI these days, Scale is now hiring more educated workers in the US. Scale also provides tools and infrastructure solutions for AI/ML teams.

Pre-ChatGPT

The shift toward LLMs has helped it jump-start business growth after it started to slow down. Before the launch of ChatGPT, Scale’s revenue growth dipped below 50% in 2022, significantly behind previous revenue forecasts it had shared with investors.

At this time, it was hiring cheap labor overseas to train models (it still is)—workers would label autonomous vehicle sensor data like images of pedestrians, stoplights, and traffic signs for companies like Waymo, and Cruise, or tag shopping images for Instagram.

Scale was losing money, spending more than $1.40 in operating costs for every $1 of revenue 2 years ago. Things started improving in 2023, in which the company generated $334M.

Scale’s shift to more educated workers

Fast forward to today, every company is investing in AI, which is great for Scale’s business. The startup is run by 27-year-old CEO Alexandr Wang, who has told investors his company expects revenue to triple to just over $1B this year.

But this fast-paced growth comes at a cost. Scale now has shifted to a more highly educated workforce based in the US—including PhDs, doctors, and lawyers, which sharply increased labor costs. While its workers in Africa, India, and the Philippines earned an average of $1.50 per hour, workers in the US earn about $40 per hour.

Labor costs have ballooned up for the company, but it has found other ways to reduce operational costs by automating parts of the labeling process.