Scale AI, which labels data for ML algorithms and was valued at $7.3B in April 2021, lays off 20% of its staff; the startup had ~450 employees in February 2022
Today I have to announce the hardest change I've ever had to make at Scale. The Economic Times : Software firm Scale AI lays off 20% of workforce
Context & Ripple Effects
Scale AI's January 2023 cut lands less than two years after the company doubled its valuation in five months with a $325M raise at $7.3B — the classic shape of a pandemic-era AI vendor caught by the funding winter. At roughly 450 employees in February 2022, a 20% reduction means about 90 people out at a startup still selling human-labeled training data.
What makes this early retrenchment notable in hindsight is the whiplash that followed: Scale nearly quadrupled sales year-over-year to almost $400M in H1 2024 and raised $1B at $13.8B, then saw OpenAI begin phasing out its Scale work after Meta's $14.3B investment — followed by a second layoff round of ~14% in mid-2025. The 2023 cut was the first signal that demand for third-party data labeling would be cyclical, not linear.
First-order effects
- Roughly 90 of Scale AI's ~450 employees lose their jobs immediately, and CEO Alexandr Wang signals the company is re-basing its cost structure for a slower fundraising market rather than growth-at-all-costs hiring.
Second-order effects
- Rivals in commercial data labeling inherit both displaced talent and price pressure, while Scale leans harder on differentiated revenue — the Department of Defense relationships that later include a $500M contract and reported Golden Dome consortium work — to offset volatile lab customers.
Third-order effects
- If the pattern holds, data-labeling vendors structurally depend on a handful of frontier-lab buyers whose allegiances shift with capital moves — Meta's investment coinciding with OpenAI's exit shows how patronage, not product quality, can decide who keeps the contract.
The trend: AI infrastructure vendors are learning that frontier-lab demand is concentrated and fickle, forcing repeated workforce resets even at companies riding headline valuation growth.