DeepMind CEO Demis Hassabis at a 2017 event in China.
Source: Alphabet
Alphabet’s AI lab, DeepMind, cut employee costs by 39% last year, according to a recent filing with a U.K. government agency.
For the 2022 financial year, staff costs and other related expenses were 594.5 million pounds (nearly $731 million), down from 969.4 million pounds (nearly $1.2 billion) in 2021 — translating to an almost 39% reduction in employee costs, per the filing. Following those cuts, in January, DeepMind announced it would shutter its first-ever international artificial intelligence research office in Edmonton, Canada.
DeepMind’s staff budget cuts occurred during the tech industry’s self-proclaimed “year of efficiency,” and Alphabet itself helped lead the charge. In September 2022, CEO Sundar Pichai announced plans to “make the company 20% more productive,” and over the past year Alphabet has conducted mass layoffs, slowed hiring, cut travel and entertainment budgets, paused construction on at least one office campus and reduced investment for experimental projects such as its Area 120 tech incubator.
Following DeepMind’s employee cost cuts in 2022, Alphabet executives discussed plans to allocate resources to key revenue drivers, such as AI, on its first-quarter earnings call of 2023. But part of that decision was bringing AI-focused groups Google Brain and DeepMind under one umbrella with “pooled computational resources.”
“Beginning in the second quarter of 2023, the costs associated with teams and activities transferred from Google Research will move from Google Services to Google DeepMind within Alphabet’s unallocated corporate costs,” Pichai said during a spring earnings call.
DeepMind’s 2022 profit was about 60.9 million pounds (nearly $74.9 million), down from 102.4 million pounds (nearly $126 million) in 2021 — a decrease of more than 40%.
“We continue to invest in fundamental research and our world-class, interdisciplinary team because we see the promise of AI to unlock scientific discoveries,” a DeepMind spokesperson told CNBC.