DeepMind, the Google-owned artificial intelligence firm on a mission to create human-level AI, had an expensive year in 2018, according to documents filed with the U.K.’s Companies House registry on Wednesday.
The London-based AI lab—founded in 2010 by Demis Hassabis, Mustafa Suleyman and Shane Legg—saw its pretax losses grow to $570 million (£470 million), up from $341 million (£281 million) in 2017, and $154 million (£127 million) in 2016.
DeepMind’s losses are growing because it continues to hire hundreds of expensive researchers and data scientists but isn’t generating any significant revenue. Amazon, Apple, Facebook are locked in an expensive battle with DeepMind and Alphabet to hire the world’s best AI experts, with the goal of building self-learning algorithms that can transform industries.
In 2018, DeepMind spent $483 million (£398 million) on around 700 employees, up from $243 million (£200 million) in 2017. Other significant costs included technical infrastructure and operating costs. In addition, DeepMind spent $17 million (£14 million) on academic donations and sponsorships.
DeepMind also spent $12 million (£9 million) on construction and $1.2 million (£1 million) on furniture and fixtures. The company is planning to move out of Google’s office in King’s Cross and into a new property around the middle of 2020.
While losses at DeepMind have grown, so to have the company’s revenues. Turnover almost doubled in 2018 to £103 million, up from £48 million in 2017. The firm sold some of its software to Google, which has used DeepMind’s AI systems to make the cooling units in its data centers more energy efficient, and improved battery life on Android devices. DeepMind does not make any money from its work with Britain’s National Health Service.
A DeepMind spokesperson provided Forbes with the following statement:
“We’re on a long-term mission to advance AI research and use it for positive benefit. We believe there’s huge potential for AI to advance scientific discovery and we’re really proud of the impact our work is already having in areas such as protein folding.
“Our DeepMind for Google team continues to make great strides bringing our expertise and knowledge to real-world challenges at Google scale, nearly doubling revenues in the past year. We will continue to invest in fundamental research and our world-class, interdisciplinary team, and look forward to the breakthroughs that lie ahead.”
In 2018, DeepMind also passed its Streams application for clinicians to Google. However, this transaction had not been completed by the time the financial statements were filed.
Yann LeCun, chief AI scientist at Facebook, said in an interview last year that he does not think DeepMind has yet proved its worth to Google, adding that DeepMind is too isolated to have a significant impact on the tech giant. “I wouldn’t want to be in the situation Demis [the CEO] is in,” he said.