Business

DeepMind A.I. Unit Lost $649 Million Last Year and Had a $1.5 Billion Debt Waived by Alphabet

Source: Alphabet
  • The London-headquartered AI lab had a loss of £477 million ($649 million) in 2019, up from £470 million in 2018, according to documents filed with the U.K.'s Companies House registry.
  • The vast majority of DeepMind's spending in 2019 went on "staff and other related costs."
  • While losses climbed slightly, revenue grew from £103 million in 2018 to £266 million in 2019.

LONDON — Losses at DeepMind, the artificial intelligence firm owned by Google parent Alphabet, grew 1.5% last year, according to its latest annual report.

The London-headquartered AI lab — founded in 2010 by Demis Hassabis, Mustafa Suleyman and Shane Legg — had a loss of £477 million ($649 million) in 2019, worse than the £470 million loss in 2018, according to documents filed Thursday with the U.K.'s Companies House registry.

The vast majority of DeepMind's spending in 2019 went on "staff and other related costs," with the annual report showing that some £468 million went toward this, up from £398 million in 2018.

DeepMind now employs around 1,000 people worldwide, including some of the world's leading AI research scientists, who can command annual salaries of more than $1 million. These top people, who often have Ph.D.s from the likes of Oxford, Cambridge, Stanford or MIT, can command this sort of money because they're also sought after by Big Tech companies like Facebook, Apple, Amazon and Microsoft.

A DeepMind spokesperson told CNBC: "During the period covered by these accounts, DeepMind laid the foundations for our groundbreaking results in protein structure prediction — a 50-year grand challenge in biology — and collaborated with teams across Google to deliver real-world impact at scale."

"Our teams were involved in a huge range of projects, from improving the predictability of wind power to accelerating ecological research in the Serengeti. We're excited to build on this unprecedented progress as we head into next year," the spokesperson added.

While losses climbed slightly, revenue grew from £103 million in 2018 to £266 million in 2019. However, the revenue is coming from other companies within Alphabet (namely Google), which pay DeepMind for research and development.

DeepMind was acquired by Google in 2014 for around $600 million. Today, the company relies on a steady stream of capital from parent firm Alphabet.

In 2019, Google Ireland waived the repayment of intercompany loans and all accrued interest amounting to £1.1 billion, according to the filing.

DeepMind has been based out of a large Google building in London's King's Cross neighborhood, but it plans to move into a new office early next year. According to the filing, DeepMind spent £1.3 million on construction, furniture and fixtures, and network and production equipment in 2019.

The filing also shows that DeepMind donated £6.3 million to academia in 2019, down from £13.5 million in 2018.

In terms of risks and uncertainties, the filing states: "Machine learning research and application is an emerging market characterized by continuous change and intense competition. As a result, the company will continue to face risks and uncertainties, which may have a significant impact on its ability to achieve continued success within its market."

Despite the losses, Alphabet has pledged to keep funding DeepMind.

"I'm very happy with the pace at which our R&D on AI is progressing," Alphabet CEO Sundar Pichai said on the company's second-quarter earnings call. "And for me, it's important that we are state-of-the-art as a company, and we are leading. And to me, I'm excited at the pace at which our engineering and R&D teams are working both across Google and DeepMind."

Azeem Azhar, founder of the Exponential View newsletter and podcast, told CNBC: "The question for me is why aren't more companies in other sectors backing core research in exponential technologies like AI, synthetic biology, battery technology, quantum computing and the like. That backing can be direct operational dollars like Google or it could be in corporate venturing or university partnerships. But outside of a few companies (like VW or Bosch) I don't see it happening."

He added: "You won't see huge societal returns in five years, more like 20 or 30 or more. The Intel 4004 launched in 1970. It was 2001 before half of U.K. households had a computer."

Copyright CNBC
Contact Us