{"id":2233,"date":"2021-01-12T09:54:15","date_gmt":"2021-01-12T09:54:15","guid":{"rendered":"https:\/\/thenextweb.com\/?p=1333755"},"modified":"2021-01-12T09:54:15","modified_gmt":"2021-01-12T09:54:15","slug":"deepminds-mounting-losses-show-why-its-hard-to-run-an-ai-research-lab","status":"publish","type":"post","link":"https:\/\/www.londonchiropracter.com\/?p=2233","title":{"rendered":"DeepMind\u2019s mounting losses show why it\u2019s hard to run an AI research lab"},"content":{"rendered":"\n

Last week, on the heels of DeepMind\u2019s breakthrough in using artificial intelligence to predict protein folding came the news that the UK-based AI company is still costing its parent company Alphabet Inc hundreds of millions of dollars in losses each year.<\/p>\n

A tech company losing money is nothing new. The tech industry is replete with examples of companies who burned investor money long before becoming profitable. But DeepMind is not a normal company seeking to grab a share of a specific market. It is an AI research lab that has had to repurpose itself into a semi-commercial outfit to ensure its survival.<\/p>\n

And while its owner, which is also the parent company of Google, is currently happy with footing the bill for DeepMind\u2019s expensive AI research, it is not guaranteed that it will continue to do so forever.<\/p>\n

DeepMind\u2019s profits and losses<\/h2>\n
\n

\"DeepMind<\/a>
<\/i><\/a>DeepMind\u2019s AlphaFold project used artificial intelligence to help advance the complicated challenge of protein folding.<\/figcaption><\/figure>\n<\/p>\n<\/figure>\n

According to its <\/span>annual report<\/a> <\/span>filed with the UK\u2019s Companies House register, DeepMind has more than doubled its revenue, raking in \u00a3266 million in 2019, up from \u00a3103 million in 2018. But the company\u2019s expenses continue to grow as well, increasing from \u00a3568 million in 2018 to \u00a3717 in 2019. The overall losses of the company grew from \u00a3470 million in 2018 to \u00a3477 million in 2019.<\/p>\n

At first glance, this isn\u2019t bad news. <\/span>Compared to the previous years<\/a>, DeepMind\u2019s revenue growth is accelerating while its losses are plateauing.<\/p>\n

\n

\"deepmind<\/a>
<\/i><\/a>DeepMind\u2019s revenue and losses from 2016 to 2019<\/figcaption><\/figure>\n<\/p>\n<\/figure>\n

But the report contains a few more significant facts. The document mentions \u201cTurnover research and development remuneration from other group undertakings.\u201d This means DeepMind\u2019s main customer is its owner. Alphabet is paying DeepMind to apply its AI research and talent to Google\u2019s services and infrastructure. In the past, Google has used DeepMind\u2019s services for tasks such as managing the power grid of its data centers and improving the AI of its voice assistant.<\/p>\n

[Read: Meet the 4 scale-ups using data to save the planet<\/span><\/a>]<\/em><\/p>\n

What this also means that there isn\u2019t yet a market for DeepMind\u2019s AI, and if there is, it will only be available through Google.<\/p>\n

The document also mentions that the growth of costs \u201cmainly relates to a rise in technical infrastructure, staff costs, and other related charges.\u201d<\/p>\n

This is an important point. DeepMind\u2019s \u201ctechnical infrastructure\u201d runs mainly on Google\u2019s huge cloud services and its special AI processors, the Tensor Processing Unit (TPU). DeepMind\u2019s main area of research is <\/span>deep reinforcement learning<\/a>, which requires access to very expensive compute resources. Some of the company\u2019s projects in 2019 included work on an AI system that played <\/span>StarCraft 2<\/a> <\/span>and another that played <\/span>Quake 3<\/a>, both of which cost millions of dollars in training.<\/p>\n

A spokesperson for DeepMind told the media that the costs mentioned in the document also included work on the <\/span>AlphaFold<\/a>, the company\u2019s celebrated protein-folding AI, another very expensive project.<\/p>\n

\n