DeepMind, Google’s artificial intelligence unit, reported pre-tax losses of £460.9m in 2019.
The London-based company published its filings with the UK Companies House registry.
It reported dramatic losses for the second year in a row, as well as £1.1bn of debts, waived by Google.
Google bought DeepMind in 2014, for around £400m, to compete against other tech companies turning to machine learning to enhance their businesses.
DeepMind recently struck a major breakthrough by demystifying ‘protein folding,’ potentially making treatments of all manner of ailments and diseases far easier in the future.
Projects for Google include reducing electricity used at Google’s data centre and creating a more natural voice for the Google Assistant.
The filings report that the companies administration expenses cost £717 million in 2019, compared to its turnover of £265.5m. The costs are due to “a rise in technical infrastructure, staff costs and other related charges”.
DeepMind employs the best-minds in artificial intelligence. Last August, eFinancialCareers estimates that the AI lab pay around £478k per head, including pensions and travel.
The filings also refer to Google’s controversial take-over of DeepMind’s health subsidiary last year.
It said: “The Company distributed intellectual property assets which had a nil book value to another group undertaking on 31 October 2019”.
DeepMind handed over contracts with the NHS for services such as their clinical task management app Streams, as well as predictive health AI research which uses patient’s data to develop predictive diagnostic models.
In total, DeepMind Health had access to 1.6m patient records. In 2017, a review by the National Data Guardian came to the conclusion that this data had not been transferred “on an appropriate legal basis”.
Dame Fiona Caldicott of the NDG, said that if the data is used in unexpected ways “we risk damaging public trust and losing support for the technological advances that could benefit us all”.