How BenevolentAI bounced back from a $1 billion drop in valuation

First published by WIRED on 5 May 2020

When London tech startup BenevolentAI secured a $90 million (£68 million) cash injection from Singaporean sovereign fund Temasek in September 2019, the deal valued the business at $1 billion. Not bad for a six-year-old company whose ambition to dramatically change the way drugs are discovered and developed remains largely unrealised.

In reality, the $1 billion valuation tag was less of an achievement and more of a come-down for BenevolentAI, which in April 2018 had been valued twice as high, at $2 billion, after raising $115 million from a number of undisclosed American investors.

The problem was that one of BenevolentAI’s early backers was one-time star investor Neil Woodford, who was forced to wind up his Woodford Equity Income fund following a prolonged period of underperformance. Having been the largest unquoted company in the portfolio and the largest holding in Woodford’s Patient Capital Trust, BenevolentAI’s value was impacted by the fire sale.

Not that BenevolentAI chief executive Joanna Shields is bitter. Her attitude is that, in the world of startups, these things inevitably happen. “It’s frustrating, but I don’t want to blame the situation [for the fall in valuation],” she says. “It is what it is.”

Shields feels that how much the Woodford situation actually contributed to BenevolentAI’s revaluation – and indeed how much Woodford’s backing may have inflated its value in the first place – is moot, given that the business has been going through a process of reinvention since she was appointed to its board in May last year.

“When I arrived, we’d just closed our funding round at $2 billion and we were much more of a traditional drug-discovery company at that stage,” she says. “We looked at the tech and decided to shift strategy. That pivot [affected the company’s valuation].”

The business, which in the year to December 2018 made a pre-tax loss of £33 million on turnover of £6.8 million, also expects that change to enable it to start making some significant returns. Having bought in pharmaceutical capabilities at the beginning of 2018, the firm no longer uses machine learning just to identify new ways of treating diseases, but to find ways to develop its own drugs too. Though it has formed partnerships with pharma giants AstraZeneca and Novartis, creating medicines in its own right is what will drive revenue.

In late January, researchers turned their attention to coronavirus and used technology that can scour literature related to the virus to pinpoint a possible treatment for the virus. It will soon be tested in a clinical trial with the National Institutes of Health in the US, according to a New York Times report.

For US-born Shields, who was made a life peer in the House of Lords in 2014, all this is a far cry from her years running early-stage operations for Google, Bebo and Facebook, and she admits that when she joined BenevolentAI “people were speaking a language I didn’t understand”. But she says the lessons she has learned in bringing numerous businesses from the startup to the IPO stage will remain unchanged regardless of the sector she ends up operating in.

“The most valuable lesson is that building an organisation is based on human beings,” she says. “We have a tendency to simplify things and put them into charts, but what I’ve learned is that organisations are much more organic and fluid; you have to build them around the people in them.”

That means listening to people and giving them the tools – tech driven or otherwise – they need to succeed.

“People in tech tend to be very analytical and driven by structure, but [growing a business] is an art, not a science,” Shields adds.


Most people think tech businesses are all about data and machines, but Shields says nothing is more important than people, especially when a firm is in startup mode. Hiring the right blend of personnel and giving them the freedom to make an impact has been the secret to her past successes.

Since arriving at BenevolentAI, Shields has steered the business on a different course, expanding its scope from researching drugs into also developing them. Although the company’s valuation has suffered, she is certain it was “the right pivot to make”.

When she moved to London in 2000 to grow RealNetworks’ European arm, Shields had just taken on another challenging full-time role as a lone parent to her then one-year-old son. Having childcare she could rely on completely allowed her to focus fully on the day job.