Call it the Nadella variation. It might sound like a diabolical move in a chess match, but it is in fact a useful term to describe the latest calculated corporate gambit by Satya Nadella, Microsoft’s chief executive.
Tech companies often try to poach teams of smart employees through so-called acquihires: acquiring a start-up to hire the people. Nadella’s novel twist is to hire the people and leave the company behind.
This week, Microsoft announced that it had recruited two of the three founders of Inflection, once one of the hottest AI start-ups in the US, as well as many of its 70 employees.
Mustafa Suleyman and Karén Simonyan will now oversee Microsoft AI, which will take responsibility for the company’s consumer-facing AI services, including its Copilot chatbot, its Bing search engine and its Edge browser.
Microsoft’s latest hiring spree, following its $13bn investment in OpenAI and its more recent partnership with France’s Mistral, highlights the company’s intent to ally with ambitious AI start-ups and command the market. It had also been an early backer of Inflection.
The company’s hyperactivity, laced with a lot of investor hype about AI, has helped Microsoft re-emerge as the world’s most valuable public company with a market value of $3.1tn, more than all the companies listed on London’s FTSE 100 index put together. Once again, Microsoft has left its arch-rival Google spitting dust.
Microsoft’s move illustrates the scramble for the best researchers among the world’s biggest tech companies as they stake their claims in the future AI economy.
Even relatively junior engineers at the leading research companies that are developing AI foundation models, including OpenAI, DeepMind and Anthropic, can command seven-figure salaries and are bombarded with job offers whenever they log on to LinkedIn. More senior engineers can earn up to $10mn.
“There is no question that there is an enormous talent war,” says Jordan Jacobs, managing partner of Toronto-based Radical Ventures, which has invested in some 50 AI start-ups globally. “At the top of the pyramid are the people who can build foundation models and make them sing. There are very few people who can do that effectively and there are organisations that will pay them an absolute fortune to do so.”
But it is also a further sign that the emerging AI economy will probably be dominated by the US tech giants with the money, human capital and cloud computing infrastructure needed to train state of the art foundation models, such as OpenAI’s GPT-4 and Google’s Gemini.
Suleyman is a valuable yet controversial hire for Microsoft. A prominent voice in the AI revolution, he was one of the three founders of DeepMind, the pioneering London-based AI research company bought by Google in 2014.
He says he was excited by the chance to join Microsoft to bring AI services to more than 1bn users in a responsible way. “My vision for how this plays out is that everybody is going to have a trusted, reliable personal AI that makes you smarter and more productive,” he told the Financial Times this week. “I think it’s going to be a cool collaboration.”
However, Suleyman has a chequered past. He left DeepMind in 2019 following an independent investigation into bullying and harassment accusations against him. Suleyman has publicly apologised for his behaviour, saying in a recorded interview that he “really screwed up” and was “very demanding and pretty relentless”.
His expertise is also in product development and policy, rather than research. But in Simonyan, Inflection’s other co-founder and chief scientist, Microsoft gains one of the world’s top AI researchers. As a principal scientist at DeepMind, he helped pioneer its powerful AlphaZero model and went on to build an impressive team of researchers at Inflection. Most of them seem likely to follow Simonyan to Microsoft.
Kevin Scott, Microsoft’s chief technology officer, says the Inflection folk who are joining Microsoft have a very good sensibility for making leading-edge AI models and adapting them to different user contexts. “You just need a lot of talent,” he says.
It is unclear where that leaves Inflection, once viewed as one of the world’s most exciting and well-resourced AI start-ups after it launched a conversational chatbot called Pi last year.
In June, Inflection raised $1.3bn from several prominent corporate and personal investors, including Microsoft, Nvidia, Bill Gates and Reid Hoffman, who was the third co-founder of Inflection but is better known as co-founder of LinkedIn, and sits on Microsoft’s board.
However, Inflection’s chatbot has failed to gain traction with users and in the absence of a viable business model its few remaining staff will now look to sell its technology to enterprise customers. In an analysis of the start-up’s traffic estimates for February, independent technology analyst Ben Thompson wrote in his influential newsletter that the “relatively sluggish usage numbers”, especially given the cost of training its models, were “catastrophic”. In his assessment, Inflection was a “failed company”.
According to Inflection, both the company and its investors will benefit from a new licensing agreement with Microsoft. Shareholders will also be made whole and get more than their invested capital back, the company says.
In a different era, the big tech companies might have bought a start-up such as Inflection outright, but they appear reluctant to launch takeover bids today, given the antitrust activism of the Biden administration.
Instead, Microsoft, Google, Nvidia and Amazon have been among the most active investors in AI start-ups, often trading access to computing power and sophisticated microchips for a financial stake in these companies.
The emergence of this interconnected network of companies and start-ups has already attracted the scrutiny of regulators. In January, the US Federal Trade Commission launched a probe into five tech companies operating in this field. “Our study will shed light on whether investments and partnerships pursued by dominant companies risk distorting innovation and undermining fair competition,” wrote the FTC’s chair Lina Khan.
Amba Kak, executive director of the AI Now Institute and a former FTC adviser, says it is heartening that regulators are already probing this concentration of corporate power. The focus will be on market structures that might stifle competition rather than more traditional antitrust concerns about consumer harm.
“It is not going to be about who has the best model but who has the best infrastructure and platform-play with access to the consumer,” Kak says. “Inflection and Mistral are the barnacles on the hull of Big Tech.”
Some venture capital investors will be disappointed if they cannot exit their start-up investments through trade sales as they have relied upon in the past. Even so, many are still happy to fund start-ups exploring the AI industry’s foothills. While the big AI companies focus on building omni-purpose foundation models, start-ups can help answer client companies’ narrower business needs.
“Writing a poem about your dog in the voice of Donald Trump requires an enormous AI model. But if you are focusing on the enterprise side, then you can train a far smaller model and make them work better for a particular business need,” says Jacobs of Radical Ventures. “We are at the very beginning of a big replacement cycle for every bit of software ever written over the next decade.”
The competition for this level of talent is intense. Jacobs’ fund has a dedicated internal team that scours the world for AI experts to join its portfolio companies. Some nation states, including Canada, Singapore and the United Arab Emirates, have also been targeting promising AI start-ups in Europe and tempting them to relocate.
Yet although the deep-pocketed five US tech companies — Amazon, Apple, Google, Meta and Microsoft — might seem in the best position to hoover up the top talent, not everyone wants to work for a giant business on the West Coast. The Nadella variation has its limits.
The research company Zeki, which tracks the top 140,000 AI scientists and engineers in 20,000 businesses in more than 90 countries, has found an increasing desire among many to work outside the US. They are also attracted to more traditional companies in healthcare, banking or manufacturing or want to join a start-up.
The UAE, Saudi Arabia, several Nordic countries and South Korea have emerged as net importers of AI scientists, with more moving to their countries to work than leaving them. Some national industrial champions, including Siemens in Germany, Samsung in South Korea and ASML in the Netherlands have also become big employers of AI engineers.
Overall, the 10,000 small companies in Zeki’s database hire more top AI talent than the big five combined. “The bigger, slower story is the shifting sands beneath the big five as the AI world deepens and diversifies away from the US,” says Tom Hurd, Zeki’s co-founder.
Even though he has now jumped ship to two big tech companies after co-founding two AI start-ups, Suleyman thinks there are still plenty of opportunities for small companies that can “go superfast, and be creative and invent things that the bigger companies never even thought possible”.
“It is hyper-competitive at both ends of the spectrum,” he says. “At the small, start-up scale everybody is on fire in terms of their creativity. And at the big end, everyone’s going head-to-head there, too. I think it’s just a ferociously competitive time to be working in tech.”
Additional reporting by Madhumita Murgia and Camilla Hodgson
Data visualisation by Alan Smith and Ian Bott