Big tech companies have bankrupted three promising artificial intelligence startups in the past six months and are rethinking new M&A strategies that threaten to relegate venture capitalists to the sidelines of the AI boom.
Chatbot makers Inflection and Character. AI, as well as AI agent developer Adept, had raised more than $2 billion combined before their top talent was hired by Microsoft, Google, and Amazon, respectively.
Three deals later, big tech companies were snapping up startup founders, researchers, engineers, and product licenses, but venture capital was ultimately back to roughly where it started.
Their early exit is an ominous sign for other AI startups trying to build their own large-scale language models, the systems that underpin OpenAI's ChatGPT and Google's Gemini.
These deals will further fuel venture capitalists' concerns that the winners from the AI boom will be the biggest tech companies that can afford the billion-dollar costs of developing cutting-edge AI systems.
Earlier this month, Google hired Character.AI co-founders Noam Shazier and Daniel de Freitas and agreed to license the startup's celebrity-impersonating chatbots. Character's founders and other shareholders (whose largest shareholder is Silicon Valley venture capital firm Andreessen Horowitz) are reportedly set to receive $2.5 billion as part of the deal.
This values Character at 2.5 times its March 2023 market cap of $1 billion, a respectable but hardly spectacular return for investors who have pumped nearly $200 million into the startup since its 2022 founding.
Microsoft's partnership with Inflexion and Amazon's deal with Adept were even less lucrative for the startups' venture capital backers, who only managed to recoup slightly more than their original investments, people familiar with the deals said.
Mike Volpi, a partner at venture firm Index Ventures, said while such deals have delivered “pretty decent” compensation for founders who went on to work for big tech companies, “it's not a good outcome” for venture capitalists.
Because the vast majority of startups fail, venture capitalists hope that a small number of them will be successful enough to cover their failed investments many times over.
“VCs, especially larger ones, want extraordinary results, and 2.5 times the capital doesn't really work for a company,” said Volpi, who has invested in a number of AI companies, including Cohere and Mistral.
Less than two years after the release of OpenAI's powerful chatbot ChatGPT sparked a surge in AI investments, many founders who left corporate jobs to start startups are returning to big tech companies.
Shazier and de Freitas criticized Google's slow pace when they left to start Character in 2022, but eventually returned. Several of Adept's leaders and Inflexion's founder Mustafa Suleiman were both Google researchers before starting their companies and now work at Amazon and Microsoft, respectively.
The acquisition of the three companies by tech giants highlights how difficult it is for AI startups to scale: the resources required to train and run cutting-edge AI models are enormous, and startups without established distribution channels struggle.
These challenges are likely to become even more complicated in the next phase of AI development, according to David Kahn, a partner at venture firm Sequoia Capital.
Over the past year, startups have tried to gain an edge with novel research methods and better training data. As AI companies race to build giant data centers, costing billions of dollars apiece, to build more powerful models, Khan wrote in a recent blog post that “the next phase of the AI race will be different — it will be defined more by physical construction than by scientific discovery.”
Recommendation
This is likely to favor large technology companies, which have increased their annual capital spending from $138 billion to $229 billion in the past year, he added.
Even startups that remain independent rely heavily on partnerships with big tech companies: Microsoft has poured $13 billion into OpenAI, while Amazon and Google have invested a combined $6 billion in Anthropic, and smaller companies like Cohere and Mistral have also partnered with big tech companies.
Raviraj Jain, a partner at Lightspeed Venture Partners, said big tech companies' access to computing resources and customers gives startups building large-scale language models an “inherent advantage.” While only a handful of companies can compete at that level, he added, there's still room for venture-backed startups to build smaller AI models, applications and infrastructure.
Regulatory intervention could shift the balance again. Antitrust authorities in the U.S. and Europe are investigating deals by Amazon, Google and Microsoft, despite efforts to structure the deals in a way that keeps the startups nominally independent. Last month, the U.K. Competition and Markets Authority said it would investigate the Microsoft-Inflexion deal.
Venture capitalists say consolidation and bankruptcies are common in the early stages of a tech boom: The collapse of the dot-com bubble in the late 1990s didn't halt the spread of the internet, and the most popular consumer applications didn't emerge until years after the smartphone.
“It's definitely a difficult time for venture capital because all the excitement is concentrated in one area,” Volpi said, “but there are hundreds of companies working on a lot of interesting projects in the AI world. There are bound to be some needles in that big haystack.”
