- VCs think more M&A is on the horizon for AI startups.
- Big Tech's demand for top AI talent could prompt a wave of "acquihires" later this year.
- Investors said overpriced AI startups could see their best talent poached.
In March, Microsoft signaled its intention to push further into consumer AI by notching an unusual deal with Inflection AI, the chatbot startup the company invested in, paying $650 million to license its AI software after poaching Inflection's cofounders and much of its staff.
Microsoft stopped short of officially acquiring the $4 billion startup. But the tech giant's play underscores the market's demand for tech talent — and venture capitalists expect a wave of "acquihires" to hit the AI ecosystem later this year.
As tech companies scramble to add AI into their offerings, they're contending with a limited pool of top engineers with the skills to train AI models or build new AI products from scratch. To draw in top talent, these companies will be interested in "acquihiring founders or teams with exceptional backgrounds," said Sri Ayangar, an investor at OpenOcean.
M&A dealmaking among AI startups is already off to a strong start this year, with 55 exits recorded in Q1 of 2024, per Crunchbase data. Veradigm's $140 million acquisition of AI-enabled biomedical startup ScienceIO was the biggest AI deal in this period. Investors told Business Insider they expect to see more deals in the coming months as the talent market becomes even more competitive and AI startups return to the market looking for more cash.
The majority of AI startups are overpriced, said Konstantin Vinogradov, general partner at Runa Capital. Even if they have impressive technology, most of these new upstarts will fail, he added — ushering in more cases of distressed M&A.
The AI startups that raised these overpriced rounds in 2023 and 2024 are likely to go to market again by the start of next year, with M&A activity picking up in early 2025, said Somesh Dash, general partner at IVP.
Buyers fall into three key categories
Investors that Business Insider spoke to highlighted three main categories of buyers in the market.
Big Tech incumbents and AI juggernauts have been the most notable buyers and have fueled customer demand for AI products by nabbing smaller startups. Apple is leading the acquisition race, quietly grabbing an estimated 32 AI startups in 2023, overshadowing Meta's 18 AI acquisitions and Microsoft's 17, according to data from Statista.
AI enterprise giants such as Databricks and Snowflake are the current frontrunners for buying companies building in the application layer, Shmuel Chafets, founding partner at Target Global, told BI.
Databricks has been on a spending spree in the past year, and its previous acquisitions of MosaicML and DBRX "show how it's looking to acquire companies building a strong AI infrastructure," said Sivesh Sukumar, VC at Balderton.
Other startups training large AI models like Anthropic, Ideogram, and Pika, are always looking for talent specializing in building and optimizing those models, and could make acquisitions to get it, said AIX Ventures founding partner Shaun Johnson.
Non-tech companies operating in traditional sectors, such as banking, finance, and consulting, are also looking to expand their AI portfolios and offer AI-enabled services to clients. Sukumar pointed to Accenture and Boston Consulting Group as potential buyers in this category.
Johnson highlighted brands like Intuit, Nike, and American Express — big companies that "are never going to train their own model, but they're going to try to leverage AI," he said, suggesting these companies' brands are "sexy" enough to attract engineering talent.
In a third category, however, are the less publicly innovative brands that want to leverage AI, like Safeway, Johnson said. He said these companies may find it difficult to recruit or even acquihire top AI talent, and could forced to buy AI startups to buy the technology itself, rather than building AI applications internally.
Acquisition targets
While companies touting technical talent have been prime targets, startups that have built a strong brand — and "moats around that brand" — are also attractive to investors, Sukumar said. He cited the likes of voice cloning startup ElevenLabs, AI search engine Perplexity, and AI image editor PhotoRoom as having strong buyer interest. All have valuations ranging from $500 million to over $2 billion.
"Perplexity is a top target given they're threatening Google search," Sukumar told BI. "I used the words "I'll Perplexity it" rather than "I'll Google it" for the first time without realizing."
VCs are also eyeing the foundational model layer as an acquisition target. The amount of computing power required to run and develop these models means there will likely be consolidation among the big "incumbents," said Mike Smeed, managing director at InMotion Ventures.
In Europe, LLM developer Mistral could be a key contender for acquisition, with Microsoft already showing interest in the year-old upstart by investing over $15 million, Chafets told BI. Ayangar also pointed to alumni of DeepMind or OpenAI as prime targets for talent acquisition.
At the application level, Dash said startups in industries with less regulation — think consumer tech and gaming rather than healthcare or biotech — will be acquisition targets. He pointed to AI avatar generation startups, which he said could be "quietly acquired" by players such as Meta, Disney, Netflix, or Amazon, to break into AI-generated media.
M&A or poaching?
While investors expect consolidation to pick up across the VC industry broadly this year, Dash said M&A is likely to surge more quickly in AI than in other industries.
"It's going to happen faster in AI because companies are raising money faster than they're growing — they're getting eyeballs and attention more quickly — but they also can fail faster," he said.
Some of these startups have recruited impressive teams and raised tens or hundreds of millions of dollars at high valuations. But Dash suggested those startups might get "stuck" when they return to market to raise again and find it difficult to raise an upround, or to raise more money at all, making some of them good candidates for acquihires "in the $50 to $150 million range," he said.
AI startups with revenue in the single-digit millions that raised money at unicorn valuations, Dash said, might not even have that option. Those deals don't make sense for buyers, he said, which could lead Big Tech companies or other AI startups to poach their talent instead.
"The smart companies will try to identify who are the best researchers, and when those startups start flailing, they'll just recruit the best from there," he said.
