AI investment is set to flow mainly to established companies such as OpenAI in the short term. |
|
Funding of AI companies has not returned to the high levels seen during the COVID-19 pandemic despite a surge of interest over the last year.
M&A activity involving AI companies is down 37% from its peak, while private equity and venture capital-funding deal volume has fallen by 44% over the last couple of years, according to S&P Global Market Intelligence.
Complexities inherent to artificial intelligence, the cost of developing AI models and higher interest rates have restrained investors from spending freely across the sector. In the short term, most AI funding will flow to established foundational model companies like OpenAI LLC and Anthropic PBC, according to Nick Patience, lead analyst for AI and machine learning research at Market Intelligence.
"Some will go away, some will go bust, some will [get acquired] and one or two might make it as the big dog in their space," Patience said.
M&A slowdown
All told, there have been a little more than 300 deals globally targeting AI entities thus far in 2023. That is down from nearly 500 in 2021, when tech buyers were flush with cheap debt and eager to take advantage of the digitalization trend sweeping the enterprise and consumer markets in the wake of the pandemic.
The fourth quarter is set to be the weakest in terms of deal volume, and the second weakest since early 2020 in terms of aggregate transaction value.
Predictive analytics is already baked into most software applications, rather than being a separate category as it was a decade ago, which goes some way to explaining the dearth of AI deals, said Brenon Daly, research director for tech M&A at Market Intelligence.
Daly expects to see continued smaller acquisitions for what he termed "smarts," mainly existing products and key engineers, by larger brands.
A recent example is Adobe Inc.'s planned purchase of Indian text-to-video generation platform Rephrase.ai for an undisclosed amount, which comes shortly after the official launch of Adobe's own family of generative AI models, Firefly. But that transaction is unlikely to spur dealmaking more broadly. Daly instead said the three biggest tech investors are likely to continue to dominate.
Google LLC in October agreed to invest up to $2 billion into Anthropic, a San Francisco-based generative AI startup that is hoping to rival ChatGPT parent OpenAI. Before the Google deal, Anthropic had also recently secured a $4 billion investment from Amazon.com Inc. Meanwhile, Microsoft Corp. is estimated to have invested $13 billion in OpenAI.
"Running AI is hugely resource intensive … so it is the domain of the deep-pocketed few," Daly said.
M&A activity in general has been in something of a slump for more than a year, with the total number of deals falling to its lowest level since before the pandemic in the third quarter. A root cause is higher interest rates, which has made potential buyers more cautious. In the AI space, that means the days when a platform could trade on its user growth are gone; investors are far more interested in profitability.
This presents a problem for many potential targets, as "[generative] AI isn't really generating any [revenue] these days," Daly said.
Funding rounds
The low level of M&A reflects the fact that generative AI is a nascent market, according to Market Intelligence's Patience.
There has been a consistent flow of funding round investment into AI companies over the last several quarters, but both the number and value of deals are significantly down from their peak in late 2021. Private equity and venture capital-backed deals in the third quarter totaled 180 in number and $4.11 billion in value, down from 497 and $13.50 billion in the fourth quarter of 2021. Those totals exclude the recent mega-investments by Google, Amazon and Microsoft.
Another key risk to dealmaking is around intellectual property, which OpenAI's recent firing and rehiring of CEO Sam Altman brought to the fore. Microsoft's share price fell immediately after the firing, which may have reflected concerns about intellectual property that may not be under the direct control or ownership of companies using it, credit rating agency DBRS Morningstar said in a Nov. 29 report.
A lot of AI funding is contingent on the company being able to demonstrate that the IP in source code is properly protected, according to Hyder Jumabhoy, co-head of EMEA financials M&A at law firm White & Case. Often IP can reside with the founder or the person who wrote the code, and it needs to sit within the businesses themselves so founders cannot simply terminate licenses and walk away if investors fall out with management.
White & Case will typically ask to see a full suite of documents showing the IP registrations, where the source code comes from, and whether it is open source.
"The moment you start pulling the threads, a lot of it starts to unravel," Jumabhoy said.
'Platform shift'
The underlying forces driving investment remain strong. Generative AI represents a platform shift that fundamentally changes the way humans interact with software and will catalyze upgrades to PCs, mobile phones and other hardware, Market Intelligence's Patience said.
For private equity and venture capital, there is still an appetite to invest in businesses that can show profitability. Sovereign wealth funds have entered the space as investors over the last 12 or 18 months, Jumabhoy said. Singapore's Temasek Holdings (Pvt.) Ltd. helped fund AI car-parking company Metropolis Technologies Inc. in May, in the third-largest funding round since 2020, Market Intelligence data shows.
But the technology must transition from proof-of-concept to production and must become more specialized before M&A deal activity in the sector gains steam, according to Goldman Sachs' Global Banking & Markets team. Policymakers across the globe will also need to crystalize legal and regulatory frameworks around the technology before potential buyers can feel confident in making an investment, the team added in an October white paper.