Here are the deals the 'AI arms race' could drive in 2025, according to 4 bankers from Goldman Sachs, BofA, and Axom Partners
Goldman Sachs; Bank of America; Axom Partners
- Macroeconomic signs, like lower interest rates and a new administration, point to a rise in M&A.
- Bankers anticipate more AI dealmaking to benefit data and infrastructure companies.
- Execs from Goldman Sachs, Bank of America, and Axom Partners outline their 2025 predictions.
AI offers the promise of a dealmaking gold rush in years to come, and investment bankers are looking to cash in on deals involving data and infrastructure companies selling the proverbial pickaxes and shovels.
"Everybody's rightly caught up in the lore of AI and what the industry leaders are doing regarding building out platforms, but people forget; unless you have good data management and good data integrity, you can't fully deploy any application of AI," said Neil Kell, Bank of America's chair and global head of technology, media, and telecom for equity-capital markets. "Companies that are looking to be acquisitive are focused on this," he said.
Brandon Hightower, a founder and partner at the tech-focused M&A advisory firm Axom Partners, attributes an increase in AI-themed deals to "an arms race" earlier this year around infrastructure and talent.
Those are two themes that are expected to see continued momentum, according to interviews with four AI bankers. Tech companies focusing on managing, moving, and securing data will be at the forefront of the AI M&A wave. Other "pickaxe and shovel" companies include those focusing on developer tools and resource optimization. AI dealmaking is also poised to touch non-technology sectors, like customer service, commerce, and industrials.
While there hasn't been a lot of dealmaking activity among pure-play generative AI companies, tens of billions have been spent on technology companies that are building infrastructure around AI like storage, server, cloud, and software companies, according to Scott Denne, a principal research analyst at S&P Global Market Intelligence.
In 2024, some $82 billion was spent on AI- and AI-related acquisitons, up from about $55 billion in 2023, according to data from 451 Research, an arm of S&P Global Market Intelligence. This includes purchases of companies selling AI products or products built with AI, as well as companies that sell software, hardware, and other tech to support AI development and deployments.
There are several reasons the timing may be right for companies to shop around. Lower interest rates have reduced the cost of borrowing. The gap between buyer and seller price expectations is shrinking as companies, AI ones included, reset their valuations. Election uncertainty is behind us, and president-elect Trump's pick of Andrew Ferguson to run the Federal Trade Commission is looking like a positive for Big Tech, according to Wedbush Securities.
Ferguson was tapped "at a key time in the AI arms race in which we expect the strong to get stronger as Mag 7 gets the engines started up again on M&A," Wedbush analysts wrote in a note to clients Wednesday.
All eyes on data
Because generative AI is still relatively nascent, bankers are focused on the raw ingredients that are critical to AI models'success.
"Two of the most interesting areas to watch next year will be data infrastructure, management, and analytics companies. The second is developer tools," Jung Min, the co-COO of Goldman Sachs' TMT division, told BI.
Also important is data security, according to Bank of America's Kell, which he said continues to be very relevant and vital for those companies thinking about M&A.
Eventually, the front-facing application layer will be the biggest area, he said. But "in that creation phase, the infrastructure and the developer tools, those are the things that really matter first" after the models are trained, Min said.
That's because companies are trying to figure out "how to get better quality data" and "more efficient flows of data," Axom's Hightower said.
It's something even the biggest AI companies are opening their wallets for. When Axom worked on the sale of database analystics firm Rockset to OpenAI earlier this year, the deal came down to enabling faster data retrieval and improving the data pipeline, according to Axom cofounder Alan Bressers.
The focus on data infrastructure is spotlights two big-data giants Databricks and Snowflake, which respectively have made a handful of acquisitions related to data and AI this year.
"If you have data and if you own the models, those are two key components. How do I bring those together? And if you've got Databricks and Snowflake holding a lot of enterprise data, that's a natural place for a future winner in the AI world," he said.
Optimizing the back end
A need to scale is also driving tech companies to acquire infrastructure players, Axom's Hightower said. He pointed to Nvidia's purchase of Seattle-based OctoAI, a deal that Axom worked on "so that it can scale and do certain aspects of AI workflows in a more efficient manner," he said. It's at least Nvidia's second acquisition this year with an eye toward scalability.
Earlier this year, Nvidia set out to buy Run:ai. The deal, which is still in the regulatory approval process, could help the chipmaker run compute more economically by allowing more work to be done on fewer chips.
There's also a greater focus on lowering inference costs, essentially the price of asking an AI models a question and having them generate a response. While it's well known that training LLMs can come with astronomical prices, getting them to beam back an answer might come at an even steeper price. It's something that AI adopters are learning the hard way, prompting a "very near-term thing where you can see M&A because people can say there's real dollar value from inference savings," Axom's Bressers said. He noted potential buyers interested in this part of the tech stack could be the newer generation of cloud companies, like CoreWeave or Lambda.
Some cross-sector action
While most of the AI and AI-related deals will likely be between tech companies, Goldman's Min anticipates some transactions in the industrial space.
"Companies that already automate or help automate the supply chain, a lot of those interaction are already software to software or machine to machine, so those are ripe for AI to really deploy there," he said.
Other fertile grounds for acquisition going into 2025 include customer service and customer-relationship management companies, potentially spurred by some of Salesforce's recent AI ambitions, Axom's Hightower said. This year, the CRM giant Salesforce made "a hard pivot" to AI agents with a product that lets clients build their own custom ones to interact directly with clientele. Competitors like ServiceNow, Braze, Klayvio, and HubSpot to broaden their suite of solutions and add data and data connectivity to their offerings, he said.
