Are we in the early innings of an AI M&A spree?
... and if so, who are the likely bidders at the table?
Since Microsoft effectively acquired OpenAI in January of this year (they acquired ~49% of the Company for a whopping $10B), there actually hasn’t been much M&A activity in the world of AI, despite massive amounts of hype and anticipation.
That all changed last month when Databricks acquired MosaicML for $1.3B and Thomson Reuters acquired CaseText for $650M on back-to-back days. These represented large acquisitions in both the model / tooling layers and in the application layers, respectively, and have been speculated to be the opening gunshots of an AI arms race.
Much has been written regarding where economic value will accrue in the age of AI (see our discussion on Moats from a few months ago). If you believe most of the value will accrue to incumbents who have a natural distribution advantage (e.g., Databricks, CaseText), it only makes sense that they would look to acquire AI talent and capabilities that they can immediately cross sell to their existing customer bases.
But with two players in very different industries now “off the board”, are we about to see a game of musical chairs in which the big incumbents in every industry pick dance partners? If so, who are the likely bidders going to be, and what types of acquisitions should we expect to see in the coming months?
The Buyer Universe
As we’ve discussed ad nauseum, building AI models can be prohibitively expensive, and as such their valuations are very high. To take a promising AI startup out in this environment, bidders will likely have to pay up. Using the MosaicML transaction as a starting point (and because I like round numbers), let’s assume that a marquee transaction in this space would be a $1B+ acquisition. While that price might sound high for such a nascent category, Mosaic was taken out well before they crossed the $10M revenue threshold, before they had their third birthday, and before they reached 65 employees (Mosaic was acquired for ~$21M per employee. Is that the new “page views” multiple of the AI era?). As the category continues to heat up, I think more and more strategics will make the leap to acquire AI teams and products.
But who out there can actually make $1B+ acquisitions of cash burning businesses? Turns out, there are more than a few suitors. To come up with a buyer universe, I think we can pretty quickly eliminate financial buyers (e.g., Private Equity Funds). Most modern AI businesses don’t have the free cash flow profile (at least for the foreseeable future) to appeal to a financial buyer, and they don’t have the benefit of the massive synergies to make the math make sense.
Turning to narrowing down the list of strategic buyers, I think there are two criteria that make sense as north stars:
Who can literally afford to buy these businesses (i.e., who has billions of dollars of cash on hand)?
Who are the horizontal technology platforms that can benefit the most from embedding modern AI into their platforms?
While businesses often use debt financing to make large acquisitions, I don’t think that will be happening here: these startups are burning a lot of cash and I’m not sure the debt markets are going to react favorably to these types of deals (not to mention, debt financing can take a lot of time, and I expect these types of deals to happen quickly).
Yes, of course companies can always issue stock to make acquisitions as well, but I think cash reserves burning a hole in a their pockets (particularly if they’re held overseas for tax reasons) make it easier to get board approval.
In terms of business characteristics, large horizontal tech platforms stand to benefit (or lose) the most when it comes to AI adoption. While vertical focused tech / media companies (e.g., Disney) have lots of cash on hand, I don’t expect them to be bidders in the near term. The same goes for large financial institutions swimming in cash (e.g., JP Morgan). While I expect these players to be big customers of AI companies, I don’t see enough strategic value there to acquire a pricey AI startup at a massive revenue multiple. They may, however, take out smaller vertical AI applications that have started to generate some real revenues already (such as Reuters and CaseText).
So where does that leave us? Below I’ve pulled together a list of 15 bidders who I expect to be “at the table” when it comes to making some of these mega acquisitions. At the bare minimum, I think that these players will all consider making large AI acquisitions at a Board level. To pull this list together, I’ve pulled 15 of the largest horizontal technology platform companies out there (excluding those located in China, for obvious reasons), and sorted them by cash on hand. Before any investment banking analyst comes after me: no, I haven’t manually scrubbed their Balance Sheets for this exercise: Pitchbook’s data is good enough for me.
Why is this group likely to buy?
Clearly, recent private markets valuations in the space and Mosaic’s acquisition price have indicated that AI companies are treated as premium multiple assets. While most publicly traded cloud businesses trade at 6-7x NTM revenue, AI companies are trading at much higher multiples (in Mosaic’s case, well over 100x ARR). To believe that’s fair, you have to believe that these businesses are going to grow exponentially over the next few years, and faster than traditional software businesses over the long term (we discussed this topic at length a few months ago). Given the massive consumer interest in AI applications such as ChatGPT (the fastest growing App of all time), more and more investors and strategics are willing to believe that this might be true.
For these buyers, however, there’s a second reason they’d be willing to pay premium prices for these assets: existing distribution advantages (and therefore, revenue synergies). These large horizontal technology platforms all already touch millions of consumers (either directly, or indirectly) with their products. If they’re able to acquire AI capabilities that they don’t have in house and can then immediately embed those into their products and cross-sell them to their existing customer bases, they have the ability to dramatically increase the revenues associated with their acquisitions. In other words, they can actually pay a lower “adjusted price” (or “Pro Forma Adjusted ARR Multiple” for my nerdiest of readers) for assets than someone without material revenue synergies.
This is widely cited as the reason that Microsoft acquired half of Open AI for $10B: they’re now embedding Open AI’s capabilities into their Office products (TBD if for a premium or not), and cross selling Azure compute to customers who want to seamlessly integrate with ChatGPT / GPT-4. While Open AI had almost no real revenues at the time, Microsoft saw an opportunity to sell best-in-class AI capabilities to their existing customer base. As of now, the public markets have continued to react very favorably to the Microsoft / Open AI deal (MSFT is up ~40% YTD).
There is a third reason, however, I think that many of these businesses will be bidding on AI startups: as a defensive maneuver. These days, everyone in the public markets is concerned about disruption from AI startups. In some extreme cases, some public stocks have fallen as much as 50% in a day after admitting that they’re facing headwinds competing with Generative AI. For massive tech companies that don’t have strong internal AI teams, acquiring a startup helps to protect their existing revenue base from disruption, and helps to assure their investors that they’re not falling behind the technology curve. As long as teams with modern AI expertise remain dramatically undersupplied in the market, companies that stand to lose a lot from AI disruption have to think critically about acquiring their way out of a talent gap.
Way too early predictions on who’s next to make a move
Revisiting the above list, I think a few names jump off the chart to me as folks who I think will strongly consider an acquisition in the next year or so. Before getting into predictions, I think it’s simplest to disqualify a few bidders first:
First of all, Nvidia stands to gain from powering everyone in the AI universe, so I don’t think they’ll pick a horse. Instead, I expect them to double down on their R&D spend to maintain their lead in AI chips.
Secondly, Microsoft has already made their bet on Open AI and is now tied up in a separate hairy acquisition in Activision Blizzard, so I think they’re immediately out of the running.
While Alphabet has yet to make a big acquisition in the space, they already have a lot of very strong AI talent in-house. While they’re always the most talked about potential bidder in the press and do have a lot to gain from cross-selling AI products, I don’t know if they have a real AI talent gap internally. I’m going to guess they actually stay out of the bidding wars over the next few months.
Lastly, Meta has already aggressively leaned into the Open Source AI movement. While I’m still skeptical that has real legs, I don’t think they’ll be acquiring an AI startup any time soon.
So who does that leave us with? Below would be my best guesses:
Apple: The company famously has a “too much cash” problem and has been pressured by investors to acquire something or buy back shares. They also don’t seem to have the cutting edge bench of AI talent that OpenAI / Google have. It’s only a matter of time until we have AI models running directly on our iPhones. Why not make that part of the native Apple experience? Notably, Apple has never been very acquisitive, but that might have to change
Amazon: While they do have an AI offering in their Sagemaker platform, much has been made in the press about how Amazon is falling behind in the AI race (so much so that they felt the need to address it publicly). As the trillion dollar cloud war heats up and Google and Microsoft lean into AI offerings, does Amazon feel the need to acquire a world-class AI team of their own to catch up?
Salesforce: Having just raised a $500M generative AI fund, developments in the category are clearly top of mind for Salesforce. I would not be surprised if the internal Corp Dev team at Salesforce looked at every one of those investments as a potential acquisition for the broader parent: there are just too many synergies to ignore.
Oracle: One of the most acquisitive companies on this list, Oracle has never been afraid to purchase strategic assets. Combine that with their expertise in a highly complementary area to AI (data management), and a renewed push to compete as a compute provider for AI businesses, and I think we’ve got another real contender.
Adobe: Despite recently retaliating against image generation companies with their launch of Firefly, I still think Adobe is a dark horse to bolster their team with more AI talent. The combination of clear distribution advantages in their existing platforms and the looming threat of disruption from Stability or Midjourney makes me think they’ll definitely think hard about buying.
Snowflake: Lastly, despite having a rather meek cash balance in comparison to the above players, I think Databricks’ acquisition of Mosaic was a shot across the bow at Snowflake. While they already picked up Streamlit last year, I think Snowflake is likely to look long and hard at supplementing their Data Warehouse dominance with an in-house AI team.
Will any of these six players actually look to make a big splash in the next 12 months, or have I drank way too much VC Kool-Aid already? Only time will tell.