I write columns on news related to bots, specially in the categories of Artificial Intelligence, bot startup, bot funding.I am also interested in recent developments in the fields of data science, machine learning and natural language processing ...
I write columns on news related to bots, specially in the categories of Artificial Intelligence, bot startup, bot funding.I am also interested in recent developments in the fields of data science, machine learning and natural language processing
The recent acquisition of Blue River Technology by John Deere is an artificial intelligence deal at its best. It shows the impact AI can have across various industries, as well as the scale of that impact. Leaders of previous technological revolutions are paying hundreds of millions for nimble AI-focused startups. They are competing with major players like Alphabet, Facebook, Intel, and Microsoft to make acquisitions that will give them a leg up in the race for AI innovation.
The history of AI acquisitions
The race to acquire artificial intelligence startups is a relatively recent phenomenon. The most notable spike in mergers and acquisitions (M&A) in the segment started in 2016 and has continued through 2017. However, many deals took place before AI gained real popularity in 2016. Tech giants closed the earliest cognitive tech deals back in the '80s and '90s, when Apple purchased Coral and the developer of Lisp programming language. Microsoft acquired Knowledge Industries around that same time. And while it's estimated that there have been more than 250 private AI company acquisitions made since 2012, our research shows that another 50 AI acquisitions took place prior to 2012.
Despite some historical M&A activity and the recent buying spree, we don't know much about AI startup valuations and the motives of those making the acquisitions. Only 112 of the 340 acquired startups identified in our research have a reliable exit valuation. And motives behind deals are also not always fully disclosed. Apple, for example, does not generally discuss its purpose or plans regarding startup acquisitions.
Analyzing the available data
Nevertheless, analysis of 50 notable transactions (with valuations of $50 million or more) is useful as it can help entrepreneurs chart a path in an already crowded ecosystem. Such analysis could also help startups and buyers frame a possible exit strategy. And entrepreneurs can dive in deeper and explore the various rationales behind M&A deals to find the combination that is the most relevant for their individual case.
Exploring these transactions helps define at least three approaches to acquiring AI startups.
Portfolio expansion: This is the most common rationale for acquiring an AI startup. Buyers commonly seek complementarity in two ways with this type of acquisition. The first is across the value chain. An example of this would be when a company that offers tools to respond to cyber threats acquires a threat detection specialist. Another way a buyer might seek complementarity could be through horizontal expansion. An example of this would be a natural language processing company acquiring a startup that is exceptionally good at handling certain languages.
Product improvement: This rationale is also relatively common among buyers. An example of a product improvement acquisition would be when a diversified company acquires a specialist that has an edge over its existing offering.
Supply chain integration or transformative deals: These deals are the least common. They appear when industry-wide changes are expected and typically involve a buyer that is trying to keep their assets and business relevant through acquiring players across the supply chain.
Acquisitions that help buyers expand their portfolio can take several forms. Some companies in our research purchased AI startups that were operating in adjacent areas for the sake of making their own existing offerings more holistic. Others preferred to add new industry or market-specific capabilities to their portfolio.
Several notable hardware acquisitions offer good examples of companies building a more holistic offering and expanding across the tech stack. For instance, Intel acquired Mobileye's computer vision expertise as an addition to its own high-performance computing and connectivity assets. And ARM acquired Apical, which developed tech for enabling computer vision on low-powered devices.
Software companies also expand product portfolios vertically. For instance, Splunk's threat response tech was complemented by Capsida's detection capabilities, and Predix expanded its platform by adding Bit Stew's real-time data integration. Pegasystems extended robotic automation capabilities on top of its CRM system by acquiring OpenSpan.
These examples demonstrate that in order to find complementarity with a potential buyer, an entrepreneur has the option to look upward and downward in its tech/product stack.
An alternative way to spot a potentially complementary company is to look across several industries, functions, and markets that a potential buyer covers or could attempt to approach. For example, Nuance acquired SpeechMagic in order to improve its position in health care and speech recognition. Through its acquisition of Guavus, Thales got exposure to applications in predictive maintenance. And Pitchbook's ability to mine data from private companies was a strong addition to Morningstar, a stock market analytics specialist.
Horizontal exposure can facilitate improvements to an existing offering, but it is more typically seen as a way to approach new markets.
When a company is acquiring startups in order to expand its portfolio, whether vertically or horizontally, it must consider not only in the technology of the target but also its commercial footprint. The majority of AI companies acquired under an expansion paradigm were commercially successful. Mobileye generated hundreds of millions in revenue, OpenSpan was running on 200,000 desktops, and PitchBook was expected to generate several million per quarter in revenue when Morningstar acquired it.
Portfolio expansion deals come latest in the startup life cycle. It takes years to build up a business and approach the stage when it becomes a viable target for a corporation.
Acquisitions for product improvement are executed when a large and typically diversified company buys a specialist and integrates the new tech into its own products. No new markets are actively pursued in this case. For example, LinkedIn scooped up Bright.com and integrated its matching algorithms into its resume search. Twitter purchased Magic Pony to improve its video processing capabilities. And Microsoft purchased Equivio to bolster the eDiscovery functionality of Office 365.
Buyers that are acquiring an AI startup with the goal of improving an existing product generally appear to be less interested in the company's commercial footprint than those who acquire with the goal of portfolio expansion. Out of 20 product improvement deals, only three cases we looked at showed evidence of a buyer's interest in the target's traction. Therefore, founders willing to dock their AI startup to a corporation's existing product may focus more on tech excellence.
Product improvement deals appear to be created expressly for early-stage startups. These deals are sealed almost 2X earlier in a startup's life cycle than portfolio expansion deals. This is probably the point in time when a piece of technology has proven its viability but commercial traction is still to come.
Supply chain integration or transformative deals
Supply chain integration is the rarest type of deal. These acquisitions appear when an industry expects a major shakeout or a certain company is attempting to project its existing products into new markets. In these cases, corporations try to gain control of assets that will either capture the highest value in the future or boost sales.
DeepMind is probably the most notable - though not the largest - AI acquisition I'll mention. It is also a good example of a transaction aimed at reshaping an entire company. Alphabet, which is still an advertising business, is rebuilding itself as a ‚??machine learning first company‚?? and is growing its tiny non-Google revenues.
Three out of five of the largest transactions in AI fall into this category. Cruise Automation and Argo AI were acquired by Ford and GM, respectively, for the purpose of keeping automotive giants relevant during the major industry shakeup associated with self-driving cars.
Elsewhere, Mako, a robotic-surgical systems maker, was acquired by Stryker, a medical device manufacturer, with the idea that this acquisition could boost adoption of Stryker's implants and help it defend itself against competitors who would now have less control over the supply chain.
Sometimes the connection between an acquired asset and a buyer's bottom line is less straightforward. For example, by acquiring Blue River Technology, John Deere will enable farmers to spend less on herbicides. This will hopefully allow the farmers to reallocate funds for the purchase of additional machinery. The most notable feature of transformative deals is the way they work across the supply chain.
Unlike portfolio expansion or product improvement deals, where a chip maker typically acquires another chip maker or a software company scoops up a smaller software player, transformative deals take place between companies that hardly play in the same field. Therefore, in order to identify a potential buyer, a founder should look at their startup from a global perspective and identify its place in the relevant supply chain. Unsurprisingly, transformative deals can have the highest value, as there is a lot at stake for the buyer.
What these classifications mean for startups
The AI marketplace is becoming overcrowded, and many companies are adding ‚??AI‚?? to their names just to get funded. As it becomes increasingly difficult to find a niche or chart an exit, imagining a future buyer's potential motive is a good exercise to help solve the equation and build a company with real value.