By Sheana O’Sullivan
I recently attended the Tech M&A Summit in Palo Alto, hosted by global research and advisory firm 451 Research. This is a leading event for Corporate Development leaders who drive M&A decisions for the largest tech companies in the Valley, and a great source of information about where acquirers’ interests currently lie and where this market is headed. People were very vocal about key trends in M&A and, in particular, had a lot to say about what’s happening in machine learning (ML).
We’re seeing unprecedented growth in ML M&A. Around 140-150 ML transactions are forecast to be completed this year: that’s almost three times more transactions than in 2015. The total M&A market forecast is 3,800 deals so machine learning makes up only 3-4% of deal volume for 2018. This may be a small slice but it’s actually the fastest and most dynamic part of the overall market.
What’s behind this growth?
It may still be early days for machine learning but it’s seen as a transformative technology with a growth trajectory that is already unparalleled in the market. Companies are using it primarily to gain competitive advantage, improve customer service, and better respond to threats and opportunities. Interestingly, the motivation here is not financial, as you might expect. Increasing sales and lowering costs are way down the list of positive benefits. Working better and smarter is the main driver behind this surge in machine learning innovation.
Who are the biggest players?
Machine learning is the top ranked theme in M&A, primarily focused on business analytics (29%), sales & marketing (22%), security (16%) and customer service (4%). Another area that is active includes chatbots, for example companies like HubSpot are using ML-enabled chatbots to help small businesses – a sort of democratization of machine learning.
CB Insights has outlined the most active ML acquirers as below:
Google is the most active of startups, having acquired 11 startups since 2012. Apple has also been ramping up its M&A efforts and ranked second with seven acquisitions. Facebook has done a number of transactions in this space, including Bloomsbury.ai (which was a deal that FirstCapital advised).
What are the issues?
One of the Corporate Development leaders I spoke with commented that they see companies increasingly saying they’re offering machine learning, just to attract attention. Engineering then has to kick the tires and get under the hood to identify if it really is machine learning. His perspective was that what most companies are advertising as ML technologies are actually just elementary data analysis and basic machine learning. If it does turn out to be “true” machine learning, it’s going to be expensive.
Of course, the value of machine learning is largely determined by the underlying data and whether something valuable can be extracted from this raw material. Big, vertically integrated data is ideal for ML technology. ML companies that can’t demonstrate access to the required level of data tend to get put into the acqui-hires bucket for general tech acquisitions. But note that acqui-hires in this space start at $2m per head and can go as high as $10m.
The general consensus is that most ML companies are still small, with very few established ML companies in the market. Areas where there are larger ML companies include voice and autonomous driving, as these ML solutions tend to solve big problems.
Should Valley acquirers be looking to Europe?
One of the points I made to this (largely Valley-based) audience is the growing relevance of ML companies in Europe. Since 2012, the top 20 AI/ML companies in Europe have raised more than $2.6bn, and we are tracking around 1,000 companies in this space in Europe.
This is not well known in the US, and indeed one attendee remarked that we tend to see the first versions of great tech in the US before copycat models ripple out into Europe and the rest of the globe. In his view, some of these companies would then have the benefit of learning from their predecessors and innovating in an interesting way, meaning that he thought there could be good deals in Europe but doubted the originality of the innovation.
But we at FirstCapital wholeheartedly disagree with this. European valuations may tend to be lower than those in the US, which is more down to differences in availability of funding and a different attitude to sustainable growth over a reliance on funding, but the data shows that much world-leading ML innovation has come out of Europe. Indeed, around 40% of all acquisitions in this space have involved European companies. One of the pioneering deals was Google’s acquisition of Deepmind in 2014, but other notable deals include Intel’s acquisition of Mobile Eye, Apple buying both Vocal IQ and Shazam, Twitter acquiring Magic Pony, as well as the Facebook/Bloomsbury.ai deal mentioned above.
It’s going to be an exciting road ahead for machine learning. Watch this space!