An overview of machine intelligence deals and exits for H1 2015
For more AI news and analysis, sign up to my newsletter here.
Machine intelligence, the field of building computer systems that understand and learn from observations without the need to be explicitly programmed, is a core investment focus for us at Playfair Capital. Here, I present a quick overview of the deals and exits for companies operating in this arena in the first half of 2015.
I queried data sources including Crunchbase, CB Insights, press releases, SEC filings and our own deal flow to draw up a dataset of 87 unique companies completing deals between January 1st and June 30th 2015. These companies were selected because they mention “artificial intelligence” and/or “machine learning” in their descriptions/marketing copy. As such, the core technology areas covered include computer vision, machine learning, deep learning, natural language processing and generation, data science, robotics, and speech. Here’s what I found:
Deal and round sizes in H1 2015
In this 6 month period we saw 93 deals, 35% of which were at the Seed/Angel stage. Series A and B financings accounted for 15% and 11% of deals, respectively. For comparative purposes, consider that on a global multi-sector level Seed/Angel, Series A and B financings accounted for 29%, 23.5% and 15% of deals in H1 2015, respectively (CB Insights). On the other end of the company lifetime spectrum, 19% of machine intelligence deals were M&A transactions. One company, Adgorithms, IPO’d on the LSE.
When looking on a $ invested basis for machine intelligence financing rounds, we see that median round sizes for Seed/Angel and Series A deals are $1m and $5m, respectively. Both of these figures are lower than their peers the broader market. The significant sums raised in growth stages are skewed by the $55m Series C for Ayasdi (industry agnostic topological data analysis) and $28m Series D for Lattice Engines (e-commerce predictive analytics). Total $ raised indicates the aggregate capital raised over the lifetime of companies in each segment.
Sub industries with activity
The most popular industry applications for machine intelligence on a deal number basis were business intelligence (BI) and analytics (29%), followed by security and monitoring (12%). Within BI, there were 2 Seed/Angel deals, 4 Series A, 6 Series B and 9 M&A transactions. Within security and monitoring, there were 6 Seed/Angel deals, 2 Series B, and 1 M&A transactions. The median deal size was less than $8m in each industry across stages, excluding debt rounds for consumer lending companies Applied Data Finance and Argon Credit.
The U.S. accounted for 72% of the deals by number and 94% of the dollars invested into machine intelligence companies. Europe came in second with 18% of the deals, but only 5% of the capital invested. Taking a closer look within Europe, 59% of the deals and 42% of the capital invested were for UK companies. In contrast, the US accounts for 60% of financing $ and 65% of deals, versus 10% and 15% for Europe when considering the global VC market data for H1 2015.
The median M&A deal came in at $105m, while Adgorithms raised $42m in the IPO valued at $127m. Of note, 10 of 12 exits were for North American companies. The median time to exit from founding was 3 years across all 18 M&A deal. In 9 cases, companies raised financing rounds a median of 394 days prior to their acquisition. The median team size across M&A exits was 7.5 employees. Twitter and Google were the most active acquirers, each purchasing 2 companies (Whetlab/TellApart and Timeful/Granata Decision Systems, respectively).
Next, we explored two indicators of value creation efficiency: exit consideration as a function of a) human capital (# full time employees (FTE) at exit), and b) financial capital (total raised prior to exit). The former returns the value created per FTE (value/FTE) and the latter the return on capital invested (ROIC), both plotted below.
Here, we find that Twitter’s $532m acquisition of 6-year-old, 14 person-strong TellApart returned the highest multiple for shareholders (30x ROIC) and did so in the most human capital efficient way ($38m/FTE). Contrasted with Saba Software, an 18 year old public company that was taken private by Vector Capital for $300m, which was the least financial (6.8x) and human capital ($0.41/FTE) efficient of the four.
Take home points
The first half of 2015 has seen mostly early stage financing activity behind companies building machine intelligence technology and only few exits. It’s clear that the U.S. is the place to be for fundraising and exiting. One must also consider that there’s probably a bias in the birthplaces of AI companies that favors the U.S. versus other continents and thus has a knock-on effect on what happens next.
There’s still room to innovate beyond the sectors traditionally addressed by AI, such as BI and sales. For example, consider that popular machine learning patents include medical diagnostics, data mining, and image/video analysis (Quid 2015, below).
These translate into investments within the medical sector, robotics, and computer vision occurring since 2010 (Quid 2015, below).
Let’s see what the next half of 2015 has in store for us! Stay tuned.