10 Thoughts from a VC on Enterprise Tech in 2017

With 2016 solidly in the rear-view mirror and 2017 stretching out ahead, I thought I’d join the ranks of VCs taking a stab at some predictions for the future. My focus has always been on investing in early-stage enterprise tech companies, and so I’m going to focus this post on some thoughts on enterprise software going forward, including a few themes and trends I’ll be watching for.

Here goes:

1 AI everywhere / AI inside. Artificial intelligence (AI), machine learning (ML), and deep learning (DL) have been the dominant buzzwords in the venture world for at least the past twelve months. I expect the significance of this technology to increase even further in 2017. I expect we’ll see the confluence of three major sub-trends here:

• Everyone is going to say it. First, all manner of startups will continue to use these terms to dress up their product offerings in the hope of attracting venture capital interest and even customer adoption. Despite the hype, in many cases, these terms are being used to describe fairly simplistic rule-based or statistical engines.
• Nearly everyone is going to be able to do it. Second, the core set of algorithms that enable AI, ML, and deep learning are going to be increasingly commoditized and are not going to represent meaningful barriers to entry for startups nor are they — in and of themselves — going to be sufficient to drive customer adoption of new products.
• The transformative power of deep learning is best understood as a force multiplier. AI (and deep learning in particular) are going to bring massive value to a subset of startups (and to their customers and VCs), but this value isn’t going to come from algorithms alone. It’s going to come from how those algorithms are packaged into powerful enterprise applications that are easy to buy, deploy, and understand. Deep learning has changed the applicative power of data, but it has not changed the fundamental physics of enterprise software. Deep domain expertise is still going to be required to deliver truly valuable applications. In a world where any startup can leverage powerful compute capabilities and deep neural networks, it’s going to increasingly be domain expertise, product management skill, and sales that will separate the winners from the pack.

2 Conversational interfaces. Another major shift in the enterprise is going to be the emergence of conversational interfaces in true enterprise-grade, customer-facing applications. In both voice and text settings, this will become the easiest way for humans (customers and employees) to engage with computers. We’re already seeing a range of startups providing both the building blocks and the applications that will drive adoption of this technology by the enterprise. My portfolio company ServiceFriend is hard at work in this space.

3 New data sources. As mentioned above, the transformative power of analytics and deep learning will not be a function of commoditized algorithms but of other factors. One of the most important is data — and the definition of “data” is rapidly expanding. We’ve lived through one massive transformation in the addition of “unstructured” data thanks to distributed databases and technologies like Hadoop, ElasticSearch, and MongoDB, but that was just the prelude. We are already in the early phases of an explosion in the amount of useful data available to software. This is driven by three factors: First, we are deploying more sensors than ever before. This includes mobile devices in the hands of customers and employees, but it also includes the world of IoT which will generate masses of (often continuous) useful data. Second, advances in audio and image processes are transforming masses of audio and visual information into useful sources of data. My portfolio company Chorus.ai, for example, is running advanced analytics on voice sales calls. Third, the availability of virtually unlimited computational power on the cloud (together with very powerful edge devices) is redefining the parameters of “parsable” data. Datasets that, ten years ago, would have been too massive to be useful (hundreds of temperature readings per millisecond, for example) are increasingly trivial to analyze. This opens up a world of possibility for software.

4 Penetration of old industries. Of the top five verticals in terms of growth in the number of deals financed in Europe/Israel in 2016, three were “old” industries: automotive, agtech, and general industrial. This trend is driven the availability of new data sets described above, by the arrival of a new generation of industrial operators eager to bring the benefits of software to traditional industries, and by sheer pressure to drive operational efficiency. I expect this trend to intensity into 2017 and beyond, creating tremendous opportunities for startups. This is going to drive further SaaS verticalization as highlighted as early as 2012 by Gordon Ritter of Emergence (see also these great pieces by Nic Poulos at Bowery and Tomasz Tunguz at Redpoint). The broader point is that new software delivery models (SaaS), new untapped data sources, and a new generation of IT buyers are going to make the enterprise software world much bigger than ever before — and that’s going to be mean a ton of opportunity for verticalized SaaS players in both verticals that are already well-served by horizontal players and in brand new verticals. For example, my portfolio company Imubit is still in stealth but is busy bringing the power of deep neural networks to some very old industries — with surprisingly powerful results.

5 Security remains the perennial problem and opportunity. I’m not going to touch the US election with a ten foot pole (see this previous post for thoughts…), but if nothing else, the election and other events throughout the year further highlighted the importance of information security. As long as there is technology, there are going to be vulnerabilities that will require new solutions to address them. That’s never going to be a core competency for any enterprise, and so there will always be demand for new security products and approaches. The challenge for investors is to identify security-related opportunities that might be large enough to define new categories and entrepreneurial teams talented enough to execute in this very challenging market. My portfolio company Snyk is addressing the challenge of code security with a developer-first approach (the dominant logic is that if software is eating the world, developers will eat pen-testing and infosec audits). A second portfolio company, Siemplify, is addressing the challenges that overwhelming amounts of data are posing to human security operations teams and -you guessed it — applying machine learning to help reduce the work-load at enterprise SOCs. There are other very compelling opportunities in automotive security, IoT security more generally, and security for new containerized/microservice infrastructure.

6 On the infrastructure side, APIs, microservices, and containerization drive flexibility, innovation, and competition. I’ll no doubt be doing a lot of thinking about IT infrastructure in 2017. We seem to be in the early innings of a massive shift in the way software is written and deployed. Software is being broken down into smaller components (microservices), deployed in highly flexible, HW-agnostic infrastructure units (containers), and communicating internally and externally with robust protocols (APIs) that enable tight application integration regardless of infrastructure, both within and outside the firewall. This creates a ton of opportunity for smart emergent infrastructure players to solve some new pain points (security, orchestration, deployment, debugging). It also creates an opportunity for other startups to start to attack newly vulnerable infrastructure stacks. I’m spending a lot of time with a company, for example, that is working on a microservices-driven application infrastructure stack specific to the healthcare world. I’m still gathering my thoughts here — but there is clearly a revolution underway with massive opportunities for both application and infrastructure players.

7Data portability. In the past, data would get generated where the compute power was located, and analytics would take place where the data was located. So if your transactional systems were on-premises, the data would get generated on-prem, and you’d set up your OLAP cubes on-prem as well. With centralized cloud compute, data gets generated on the cloud, lives on the cloud, and gets analyzed by cloud-based tools. I’m not sure this model is going to hold going forward. Data is getting generated on the edge (mobiles, IoT, drones, etc.) and edge compute is becoming increasingly powerful, which may draw application software towards the edge, even if the bulk of data lives on the cloud. Finally, distributed applications would benefit from truly distributed databases. Today’s databases aren’t really built to deal with the challenge of data that needs to live in multiple locations, and needs to be consumed both on the edges and the core of the network. Data and databases remain one of the last obstacles to application integration and infrastructure flexibility, and I think we’ll see some interesting companies emerge to address this problem.

8 Open source get put into perspective. Open source has transformed over the past decade from taboo to core. With software eating the world, developers are eating the enterprise — and they are winning increased freedom to deploy whatever solution they prefer. In many cases, open source is a critical driver of enterprise adoption because it limits lock-in and prevents over-dependence. That said, driving profit growth from the open source model seems to be increasingly challenging for many startups. The days of software entrepreneurs blinding rushing to go fully open source in order to drive massive adoption and then “make it up on volume” later (or turn into glorified services organizations for their biggest users) are over. I think we’ll continue to see a proliferation of high-quality open source tooling (software does “want” to be free), but the challenge for entrepreneurs (and VCs) is going to be less about how to make money from open source software and increasingly about how to build a valuable product that the enterprise will pay for in an environment of open source competition. Open source is, I suspect, going to be best understood as part of a strategy but not a strategy in and of itself.

9 Development tools are going to look a lot more “enterprisey” and less “dev tooly.” As the lines between business and software blur, the lines between “developer tools” and “enterprise software” are going to get increasingly blurry. I’ve argued above that open source is both increasingly core and increasingly problematic as a business model. But the other major shift is that development teams are increasingly strategic to the enterprise and therefore increasingly in control of significant budget for tools that have a meaningful impact on the ability of the enterprise to achieve agility. I suspect that in 2017 and beyond, the most interesting dev tooling companies are going to possess some serious enterprise-grade chops — more emphasis on sales quotas than on Github stars.

10 Drones, VR, and IoT. I’m a huge believer in the value of VR/IoT/drones for the enterprise. Companies like Dronomy, AiRobotics and many others are beginning the prove this out. Similarly, companies like Mindesk are exploring enterprise-grade CAD applications for VR. Companies like Resin.io (a portfolio company) are building the infrastructure that enterprises need to deploy IoT at scale. I’m looking forward to more such opportunities to invest in the intersection of the enterprise and the physical world.

So that’s some of what I think I’ll be looking out for this year. If you’re building something for the enterprise in Europe or Israel, or you think I should be thinking about any of this differently (or about something else entirely), ping me. Let’s talk.

My AngelList syndicate to back the best in European & Israeli enterprise companies is now one of the largest syndicates based outside the US. We are now at well over $1M in backing per deal.

The syndicate has made five investments so far: all oversubscribed and all with quality co-investors. I’d be honored if you’d consider backing the syndicate — you’ll be in pretty good company and there are quite a few awesome companies in the pipeline…

If you like this sort of post, please consider subscribing to this blog.