Over the last few months, the usual debate around unicorns and bubbles seems to have been put on hold a bit, as fears of a major crash have thankfully not materialized, at least for now.
Instead another discussion has emerged, one that’s actually probably more fundamental. What’s next in tech? Which areas will produce the Googles and Facebooks of the next decade?
What’s prompting the discussion is a general feeling that we’re on the tail end of the most recent big wave of innovation, one that was propelled by social, mobile and cloud. A lot of great companies emerged from that wave, and the concern is whether there’s room for a lot more “category-defining” startups to appear. Does the world need another Snapchat? (see Josh Elman’s great thoughts here). Or another marketplace, on-demand company, food startup, peer to peer lending platform? Isn’t there a SaaS company in just about every segment now? And so on and so forth.
One alternative seems to be “frontier tech”: a seemingly heterogeneous group that includes artificial intelligence, the Internet of Things, augmented reality, virtual reality, drones, robotics, autonomous vehicles, space, genomics, neuroscience, and perhaps the blockchain, depending on who you ask.
The venture world seems to be divided on the topic. The hype is certainly here, and VC money has been pouring in, but my sense is that, deep down, many investors are uncomfortable with those areas. Many worry that “frontier tech is the new cleantech”: a sign that VCs are just desperate to find the next big thing, rushing into domains they know nothing about.
As is obvious from this blog, I’m a big fan of frontier tech – but hopefully not blindly enamored with it, as an investor.
In defense of frontier tech
I love the term “frontier” because it beautifully captures the opportunities and perils involved. It sounds like a wide-open opportunity, adventurous, perhaps even romantic. But it’s also a place where you have a non-trivial chance of ending flat on your face with a bunch of arrows in your back.
A few thoughts:
First, it’s awesome. It’s hard for anyone with a reasonable level of inner-geekery not to get excited. We’re talking about a whole new world of appearing in front of our eyes, one that was complete science fiction not just twenty or thirty years ago, but also as recently as three or four years ago. What’s fascinating to me is that all those areas are going from science project to functional products that actually work at the same time. It’s spectacular, and also truly, exponentially world-changing. (for examples, see Chris Dixon’s fun post here)
Second, it’s not as heterogeneous a group as it may first appear. Most of those areas have the same common core, which is the newly-found ability to capture, process and analyze massive amounts of data, cheaply and quickly enough. Of course, they benefit from many other key trends (mobile, cloud), but the IoT, drones, robots, self-driving vehicle, etc. are really fundamentally all about data – both capturing tons of data, and fundamentally powered by data and AI (machine learning, computer vision).
Third, it’s inevitable. We VCs can pontificate all we want, and to some extent we’re gatekeepers, but ultimately entrepreneurs drive the bus. And today, frontier tech is increasingly what many of the smartest uber-geeks want to work on. Different parts of frontier tech are at different parts of the hype cycle. But on the whole, this is where the world is going.
What’s tricky from an investment perspective
Before there was open source, AWS and the lean startup methodology, it would take millions of VC money for any startup just to start building a product (most of that money going to Sun and Oracle). Then for the last 10 years or so, startups have been able to do more, faster and with less money. It became possible to get a sense of product/market fit much faster.
From that perspective, frontier tech is a throwback to the old days. It most often involves building some “deep tech”, rather than assembling pre-existing pieces. That can take a long time. Things get even trickier whenever hardware is involved, which happens quite often.
In addition, frontier tech startups operate, by definition, in very early markets. Any customer using their products is a super early adopter, generally in deep experimentation mode.
As a result, frontier tech startups often show up at Series A and Series B rounds looking considerably “behind” in terms of traction and metrics, compared to their non-frontier peers. For example, most IoT startups in the US seem to need a full Series A (in addition to large seeds), just to be able to actually ship their product. By the time they show up for their Series B, they’ve hopefully had some early sales success, but typically don’t have enough units in the field to be able to evidence their claim that, ultimately, they’re all about the software and the data.
Sometimes, traction is further constrained by regulation, for example in drones, self-driving cars and genomics.
The lack of metrics makes things uncomfortable for investors. If you don’t have MRR and churn numbers, then you need to make a leap of faith, which is always scary. To anchor their conviction, frontier tech investors need to dive deep into many new technical areas — a tremendous amount of work to do it well, and perhaps a recipe for disaster, depending on who you ask. You may get comfortable making the investment, but it’s unclear whether next round investors will – as an early stage investor, you may find yourself occasionally needing to support the company in later rounds more than you normally would.
The new incumbents are all over it
There’s an additional wrinkle for frontier tech startups.
In tech, there’s always a 800 pound gorilla somewhere. In the 90s, VCs would ask entrepreneurs “couldn’t Microsoft replicate this overnight?”. In the 00s, it would be “why can’t Google do this?”. But for the most of the history of the Internet, startups have had the relative luxury of facing mostly big, slow, traditional non-tech incumbents in their respective industries.
Fast-forward to today, and startups are increasingly facing a very different set of incumbents – companies that often were themselves startups not that long ago, and have now grown to be massive. The list is well known: Amazon, Apple, Google, Facebook, Salesforce. Others, like Uber, Tesla or SpaceX, are rapidly joining the club.
Those “new incumbents” behave in a very different way from the old ones – they’re digital natives, they’re aggressive, they’re fast. They’re obsessed with their own disruption, and they’re willing to take moonshots.
And they’re all looking at frontier tech and saying: “mine”. This is happening remarkably fast. In most segments of frontier tech, there wasn’t much of a time period where brand new startups truly had a chance to grow before a large incumbent showed up. Google jumped very quickly into IoT through the Nest acquisition, Facebook essentially jumpstarted modern VR with the Oculus deal, and self-driving cars seem to be shaping up from the very beginning as a race between Google, Uber, Tesla and others. Amazon came up with the first bona fide IoT hit (the Echo), more so than any startup or Kickstarter project, has a substantial drone program (as does Google).
AI is shaping up to the ultimate battleground, with everyone making a huge push.
Much has been said about the new incumbents’ formidable advantage when it comes to owning gigantic amounts of data. There are nuances to this – some of the smartest and most deeply technical entrepreneurs I know are precisely working on the problem of making AI work with small data, and already doing quite well. But we’re not quite there yet, and volumes of quality data do matter.
Of course, incumbents are also willing and able to attract and retain top talent at all costs, through compensation, startup acquisition or occasionally outright poaching (Uber and the CMU robotics department).
On the other hand, the new incumbents open source very significant amounts of technology (Tensor Flow, etc.). Depending on how cynical you are, you can view this as “doing no harm” or as a very effective tactical weapon to claim territory in key areas (before a startup does) and to keep their top tech talent happy by satisfying their need to change the world.
Without a doubt, the new incumbents’ appetite for frontier tech has already made a number of entrepreneurs very rich, very quickly, either directly (Nest, Dropcam, Oculus, Skybox, DeepMind, Otto) or indirectly by putting pressure on non-tech incumbents (Cruise), and created great returns for their investors. There’s also been countless smaller acquisitions, especially in AI, that worked out quite well for the founders financially. In fact, one of the key difficulties of investing in AI is that so many companies get snapped up so quickly.
However, for frontier tech entrepreneurs who are less interested in a quick outcome and want to go the distance, and ultimately aim to become tomorrow‘s Google or Facebook, the path to success is narrower.
What I look for
Given the above, what do I look for in frontier tech startups?
Technical founders with deep intellect. Of course, in any startup, you want smart founders. But in frontier tech, given all the above, we’re talking off-the-charts smart people with deep technical knowledge of the areas they operate in, and some significant relevant professional experience on top. This type of intellectual caliber can be evidenced in different ways, but more often than not, frontier tech is the land of PhDs (or PhD dropouts) in technical areas, ideally from top schools.
Deep customer focus. One key risk with the type of founders mentioned above is that they build “tech for the sake of tech”. You want founders who are also naturally customer-oriented, and happy to start working with at least one key beta customer very early in the life of the company.
Pragmatism and patience. This is particularly important in early, undefined, yet already competitive markets. It can mean keeping the burn rate low, to make sure the company is still alive by the time your market actually takes off. Or building a team that is at least partly internationally distributed, because you’re probably not going to be able to outbid Google or Facebook for that deep learning engineer in SF or NYC.
Thoughtful market positioning. Easier said than done, but stay away from areas that are likely to put you in direct competition with the giants sooner rather than later (for example, in AI, image recognition, video recognition or language translation are likely to be tricky). Everything else being equal, in terms of initial market positioning, enterprise is safer than consumer, vertical is safer than horizontal, and tools are safer than platforms. Whenever you can, position yourself as “the glue” between multiple platforms – for example, a solution that enables cross-compatibility across all VR headsets. Or as the “last mile” between their platforms and the end user.
Especially in AI, clear focus on smart data acquisition strategies in the short term (here‘s a a great post on the topic) and building data network effects in the medium term.
Conclusion
I believe the current wave of innovation (mobile, cloud) still has plenty of legs to it, and will continue to produce great companies. Enormous industries (financial services, healthcare) still are early in their adoption of those technologies.
At the same time, while frontier tech covers different areas at different stages of advancement, it altogether feels like an inevitable future. It may offer interesting challenges to investors, but it is where the world is going, and the genie is out of the bottle. For entrepreneurs, as long as you go into it eyes wide open, it is perhaps one of the most exciting times ever to start a new venture.
Good article Peter. I would also add that the way in which these frontiers are tackled has also changed through greater collaboration between start ups, technical PHDs ( or drop out PHDs) , entrepreneurs who have come out of the technical industry and visionaries. Working on shared ideas and solutions to build and improve to the point it creates the new market.
Matt,
As a startup focused on working at the intersection of data acquisition hardware+AI+delivering actionable solutions for agriculture clients your post struck a chord with me on many levels.
I hope more startup capital will take a more nuanced approach like you mentioned to frontier tech rather than viewing all tech through the consumer/cloud/instant scale model lens.
We believe tech+agri to a great extent solve the impending food supply problem (http://bit.ly/1qN9Ww5) and your post reaffirms our mission.
Thank you!
-Jai
Not sure if the paragragh about “pragmatism” comes from experience or it’s just a prediction, but our direct experience confirms it 100%. Just to get the best talent in the AI space (ML and NLP), we’re operating out of two cities (Boston & Montreal) and will soon extend to 3 and possibly 4.
Very much from experience. I hear great things about the ML/NLP talent in Montreal. Will need to come visit at some point, long overdue.
We’ll be happy to show you around if you happen to be in Montreal. Lots of talent, specially in Deep Learning and NLP.