Stay on Top of Enterprise Technology Trends
Get updates impacting your industry from our GigaOm Research Community
Matt Ocko had been investing in data-based companies for years before he and Zack Bogue officially launched Data Collective Venture Capital in 2012. Since then, the firm has made seed investments in nearly every hot startup in the space, ranging from database companies such as MemSQL to satellite companies such as Planet Labs. This week, Ocko came on the Structure Show podcast to talk about what has him excited as an investor and what’s just overdone.
He has interesting opinions on a lot of topics in the tech space, all of which are worth hearing, but here are the highlights from our interview. If you want to hear more from Ocko, he’ll be speaking at Structure Data (March 18-19 in New York) alongside last week’s podcast guest, Hilary Mason. It should make for a really fun talk about the future of data.
Useful doesn’t always means sexy
As sexy as the Snapchats and Slacks of the world are, Ocko is excited to see elite developers — who he says can do wonders with today’s infrastructure technologies — turn their attention toward applications such as supply-chain management or agriculture that can have “a material impact on a pretty big swath of GDP.” These spaces might seem unglamorous, and the companies and technologies might have to sneak up on the world, “a little contrarian and a little unloved,” but the rewards can be huge, Ocko said.
Just look at [company]SAP[/company], he explained:
“[T]hey said, ‘Hey, let’s integrate your accounting, and your supply chain, and your manufacturing and planning, and we’ll tell you what’s coming into your factories, how much of it is being made, how much it costs you and how much you sold it for.’ And that was transformative for manufacturing. That was more transformative, I would argue, than early industrial robots. People’s brains melted out of their ears. This was a massive operational advantage for these companies.”
Or the personal computer, which Ocko said didn’t get a whole lot of love when the concept was first introduced in the 1970s: “And it was way more transformative than any 1,000 mainframes you could have built. It created the industry that we have today.”
‘Software eats glassware’
Specifically, Ocko said Data Collective is excited about the potential for new technologies to underpin companies that can fundamentally improve lives, often by focusing on difficult scientific challenges.
“We call it ‘software eats glassware,'” he said. “We see stuff happening in computational biology and related informatics fields where you can model living things without huge capex, or even opex, and get insight into making people and animals and, heck, the planet itself a lot healthier in a way that’s consistent with the capitalist system that we all have to live in, but has profound positive impact while you’re making money.”
Betting on applications, not technologies — even deep learning
Even in the hottest of hot technology areas, Ocko says the focus from an investment standpoint still needs to be whether the technology has real and necessary applications, not just some cool research and maybe a big name. He invested in deep-learning-for-medical-imaging startup Enlitic, for example, but isn’t keen on building a portfolio of deep learning startups just because they’re getting acquired like mad right now.
“Just kind of crossing our fingers and praying for a DeepMind $400 million exit … because people in the company are so brilliant feels kind of dot.commy to me,” Ocko said. “. . . That’s just a recipe for blowing your [limited partners’] money.”
We’re good on Hadoop and marketing software for now
Even with some truly innovative and truly transformative technologies, though, there comes a point when markets become saturated. That doesn’t mean startups in those spaces will crash and burn — they could build nice companies — but they do become less attractive as investment opportunities.
One of them is Hadoop, which Ocko says still could use a lot of finessing, but has probably peaked in terms of producing huge valuations. “If you are lighting up a Hadoop cluster of some sort, there’s still a lot ‘Hadoop for x’ that’s probably needed to make your life easier,” he said. “But to your point, I’m not sure those are giant companies anymore. They may well be very nice, but not homerun — from a VC perspective — acquisitions for a Hortonworks or MapR or Cloudera.”
Another is sales and marketing software souped up with machine learning. “The number of companies I’ve seen pursuing closely related pipeline-mining opportunities, either for marketing optimization or for sales optimization, is literally over 100 now,” Ocko said. “When there are that many closely related companies with similar ideas, I think you’re potentially headed for tragedy.”