Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


Interview: Thomas Thurston on Bringing Data Science to StartupHPC

Thomas Thurston will make a special appearance at the StartupHPC Meetup at SC14 on Monday, November 17, in New Orleans. In this special feature, he gave us his candid perspectives on three areas: industry advisor, startup wisdoms and personal/professional development. 

An Industry Advisor

insideHPC: What (types of companies/products/ideas/individuals) do VC’s typically look at in the HPC space for their portfolio?

Thurston-Pic-copy-300x300Thomas Thurston: Our fund is different because it uses algorithms and compute-centric tools to help us decide which startups to invest in. The things we care about probably aren’t representative of what “normal” VCs do. Most VCs pay a lot of attention to the founding team, who else is going to invest in the deal, and what their intuition tells them about the business as a whole. It’s all about backing successful people, teaming up with sexy co-investors and following a hunch.

This surprises a lot of entrepreneurs because focusing on teams and co-investors doesn’t say much about the business itself, and it doesn’t help first time entrepreneurs or companies in other situations. Some of the most successful VC investments like Facebook didn’t start out with high-profile founders and investors, but they were obviously incredible investments.

In contrast, our models pay a lot of attention to the business itself, and to longer-term technology trends, rather than the team or co-investors. We’re looking for businesses that innovate to lower the costs of current solutions, or otherwise tap into new customer populations by simplifying solutions or making them more accessible. For example, I’ve been a long proponent of ARM, arguing long before it was fashionable that ARM cores would move up-market into more advanced computing platforms and even HPC. People thought I was pretty stupid for saying that eight years ago, and a lot of them still think that, but every year it’s a lot more obvious we called it right.

On the other hand, we don’t fund companies that are bringing better performance solutions to existing markets. For example, years ago our models predicted SiCortex would fail. SiCortex was an amazing HPC startup; great team, great technology. But our algorithms predicted failure, and that’s what happened. A challenge in HPC is that most startups are, almost by definition, bringing higher performance stuff to existing markets – and those are the deals we avoid. HPC is defined as the frontier of high performance, so it’s especially challenging. However when those rare gems come along that fit our profile, we get really excited. For example, we’ve been bullish on Pure Storage and Flash arrays for quite a while.

insideHPC: Who do you (investors) track and follow as part of finding opportunity? Advisement? Etc. Everyone is following you, Thomas, so who are your (and the VC) influencers?

Thomas Thurston: As we were just discussing, VCs are very influenced by what other co-investors are circling around a deal. For some, it’s a lazy substitute for diligence. They think “hey, if Andreessen or Kleiner are in the deal, it must be good.” A sad reality of present-day venture capital is that investor branding is often substituted for substance, especially in HPC, biotech or other highly technical areas where the average investor doesn’t really understand the subject matter. I have a ton of respect for Andreessen, Kleiner and other amazing funds, but I’d be pretty lame if that was all I cared about.

We spend more time tracking the performance of various technologies over time, and modeling how changes in one market might impact another. We take a systems view. We see things more as an ecosystem that’s always shifting, and try to anticipate where those shifts create opportunity. For example, today SaaS is old news, but a decade ago it was still bubbling up. We realized that SaaS was extremely disruptive to many client-server markets and placed a lot of bets that performed extremely well. However today things are changing again, so we try to model what the next surge will look like and what startups are likely to lead.

insideHPC: Do VC’s in general take advantage of conferences to go opportunity shopping (like SC14, and its ‘disruptor show case’ for example)? In other words How do VCs interact with/engage Startups? Is it luck? Don’t call us, we’ll find you? Pitchfest? Angel Network …???

thomas-thurston-1000-304Thomas Thurston: There are a variety of VCs, so maybe I’ve already overstated what the “typical” VC does, but in their mind there’s a small cadre of elite entrepreneurs in the world who have been successful in the past. There’s an assumption that anything these cool kids touch is going to turn into gold, so the whole game is about getting the cool kids to take your money. This basic assumption, which I’d argue is flawed, drives an enormous amount of VC behavior. So while they go to lots of startup events and conferences, it’s usually to keep a finger on the pulse of an industry rather than to find actual deals. Some deals happen this way, but it isn’t how VCs prefer to source opportunities. There are exceptions such as Y Combinator events, but I wanted to touch on the rule more than the exception.

Think of it like a middle school cliché; everyone’s sucking up to cool kids and ignoring everyone else. So if you ask VCs, there aren’t enough good deals, but if you ask startups, there aren’t enough investors.

insideHPC: At what point in VC investments, and portfolio, for example, is a start-up perceived as successful? When does a company advance from start-up to ‘mature’?

Thomas Thurston: There are a few ways to answer this. If they’re raising a new fund, a successful startup just has to show ever-higher valuations from follow-on investments. This lets the VC give good, concrete signs that the portfolio is growing in value. Later down the road, success is defined by a high exit – either a startup gets acquired or goes IPO.

Start-up Wisdom

insideHPC: What would be your top 3-10 ‘must do’s’ for start-ups? Top 5 don’t do’s!

Thomas Thurston: Let me approach this from a different angle. The three biggest mistakes I see startups make are: (1) underestimating the competition, (2) underestimating the costs, and (3) underestimating the internal forces that will emerge within their own organizations over time. This is especially the case for number 1– underestimating the competition. Startups habitually downplay their competitors rather than acknowledging them and tackling the issue head-on. I almost always go online and search for a company’s competitors while they’re giving me their pitch. If I find significant rivals the startup either didn’t mention or address, I’ll call them on it right away. Startups who try to sugar-coat the competition give me an allergic reaction.

insideHPC: What has been your greatest challenges/successes with your own start-up?

Thomas Thurston: The hardest thing about growing my current start-up is scaling rare expertise. It’s very hard to grow a business when it’s very dependent on specific bases of knowledge. We’ve been very aggressive about automation as a way to scale around these constraints, and we’re somewhere around the 80 yard line, to use a football metaphor, but aren’t in the end-zone yet automation-wise.

Personal/Professional Development:

insideHPC: What advice would you give to individuals – STEM students/professionals – as your top 3 personal/advisory ‘must do’s’ for their own personal/professional development? How does someone elevate themselves to ‘your’ level of industry recognition/regard … what’s your recipe?

Thomas Thurston: Don’t try to out-innovate the competition. Instead, try to find populations of people who have zero solution today because they lack the money, expertise or access to solutions. In other words, don’t ask existing customers in the market why they buy, ask non-customers why they don’t.

There’s still time to buy tickets for the StartupHPC Meetup at SC14, which takes place on Monday, Nov. 17

Comments

  1. Great article Mr. Thurston. It seems so obvious / logical when you state things so clearly. Well done.

Resource Links: