Qumulo’s New CEO: ‘We Do Need An AI Story’

“[We’re focused on] applied AI as opposed to building a training and learning cluster. And to me applied AI means inference workloads, using enterprise core data, making that data part of the answers that AI is generating for you while ensuring accuracy. This, without that data becoming part of the model, without ceding control of an enterprise’s data. I'm gonna do that,’ says Qumulo’s new CEO, Douglas Gourlay.

New CEO For A New Era

Distributed file and object storage technology developer Qumulo this month unveiled Douglas Gourlay as its new CEO, taking over from Bill Richter who is retiring after nearly eight years of running the Seattle-based company.

Gourlay, who spent the last five-plus years as vice president and general manager for software products at Arista Networks, told CRN in an exclusive meeting that Qumulo’s technology, which brings cloud-based and on-premises data together into a single global namespace that makes it available from anywhere, is ready to tackle the need for data that comes from the widespread adoption of AI technology.

But first, he said, he has to make sure channel partners and enterprise customers are aware of what Qumulo has to offer for AI.

“We do need an AI story,” he said. “I know what it is. [It is] applied AI as opposed to building a training and learning cluster. And to me applied AI means inference workloads, using enterprise core data, making that data part of the answers that AI is generating for you while ensuring accuracy. This, without that data becoming part of the model, without ceding control of an enterprise’s data. I’m gonna do that.”

That does not mean chasing the massive GPU installations, Gourlay said.

“I want to build the systems that support municipal governments, police, body cam storage, and every security camera,” he said. “I want image recognition to better work across all of those and provide a better defense environment than they've had before.”

Here’s more of CRN’s interview with Gourlay, slightly edited for clarity.

How do you describe Qumulo?

Qumulo to me is the coolest storage company I’ve seen because it’s the first one I’ve seen that broke the barrier between cloud and on-prem. It’s implementing a global namespace in a way that your data, to paraphrase a movie I saw, is literally anywhere, everywhere, all at once. Everything I have, it’s everywhere I want it to be instantaneously with high durability. This means you can shift from having to think about storage, what media, what physical location you're putting it in, and the economics associated with it, to a different conversation: Where does my data need to be for my business?

If I can treat cloud and on-prem with relatively consistent unit economics, it eliminates the entire conversation we’ve been having as an industry about cloud repatriation, or do we go all in on cloud, do I go hard on cloud, do I come back from the cloud, do I refactor my apps for the cloud? A large portion of that is storage-related, but not all, but a solid double-digit portion is. And that argument disappears. I spent the last five years having conversations with customers where, if you have scale and expertise, you need to maintain a data center, but if you don't have the scale or expertise, you should consider the cloud. The key attributes cloud has that you can't do in an on-premises data center are elasticity and the forward projection where you don't have physical presence. So if you want to spin something up in Singapore, you can spin something up in Singapore in five minutes and not have to fly kit over there to do it. And the cloud is phenomenal for that.

Where the cloud is generally weaker is in governance and observability and visibility into the infrastructure substrate and its impact on applications and data storage performance. But all of a sudden, 90 percent of the arguments I used to make are gone. Now I can use Azure, Amazon, and on-prem with a common management platform. I'm only paying for what I use. So many of the arguments start disappearing, and then it's a shift of value.

What are some of your strategic priorities for the second half of 2024?

First, I want to update the company's narrative and storytelling and message. I think we have such an incredible base of technology that addresses a broad set of market requirements. But the hard part of that is it sometimes when you address a broad set of market requirements, customers are afraid to try you on something small. So I need to think through the narrative, a bit of storytelling, the emotional part of it, that captures people's hearts and says this is something we need here in our business whether it's for cost, agility, or risk reasons. I think we address all three of the primary purchasing criteria that a mainstream enterprise has. But I want to get the narrative right that tells the story of the technology and its impact on the business.

Second, we do need an AI story. I know what it is. I already told you part of it, which is applied AI as opposed to building a training and learning cluster. And to me applied AI means inference workloads, using enterprise core data, making that data part of the answers that AI is generating for you while ensuring accuracy. This, without that data becoming part of the model, without ceding control of an enterprise’s data. I'm gonna do that. That's gonna be really important. A top priority I hear from CIO after CIO is that they need to do something with AI. ‘What do you want to do?’ ‘I don't know. But we need to be ready in case our competition does something with AI that's really valuable. We need to do something quickly thereafter.’ Or they have a research project on AI and they’re kicking the tires. A big tech titan, one story. A hedge fund trading on data, a different story. Nation states doing whatever nation states do to each other, different story, different resourcing levels. [These companies need] the capability, the human capital, the infrastructure spend, to build their own models. And some of the largest ones I’ve looked at, we’re talking 100,000 GPUs. … I say to myself that I don't want to chase that market if I’m not already leading it. Instead, I want to build the systems that support municipal governments, police, body cam storage, and every security camera. I want image recognition to better work across all of those and provide a better defense environment than they've had before.

What is Qumulo’s AI story?

Well, it's not chasing the 100 organizations in the world that have the technical chops and wherewithal and extremely large budgets to go build their own 25,000-node clusters for learning and training and building a model. It’s back to the mainstream enterprise.

I talked to one of the top four banks in America and they said, ‘Yeah, we got eight DGX-1s. We're not really sure what we're going to do with them. We're figuring it out. It's a good research project.’ They said in all likelihood, OpenAI or somebody’s going to build the world’s best large language model, and then somebody will do one better, and somebody will do one better.

What they wanted was, ‘How do I use that LLM with my data without my data becoming part of that LLM? I want a prompt that says, “Write an email to my top performing sales rep last quarter, and thank them for the great job and include the numbers.” And then I want to hit enter and have the data come out.’

You are taking over as Qumulo’s CEO from Bill Richter. Has he left the company?

Bill is retiring. I asked Bill to stay on as an advisor to me and as an advisor to the board. We didn’t put a specific end date on that. Bill has a wealth of knowledge about the company, about what’s worked and what hasn’t, about why decisions were made. And I have to say, I could not have asked for a better and more professional handoff and transition that I've experienced with Bill, and I’ve seen them a couple of times in my career. They can be rocky.

A random story: At a house I was once doing a remodel, and I told the contractor, OK, I want high quality, I want it on budget, and I want it on time, pick two.’ And he said, ‘Well, I’ll get you high quality, and I’ll get you budget, but it might take a few extra weeks.’ When Qumulo started its CEO search, Bill made a similar comment to the board. He said, ‘No one’s rushing me out of here. I’m not rushing to something. Let’s take the time to find the right candidate. And you’ve got me as long as you need me to ensure we get the right person and then to make the transition smooth.’ He was just so professional, and so great to work with.

What can you share about Qumulo’s financial side?

As a private company? What I can say is, we’re growing. I can say we have over 1,000 customers, and our 1,000 customers span a very broad array of markets from media and entertainment to national defense, municipal governments, universities, research labs, healthcare, consumer electronics, big box retail, banking, and mortgage. It's almost scary in the diversity of the customer base we have because it makes it hard to tell a sales rep to pick one market and lead the charge. I think what we're finding, back to the company's charter, is broad market applicability for mainstream enterprise. We are seeing significant international growth as well, which I found surprising. A very positive surprise. I like a good international-domestic mix. But we are seeing excellent growth outside the U.S.

How about the channel and Qumulo?

We introduced some new products last fall, an Azure-native version of Qumulo. And now we have in private release our AWS-native version of Qumulo. And without disclosing the numbers, early indicators are that these could be the fastest-growing products we have offered. They will offer a very quick rate of return and good turnover for the business. And bluntly, they're things that should be very attractive to the channel. We have already reached out to several channel partners, of the largest in the U.S. as well as some regionals that I've developed good relationships with over the last quarter century now, and asked them the simplest question I could: What kind of channel program can I put in place that blows your mind? And what is the best channel program possible for us to engage with organizations like WWT, Presidio, Bear Data, or Nereus Cyber Systems? I mean, full spectrum small to large.

Is Qumulo cashflow positive?

We are not cashflow positive, although we have significant operating incomes and I believe we have the ability to be managed to be cashflow positive if necessary. Without putting a number on it, there are two ways to solve that problem. One is growth. And the other is a little bit of headcount management, which is our major spend. The growth target to achieve cashflow positive isn’t an unattainable one.

Any plans to go back to for another round of funding?

The best time to raise money is when you don’t need it. The best way to not need it is to be cashflow positive and have a zero- to 1-percent operating margin and then run your business sustainably, profitably, and properly. I'm starting with the thesis of growth, supporting our organization with an up tempo that they're probably not used to. I'm used to doing 95 percent of my time on the road and 10 to 15 customer meetings a week. I'm going to have to dial that back a little bit and be present in corporate now. But an increased op tempo, especially with an updated narrative that addresses key customer priorities, should result in increased sales. The increase in sales that is necessary for us to not need any outside capital, as I had earlier indicated, is a manageable number that is not out of reach. And I'm not looking at it going, “Oh my god, we’ll never get there.’ This is measured in quarters or years. And to your point, it’s then we have the option. And I like having options. That’s safety. That’s risk aversion.

Any plans for an IPO?

I get asked that a lot. I’ll tell you a slightly different story. I don’t have an exit strategy. I don’t believe in that. I believe if I build a sustainable, profitable, high-growth business that generates good gross margin because it delivers valuable products that customers appreciate and give us purchase orders for, if we have a solid net dollar retention rate and great customer support and a high NPS (net promoter score) score, we get to choose our destiny.