‘When you believe a technology is a commodity, you stop investing in it or you focus only on cost, whereas if you look at it as high technology you want to consistently be doing two things. One is pushing bigger, faster, cheaper. That’s been high technology since the beginning. But you should also be changing paradigms, changing the market. And that’s how we look at it,’ Pure Storage CEO Charles Giancarlo tells CRN in an exclusive interview.
Pure Storage has over the last decade-plus grown from an upstart flash storage array builder to become one of the top providers of data storage in terms of capacity shipped.
But along the way, the Santa Clara, Calif.-based company changed its focus from data storage to data management, a move that not only culminated in its recent introduction of its Enterprise Data Cloud for managing multiple data stores as a single system, but also made it a key part of the burgeoning AI market.
Pure Storage Chairman and CEO Charles Giancarlo, in an exclusive meeting with CRN said that his company, by focusing on storage as high technology instead of a commodity, now outshines its competitors including NetApp, Hewlett Packard Enterprise, Dell Technologies and Vast Data.
[Related: Pure Storage CEO: New Enterprise Data Cloud Gives Partners A ‘Huge Consulting Opportunity’]
“When you believe a technology is a commodity, you stop investing in it or you focus only on cost, whereas if you look at it as high technology, you want to consistently be doing two things,” he said. “One is pushing bigger, faster, cheaper. That’s been high technology since the beginning. But you should also be changing paradigms, changing the market. And that’s how we look at it.”
Giancarlo also said that Pure Storage is the only vendor to make data truly ready for AI.
Traditional storage offerings require customers to acquire more storage capacity to make data available for AI and don’t allow that data to be used for AI in real time, he said.
“What our Enterprise Data Cloud does is allow all the data in a customer’s environment to be seen as a cloud, and therefore it’s accessible as raw data so AI can work on it directly,” he said. “So this concept of the Enterprise Data Cloud opens a customer’s data environment to be useful for AI. We think that’s a different architecture, a big opportunity, for Pure.”
Giancarlo also discussed his views on high-performance computing and AI, tariffs, Broadcom and VMware, and Nvidia.
There’s a long going on in the fast-evolving data storage industry and Pure Storage’s place in it. Here is more from CRN’s conversation with Giancarlo.
How has the definition of Pure Storage changed in the last year?
Before I get to the definition itself, I think what’s really changed is that we’ve gone from our start with just a single product whose innovation was to introduce flash in a significant way to the enterprise to now effectively offering a full-service data storage environment for our customers overall. But I think the best way to define Pure is that we’re the company that continues and will continue to innovate in the data storage space. What really differentiates us from our competition is from our beginning we looked at data storage as high technology whereas most of our competitors, for a very long period of time and I think this haunts them today, looked at it as a commodity. When you believe a technology is a commodity, you stop investing in it or you focus only on cost, whereas if you look at it as high technology you want to consistently be doing two things. One is pushing bigger, faster, cheaper. That’s been high technology since the beginning. But you should also be changing paradigms, changing the market. And that’s how we look at it.
When you look over your shoulder, who do you see?
I don’t look over my shoulder very often to the extent we see anyone challenging us. To put this in context, we’re still, in terms of total storage, No. 3 or No. 4. And so we have to focus a lot on the guys that are bigger than us because a lot of our opportunity is taking market share. When you talk about looking over your shoulder, I don’t look at the guys that are losing velocity. We have to look at the teams that are increasing their velocity. I would say the only one that has really given us some challenge, partially technologically but more from an overall marketing standpoint, would be Vast [Data]. But I feel technologically, we’re not going to have a problem with them. We still have some catching up to do in terms of reputation in the HPC [high performance computing] space.
Is HPC a focus of Pure Storage or not?
Well, AI has its roots in high-performance computing, and so a lot of the GenAI teams that have been constructed, whether it’s in the hyperscalers or in large enterprise tech titans, come from that HPC background. HPC has always been a niche area, but now, in a sense, as AI goes mainstream that HPC niche area is going to be more mainstream. We never participated in the HPC environment. Therefore, we didn’t have the background, we didn’t have the sales teams that were conversant in that area. We didn’t specifically have some of the product attributes that were necessary to be successful in that area. We’ve assembled all of that around our EXA product line. We’ve just announced its general availability, and it will be five times faster than anything else any other vendor has so they won’t even come close to us. And it has all of the ‘ilities’ or abilities of Purity. We’ll still be catching up on features probably throughout the year. But for the most part, we’ve re-established Pure and Purity as being able to cover the entire landscape of customer needs, and now in particular in the largest of all scales for AI.
Let’s twist around a previous question. Who’s looking over their shoulder at Pure Storage?
Oh, NetApp, for certain. … They’ve come out now with something called Fusion. It’s not what Pure Storage Fusion does, but it’s called the same thing.
I would [also say] Vast Data. I think we really shocked them with our EXA announcement. Why do I not put Dell in that category? It’s because effectively the entire company has at the core of its model a commodity mindset. So I think they’re very concerned about us. There’s no question about that, but I don’t think they can modify their model to be able to address it. We’re at antipodes with Dell. Dell is by far the leader in commodity products. But we compete on the basis of being a high-technology vendor. So it’s a fair fight, just different.
How about HPE?
We don’t worry about HPE. I think they’re worried about us. I think they’re very much focused on the networking side of the business now. And I think they have consistently bought their way into storage. … And I think they’ve decided that they’re going to try to maintain what they can, but their real focus is on networking.
We’re living in a changing environment, with a couple things of going on that we’d like to get your take on. First, VMware was a big partner of yours. How has that relationship changed since Broadcom acquired VMware?
Look, Hock [Tan, Broadcom CEO] has a very clear point of view in terms of how he operates organizations. He operates to consolidate power with the product, and as much as possible consolidate functionality in the product. So I think partners for Broadcom is a secondary consideration.
When you say ‘partner,’ do you mean technology partner or channel partner?
Any type of partner. Anyone that could be considered a partner. Either someone who takes some of the margin out of the deal, which would be a go-to-market partner, or someone that is, in a sense, in co-opetition, which is because VMware has always sold VSAN which, in some cases, not all, can be a competitor to a third-party storage array. But VMware realized when it was on its own that there are times when customers prefer external storage arrays. So co-opetition, right? Whereas I think the way that Broadcom looks at things is, it’s all competition. And so they’ll do what they have to do to minimize customer flight away from them, but no more than that.
Another big issue is tariffs. Any impact on Pure Storage so far?
The only impact so far is a lot of extra work planning all the possible scenarios that might take place when the tariff definitions are finally final and stop moving. Because as long as they continue to evolve—and they don’t evolve mildly, I mean, we’re talking about huge changes that occur every week—the only thing you can do is plan. You can’t take action because, unlike with changes going on in the federal government, we can’t go back a week later and say, ‘Just kidding.’ Our decisions have consequences. They have consequences for our supply chain partners, and they have consequences for us. So, a lot of planning, a lot of work. The way things seem, the problem when I say this, is they change almost every other day. The way they seem, we don’t think there’s going to be a significant effect.
Have changes in federal government organizations, the DOGE effect, had a major impact on Pure Storage?
It hasn’t. That’s both a good and a bad thing. The bad thing is, we didn’t have enough business in the federal government for it to be really meaningful for us. On the flip side, I think we are seeing [the government] as a ‘new age’ partner, if you will, a high- technology partner, rather than part of the existing infrastructure. And that seems to be a positive in the new environment.
We talked about how Broadcom’s changes are affecting the Pure Storage relationship. A couple of other big partners have made news with Pure Storage, including Nvidia and Nutanix. How important are those partnerships to Pure Storage, and what’s happening on those fronts?
Those are two very different areas. Why don’t we touch virtualization first, and then we’ll go into AI. For well over a year now, we’ve identified four different options for our customers that we have worked hard to give them to address their VMware cost issue. That’s what it is. It’s a VMware cost issue. One is that actually we can help if they stay on VMware. We can help them reduce their VMware costs, even if they were to use VSAN, for example, which supposedly is free. But as it works out, you need more cores, and so you’re still paying per core. So there may not be an extra license, but there are extra cores. We are able to reduce the cost not only by reducing the number of cores, but also because some of those very same cores also run the application itself. It could be Oracle. We reduce the application cost as well. So we provide significant cost reduction for our customers. For those customers that want to move away from VMware, we give them three different options. One is, with our work with Microsoft, they can now move to AVS [Azure VMware Solution], which is Microsoft’s VMware license. This is a very smooth transition because customers don’t have to rewrite their application, and they don’t have to do any refactoring of the application to deal with different storage because it’s our CBS, or Cloud Block Store, that works on AVS. So it makes it a very simple migration. However, you do have to choose to work in Azure rather than on-prem.
For customers that want to stay on-prem, there are two different options. One is the Nutanix option. There are many customers that are already on Nutanix for hyperconverged. We now give them a disassociated third-party storage model to be able to operate on Acropolis, which is the Nutanix hypervisor. You also have a KubeVirt, or Kubernetes virtualization, option. They can use any Kubernetes KubeVirt distribution. Red Hat OpenShift is one of those that works very well. But there are others. There’s open source, there’s SUSE, there’s a variety of others. So we give customers a choice in all four of those areas.
And Nvidia?
Nvidia is still and probably will remain for quite a long time the ultimate arbiter and driver of AI environments. Customers large and small look to Nvidia to effectively bless architectures for inference RAG [retrieval-augmented generation] or training. We’re very fortunate to have been certified by Nvidia. We’re very busy getting EXA certified by them. We believe EXA will through the remainder of this year start to become the preferred solution for Nvidia training in particular, but training and influence and inference. One of the things we do uniquely well is many of our competitors in that space require retuning their product to shift from training to inference, but we don’t. Our technology can go from training to inference, do both simultaneously in those models. We think that’s going to be a big advantage.
How important is AI to Pure Storage’s current and future product and technology plans?
It’s important in several ways, and not merely in the fact that it opens up new opportunities for some data storage. In fact, that might be the smallest of the opportunities. A very large opportunity for us is the fact that we’re introducing the Enterprise Data Cloud. It becomes extremely important in an AI environment. Today, there are only two ways for customers to use AI on their data. One is with the use of connectors into the applications where storage sits behind those applications to get the data tags that they need to run AI. But you’re going through the entire application environment to be able to get that data. As it turns out, one no longer needs to do that anywhere with AI. AI can work on the raw data. But customers’ environments are not set up to make that raw data available to AI. So instead they replicate the data from production systems into a brand-new and separate storage system. And often those storage systems require high performance, and so they’re expensive. So you have two disadvantages. One is you’ve just bought more storage even though the data already is sitting there. And two is the fact that you’ve copied the data, and it’s automatically old. It’s no longer production or real-time data. What our Enterprise Data Cloud does is allow all the data in a customer’s environment to be seen as a cloud, and therefore it’s accessible as raw data, so AI can work on it directly. So this concept of the Enterprise Data Cloud opens a customer’s data environment to be useful for AI. We think that’s a different architecture, a big opportunity, for Pure.
We are now also using AI to allow our customers to automate their data management so they can not only ask simple questions, but they can provide simple commands. They can place their policies in what we call Fusion in this Enterprise Data Cloud, and automatically all their data gets managed according to those policies.
Is there anything else you think the readers of CRN should know?
Well, yes, particularly because it’s CRN, the concept of this Enterprise Data Cloud opens up value-add opportunities for resellers. The value-add opportunity for resellers in the data storage space historically has been setting up, managing and tuning data storage, which our competitors made fairly complicated and therefore required a lot of expertise. Now, one of the things that has differentiated Pure and maybe has not created as much opportunity for our resale partners is that our products were designed to be simpler, to not require a lot of tuning, a lot of configuration. Because of that, it reduces the services opportunities, the consulting opportunities for our partners. What we’ve done with Fusion is not make the product any more complicated. The product stays just as simple. In fact, if anything, now it’s even easier to manage a fleet of these products. But what it’s now doing is allowing customers, and therefore partners, to set up data governance rules in software to manage the datasets. If you think about the issue of creating rules around compliance, rules around cybersecurity, rules around data recovery, these are very strategic issues that you want to plan at a very high level with customers, and make sure they’re rock solid, right? And then you implement them in software, in Fusion. And so that sets up a strategic opportunity with the customer. It’s a very high-level area for creating policies, for checking and reviewing policies and implementing them. It’s a wonderful service opportunity for partners to go in and operate at a strategic level with the customer.
