Prepared remarks
Good day, and welcome to Pure Storage Third Quarter Fiscal 2025 Financial Results Conference Call. Today's conference is being recorded. All lines have been muted during the presentation portion of the call with an opportunity for questions-and-answers at the end. The operator provided instructions. At this time, I'd like to turn the call over to Paul Ziots, Vice President of Investor Relations. Please go ahead.
Thank you. Good afternoon everyone and welcome to Pure’s third quarter fiscal year 2025 earnings conference call. On the call we have Charlie Giancarlo, Chief Executive Officer, Kevan Krysler, Chief Financial Officer, and Rob Lee, Chief Technology Officer. Following Charlie’s and Kevan’s prepared remarks, we will take questions. Our press release was issued after close of market and is posted on our website where this call is being simultaneously webcast. The slides that accompany this webcast can be downloaded at investor.purestorage.com. On this call today, we will make forward-looking statements, which are subject to various risks and uncertainties. These include statements regarding our financial outlook and operations, our strategy, technology and its advantages, our current and new product offerings, and competitive, industry and economic trends. Any forward-looking statements that we make are based on facts and assumptions as of today, and we undertake no obligation to update them.
Our actual results may differ materially from the results forecasted, and reported results should not be considered as an indication of future performance. A discussion of some of the risks and uncertainties relating to our business is contained in our filings with the SEC, and we refer you to these public filings. During this call, all financial metrics and associated growth rates are non-GAAP measures other than revenue, remaining performance obligations or RPO, and cash and investments. Reconciliations to the most directly comparable GAAP measures are provided in our earnings press release and slides. This call is being broadcast live on the Pure Storage Investor Relations website and is being recorded for playback purposes. An archive of the webcast will be available on the IR website and is the property of Pure Storage. Our fourth quarter fiscal 2025 quiet period begins at the close of business Friday, January 17, 2025. With that, I'll turn it over to Charlie.
Good afternoon and welcome to our Q3 FY25 Earnings call. Pure Storage delivered solid third-quarter results, with both revenues, and operating income, exceeding our guidance. As I have discussed in previous calls, we have been engaged on achieving a significant hyperscaler design win by year-end, and I am pleased to announce that we signed a design win with a top four hyperscaler in the last few weeks. This is the first-ever design win to provide flash for standard hyperscaler storage and it is the vanguard for flash storage providing all online storage in major hyperscale environments in the future. For reference, the hyperscale market is responsible today for 60% to 70% of all Hard Disk Drives purchased globally. In addition to providing cost effective data storage, this Top 4 hyperscaler’s use of Pure technology is expected to free up significant amounts of power and space in their data centers.
It is also expected to significantly reduce the failure rates and maintenance costs associated with legacy disk storage, while doubling the expected lifetime of their storage infrastructure. Pure Storage Purity Software and DirectFlash Technology will provide this hyperscale customer the necessary price, performance, density and power to deliver their cloud-based services with unparalleled performance and energy efficiency. The close engineering engagement between the companies, and extended testing by this major hyperscaler, has proven Pure’s DirectFlash technology is now capable of providing cost effective data storage at hyper-scale capacity, even at low cost bulk data, price ranges. To fully outline the opportunity here, we are working with hyperscalers to utilize our technology, with a single consistent architecture for all of their on-line storage, inclusive of low priced bulk storage, nearline higher performance storage as well as high speed storage for their most demanding use cases, including AI.
To support the expected increase in Flash demand for the hyperscale industry, we also announced today a deepening collaboration with Kioxia, a global leader in NAND Flash technology. Kioxia has been a steadfast partner in our engagement with the hyperscale community. Our design win signals that Pure’s DirectFlash technology is now ready to replace hard disks everywhere, and NAND vendors are taking notice and planning their opportunity to address this 700 Exabyte per year market. As data volumes continue to increase, our combined technologies enable hyperscalers to meet the challenge of increasing data volumes while reducing power consumption, labor, and the physical footprint of hyperscale data centers. The work for the design win we announced today started over one year ago. While we have had sales of standard product into hyperscale customers, Hyperscalers have developed their own software for their storage services which operates on commodity Hard Disk Drives and SSDs.
Our early outreach to hyperscalers was first met with skepticism that we could achieve the price and performance necessary to replace cheap hard drive storage. However this hyperscaler was open to investigating us further, and, working together, Pure optimized the design of Purity and our DirectFlash technology to fit smoothly into their compute and storage architectures, and optimized the economics to fit their financial targets. With this win, Pure is entering an exciting new hyperscale market. The design win itself signals that this top four hyperscaler’s future data centers are approved to use Pure’s technology as their data storage standard. We expect early field trial buildouts next year, with large full production deployments, on the order of double-digit Exabytes, expected in calendar 2026, which corresponds with our fiscal 2027. We continue in our dialogues with other major hyperscalers as well.
Given the significant opportunity that exists for this and other hyperscalers, we anticipate increased investment in our hyperscale Line of Business over the next year, for which Kevan will provide additional details. I would like to turn now to another significant area of opportunity for Pure, namely Artificial Intelligence. AI creates several key opportunities for Pure. First, we continue to provide leading-edge high-performance storage for public and private GPU farms in machine learning and training environments. This past quarter we were officially certified for the NVIDIA DGX SuperPOD architecture, designed to provide turnkey infrastructure for the world’s largest training environments. We recently announced a strategic investment partnership with CoreWeave, a specialized GPU cloud provider to better serve our AI customers. Building on our successful, existing, super-computing scale deployment serving thousands of GPUs, we partnered with CoreWeave to make Pure Storage available as a standard option within the CoreWeave dedicated cloud environment.
Second, many enterprises are considering inference engines and retrieval-augmented generation or RAG environments as they look to apply commercial large language models to analyze their proprietary data. This quarter, we introduced the Pure Storage GenAI Pod, a set of full-stack solutions which reduce the time, cost, and expertise required to deploy generative AI projects. In the quarter, we signed a deal with a medical device manufacturer who faced multi-million-dollar interruptions because their legacy storage technology couldn’t support a real-time AI imaging system to catch defective products. With Pure, they can now run AI analytics, capture metadata, and train their machines to identify and prevent defects, significantly improving their operations and quality assurance. Third, AI continues to drive customers to modernize and break down infrastructure and data silos to enable easier access to data.
Unlike with other vendors, Pure customers will not need to manage different storage operating environments to meet their varied AI needs. This quarter, one of the world’s leading suppliers to the defense and aerospace industry chose Pure for their AI data storage infrastructure. This Fortune 200 customer chose Pure’s platform to support a wide range of training, inference, and fine-tuning, sharing many data sets and storage environments seamlessly across multiple groups and AI activities. The Pure Storage Platform will be used to develop multiple AI technologies to enhance human capabilities, improve aviation safety, reduce pilot workload, and develop human-centric autonomous solutions. Expanding on the Pure Platform and turning to the enterprise, Pure is driving the biggest shift in enterprise storage since Flash. With Pure Fusion, we are transforming enterprise data by virtualizing data management and storage and enabling enterprises to create their own data cloud environment across their global enterprise.
Pure Fusion will be available this quarter as a non-disruptive free upgrade to all existing Pure block storage arrays, and will be standard in all new Pure block products and storage service offerings. Fusion will be extended to our file and object platforms early next year. Our advances in data storage innovation for enterprises and now hyperscalers are transforming the industry. Our experience and technology in optimizing Flash Storage for Enterprises has now enabled us to begin to penetrate the largest hyperscalers with our Purity based DirectFlash technology at the largest scale. Because of their scale, Hyperscalers manage their storage far differently than traditional enterprises. Traditional Enterprises manage individual storage arrays which are dedicated to specific workloads. Storage dedicated to a specific workload cannot be shared with other workloads. Therefore data stored for a particular workload is generally inaccessible for other workloads.
Traditional enterprise storage architectures and products create data silos. By contrast, hyperscalers only have a small handful of storage environments, segmented only by price-performance levels - low, medium and high, for instance. All data from all workloads and customers utilize the same storage environments. This makes data access far easier. Different storage capabilities are enabled by software, not dedicated hardware. Our experience in working with Hyperscalers has allowed us to bring the best attributes of data cloud architectures to enterprise data centers with Pure Fusion version 2.0, which will be released this quarter. Pure now makes it possible for businesses to build their own enterprise data cloud, seamlessly combining on-prem and cloud environments to stay agile and competitive in the age of AI. Pure Fusion automates data management, simplifies operations, and enhances the devops developer experience.
Fusion empowers enterprises to build their own enterprise data clouds that federate storage across both cloud and on-prem environments, enabling effortless scalability, global accessibility, automated job placement, load balancing, and importantly, AI-ready data access. Fusion allows organizations to define and standardize their own customized, global, data management classes, inclusive of performance, cost, resiliency, recovery, and location. And to automate delivery of data services to users via API, and according to enterprise policy. Fusion fully unifies, automates, and delivers the Cloud Operating Model across the Pure Platform, on-premises and in-cloud. We have also deepened our partnership with major public cloud vendors for enterprise services. With the official preview of the Pure Fusion powered Pure Storage Cloud for Microsoft Azure VMware Service or AVS for short, we simplify enterprise migrations from on-premises VMware environments to AVS, enabling independent scaling of storage from Azure compute nodes.
This first-of-its-kind solution ensures a smooth, efficient cloud transition with minimal IT disruption, boosting cost efficiency, data resilience, and storage simplicity. Turning now to the market and broader macro environment, we have not seen any meaningful change in the overall landscape, which remains relatively consistent with the muted IT spending and heightened competitive environments we have seen all year. Customers continue to contend with higher software, SaaS and Cloud costs, as well as AI spending uncertainty, placing unanticipated pressure on operating budgets. While I would have liked to have seen more strength from Evergreen//One in the quarter, we are confident that we are strongly positioned across all our segments. The cloud is not a location, it’s an operating model enabling self-service, speed, consistent operations, and faster scaling with greater efficiency, at lower costs.
With Fusion and our data storage platform, we’re turning the vision of an enterprise data cloud into a reality for enterprises. Our consistent innovation in our industry has been recognized annually by industry analysts, such as our recent 11th time Leader position in the 2024 Gartner Magic Quadrant for Primary Storage Platforms, and 4th time leader position in the 2024 Magic Quadrant for File and Object Storage Platforms. Energy availability is a global concern, and has become a critical risk to hyperscalers’ operations. Some are even contracting with nuclear power plants to secure a reliable supply of electrical power. Hyperscalers no longer seek low-cost power, they’re looking for power at any cost. Expanding electricity production cannot add significant capacity for many years. Alternatively, replacing inefficient hard disk data storage with Pure DirectFlash technology represents one of the largest power sources presently available to hyperscalers.
As power limitations increasingly hinder data center growth, Pure Storage is the only company that can simultaneously enable hyperscalers to cost effectively upgrade their data storage while simultaneously freeing vast amounts of electrical power and data center space for other applications, such as AI. Overall, we are very pleased with our progress on our hyperscale opportunity and with the expansion of our enterprise capabilities. I am personally more excited than ever about Pure’s opportunity ahead, as we drive a new era in data storage management. With that, I will pass the microphone to Kevan.
Thank you Charlie. We are pleased with our Q3 financial results, exceeding both our revenue and operating profit guidance. Revenue of $831 million grew 9% year-over-year while also delivering strong operating profits of $167 million. Strong demand continues for our //E family and FlashArray//C solutions, enabling customers to move their cost sensitive workloads to all-flash. Additionally, renewals of our Evergreen subscriptions across our install base remain robust, demonstrating strong year-over-year growth. Total contract value, or TCV sales for our storage-as-a-service offerings during Q3 was $96 million, reaching $253 million for the nine months of FY 2025. Conversion of larger Evergreen//One opportunities valued greater than $5 million is consistent based on our expectations that we reduced last quarter. Higher velocity Evergreen//One pipeline that we define as less than $5 million is strong, though, in Q3 we experienced a meaningful increase in these opportunities converting to a traditional sale.
This contributed to higher than expected product revenues in Q3, while also resulting in lower than expected TCV sales of Evergreen//One. Subscription services annual recurring revenue, or ARR grew 22% to $1.57 billion. Total RPO exiting Q3, which includes both subscription services and product orders, grew 16% year-over-year to $2.4 billion. RPO, excluding product orders, and associated exclusively with our subscription service offerings grew by 17%. RPO growth is impacted by lower than expected TCV sales of our Evergreen//One offering. US revenue for Q3 was $562 million and International revenue was $269 million. Our new customer acquisition grew by 340 customers during Q3, and we continue to serve 62% of the Fortune 500. In Q3, total gross margin remained strong at 71.9%, reflecting strong Subscription services gross margin at 77.4% and solid product gross margin at 67.4%. As anticipated, product gross margin was influenced by our strategic efforts to help customers transition cost-sensitive workloads to our //E family and FlashArray//C solutions.
This approach, while resulting in a modest decline in product gross margin, underscores our commitment to delivering cost-effective, high-value solutions for our customers across our data storage platform. Looking ahead to Q4, we expect continued strong growth of our //E family and FlashArray //C solutions resulting in a sequential modest decline in product gross margin, similar to the trend observed from Q2 to Q3. Operating profit of $167 million and operating margin of 20.1% during Q3 were positively impacted by revenue overachievement, strong gross margin performance and continued operating expense discipline. Our headcount increased sequentially by nearly 140, to approximately 5,900 employees. Pure’s balance sheet and liquidity is strong, including $1.6 billion in cash and investments at the end of Q3. Cash flow from operations during the quarter was $97 million, and reflects a large upfront payment for new software technology that we licensed.
Capital expenditures were $62 million and include significant investments to scale operations for our hyperscale opportunities. During Q3, we repurchased 3.6 million shares, returning approximately $182 million to our shareholders. We also paid $55 million of withholding taxes due on employee equity awards, which also offset dilution by approximately 1.1 million shares. We have approximately $213 million remaining on our existing repurchase authorizations. Turning to guidance, we are raising our FY 2025 revenue expectations to $3.15 billion, representing approximately 11 and one-half percent year-over-year growth. The raise to our FY25 revenue expectation is the result of seeing an increase in Evergreen//One opportunities, under $5 million, converting to a traditional sale. While this dynamic increases product revenue expectations, it also reduces FY 2025 TCV sales growth expectations for our as-a service offerings.
With lower TCV sales growth for our as-a-service offerings we do not expect the growth of our consumption and subscription offerings will have a significant impact on our FY 2025 revenue growth. As such, we are not updating our FY 2025 TCV sales guidance for our as-a-service offerings. For Q4, we anticipate revenue of $867 million, reflecting 9.7% year-over-year growth. Now moving to operating profit expectations. Aligned with our increased FY 2025 revenue guidance, we are also raising our FY 2025 operating profit expectations to approximately $540 million, reflecting an operating margin of 17%. For Q4, we are guiding operating profit of $135 million and operating margin of 15.6%. Before closing, I’d like to share some preliminary thoughts on our first transformative design win with a top-four hyperscaler to help inform your models. The commercial framework for this design win involves licensing our technology and delivering support services.
Hardware will not be included as part of our sale to the hyperscaler. We anticipate meaningful revenue contribution and operating margin expansion from this win beginning in FY 2027, which aligns with our expectations for full-scale production deployments reaching double-digit exabyte capacities by that time. To capitalize on this milestone, we will increase operating investments in FY 2026, aimed at scaling operations and accelerating our opportunity to deliver Pure’s differentiated technology for hyperscale storage. When considering these investments, we expect FY 2026 operating margin will be approximately 17%, consistent with our FY 2025 operating margin guidance. In closing, as we look ahead, our strategic investments and innovation position Pure as a leader in transforming the data storage landscape. We remain focused on execution, while navigating a persistently muted IT spending environment.
The growth opportunities ahead are fueled by our advancements across our data storage platform, empowering organizations to unlock greater efficiency, scalability, and resilience in their operations. We are excited about what lies ahead and confident in our ability to drive sustained growth and innovation. With that, I will turn it back to Paul for Q&A.
Thanks, Kevan. Before we begin the Q&A session, I'll ask you to please limit yourselves to one question consisting of one part so we can get to as many people as possible. If you have additional questions, we kindly ask that you please rejoin the queue and we'll be happy to take those additional questions as time allows. Operator, let's get started.
Questions and answers
Thank you. The operator provided instructions. Your first question comes from the line of Amit Daryanani of Evercore ISI. Your line is open.
Thanks a lot. Good afternoon, everyone, and congrats on the hyperscaler win. I know you folks have been working on this for a while. Could you perhaps just touch on what drove the decision by this hyperscaler to choose Pure Storage versus building this on their own, given most of these hyperscalers do have a fair amount of resources. And then, just in terms of the financial impact, I heard you folks talk about sizing it up a bit, but how big do you think at scale this revenue opportunity or TAM could be for Pure? Thank you.
Thank you, Amit, and good to hear your voice. As you know, we have been working on this for a long time, and there are a number of different factors. Ultimately, our ability to deliver a set of unique advantages convinced this hyperscaler that our technology was the best fit and that they would not need to develop it themselves. First, and in no particular order, unmatched reliability at scale: we can scale to exabyte levels and we are driving reliability down to about a 0.15% failure rate per year, which is unbelievably low and five to ten times better than typical SSDs and hard disks, reducing labor and outage costs. Second, massive power savings: we deliver roughly a 5:1 to 10:1 improvement in power usage versus hard-disk environments, which can translate to around a 20% reduction in total data center power — effectively freeing up power for the customer to use for other scale-out needs like AI.
Third, accelerated time to value: we can deliver this solution immediately with no further R&D required, and it integrates well with their architecture because our software is a processor-based architecture rather than an SSD-based architecture, fitting their virtualized, software-defined model. Finally, unparalleled price and performance: we can achieve comparable total cost of ownership to hard disks while delivering roughly five times the performance. For the economic side of this, let me pass it over to Kevan.
Yes. Thanks Charlie, and thanks Amit for your question. A couple things I'd point to just a high level for you on the financial front, including what this opportunity potentially could look like at scale. Charlie mentioned that about 60% to 70% of hard disk sold today is really driven by the hyperscaler market. And so, I think that gives you a good data point in terms of what the opportunity looks like for us long term. Obviously, that will result for us in terms of significant revenue contributions over time, as well as operating margin expansion over the long term as well. In the shorter term, I think there's a couple of considerations I would highlight that I also included in my prepared remarks. In terms of operating margin for FY 2026, we're expecting that to be flat with this year's guidance that we gave at the beginning of the year of 17%. And that's really to enable funding for scaling operations and accelerating the hyperscale opportunity that's in front of us.
We don't expect meaningful revenue contribution from this opportunity until FY 2027 and that would also include operating margin expansion. It's also important to note that we are not going to include hardware in our sale to the hyperscaler. And that's an important consideration for us as well. And then lastly, when we're thinking about revenue contribution from this opportunity, and we think about it in relation to data storage capacity purchase, it'd be significantly lower than when compared to say, a sale of our FlashBlade//E solution. So I think those are some important points for us to be considering.
Thank you, Amit. Next question, please.
Your next question comes from the line of Aaron Rakers of Wells Fargo. Your line is open. Aaron, your line is perhaps on mute.
Yes, sorry about that, guys. Hopefully you can hear me now. Thanks for taking the question and congrats on the announcement as well. I want to unpack a bit of what was just said. Kevan, my question is about the financial attributes: it sounds like this is a licensing engagement. Are you licensing Purity on a capacity-deployed basis and also licensing the DFM module, and therefore not manufacturing those components? Could you unpack those mechanics? And as we think about this, do you expect this hyperscale customer, as they move into next-generation data center footprints, to go 100% flash or to use a mix of hard drives and flash? Any thoughts would be helpful. Thank you.
Yes, thank you. Let me start by saying you're thinking about it the right way. We're effectively licensing both parts of the technology, the software and the hardware design. The customer will buy directly from their integrator, including the hardware. So that's correct. For us, the revenue comes largely as licensing fees, software fees and support fees. In terms of how the revenue will roll out, as we've indicated, more of it is expected in our FY2027. Next year will be the ongoing development by this customer of their next-generation data center design, and we are intimately involved in that. Right now, we had to prove ourselves by demonstrating price-performance at the level required for the majority of their storage, which for all the hyperscalers has been hard disk, but we expect this will be a consistent architecture they use across the vast majority of their storage, including their higher-performance storage, which is generally provided by SSDs today.
Thank you, Aaron. Next question please.
Your next question comes from the line of Howard Ma of Guggenheim Securities. Your line is open.
Great. Thanks for taking the question and congratulations on the hyperscaler design win. I believe it's the game changer for Pure and a potential key change in the industry. Charlie and Kevan, I want to ask a little more about the size and structure of this deal. First, could you share what the deal length is? And on the revenue opportunity, if I do some back-of-the-envelope math — double-digit exabytes in calendar 2026, and if you assume $0.10 per gigabyte and maybe half of that is recognized since it's software only, you get to about $500 million in calendar 2026, which means that 2025 could be either zero or close to $500 million. Any color you can share, assuming those ranges are right for 2025, would be helpful. And if I could sneak in a follow-up on margins, could you comment if the deal will be gross margin accretive and what exactly the investments are that you're making on the OpEx lines? Thank you.
Yes. Thanks, Howard. There are a lot of questions in there. Let me start with margin and the gross margin impact. We are still at a fairly high level on how this looks and will provide more details as we go. You are thinking about it correctly: the majority of the deal's value will still be hardware, and that is important to consider for your models. For gross margin, using company gross margin as we have been tracking it is a good way to think about it right now. We will give updates as we move forward. As for duration, it is multiyear, although the volumes are still to be determined.
Great.
Okay. Thank you Howard. Next question please.
Your next question comes from the line of Pinjalim Bora of JPMorgan. Your line is open.
Great. Thanks for taking the question and congrats on the quarter. Charlie, I wanted to ask you, how does this opportunity change or evolve as you kind of go through your motion of density expansion of the DFM modules. Is this currently based on the 75 terabytes or 150. As you move towards the 300 terabyte, how do you expect the volumes kind of evolve in this particular case?
Yes. It doesn't specifically depend on any one factor, although they know our roadmap, which was a major reason they chose us. They recognize we tend to lead on density. A large portion of storage for most hyperscalers, including this customer, is focused on low cost per gigabyte, so maintaining cost leadership is very important. At the same time, they're impressed that the same technology can be used for higher-performance workloads by using lower-density DFMs. The real benefit is that it's essentially one technology across nearly every price-performance tier, and for a hyperscaler, having a single technology that meets the various requirements to run the operation is a significant advantage.
Yes. And I'll just add one thing, Pinjalim. As Charlie mentioned earlier in the prepared remarks, we've been working with hyperscalers to drive this technology into disk replacement for some time. I would say that it was really the introduction of the 75 terabyte modules that really got this particular customer to really pay attention. And as we look forward, right, as we look at the 150-terabyte generation we're shipping now and our road map ahead, each step along that path just makes a solution that much more compelling from really on the basis of all the superlatives that Charlie outlined earlier, including costs.
Thank you, Pinjalim. Next question please.
Your next question comes from the line of Mike Cikos of Needham & Company. Your line is open.
Hi, team. This is Matt Litchi on for Mike Cikos over at Needham. Thanks for taking our question. I wanted to clarify, is the commentary around Evergreen subscription TCV and revenue, a reiteration of previous expectations? And is there any directional color you can provide on how you're expecting subscription revenue to trend as a percentage of the overall mix?
Yes, it's a great question, Matt. And yes, that's correct. It's really a shift that we saw this quarter, in particular, with our Evergreen//One higher velocity opportunities where we saw those opportunities converting to a traditional sale, meaning CapEx sale. And so, what that meant is, obviously, we saw a higher product revenue, and that's really what drove our guide increase for the year.
Yes. Matt, let me expand on this a little bit. I think what we're seeing in the market is that there's been increased pressure on customer OpEx budgets due to the increase in software costs that unexpected increase that came in software and SaaS cost this year as well as AI uncertainty. And so we're seeing more customers than we had seen in the past who had been considering a Storage-as-a-Service offering, which obviously comes out of an OpEx budget, moving them to a standard product purchase, which comes out of the CapEx budget, which has not been under such pressure this year. This market is still evolving. It's still in its early phases. We think this is a temporary phenomenon. We would expect just because of the superior economics and the superior capabilities that we bring in a Storage-as-a-Service offering that we'll start to see growth again, but I think this has been a temporary phenomenon due to the unexpected high OpEx expenditures by organizations this past year.
Yes. And I think that's actually really important as well, right, that when we think about selling data storage technology as a service, it's still early in its maturity. And with it, there's a tendency to default to what's easier, and that's, frankly, purchasing technology through a CapEx sale or traditional sale.
Thank you, Matt. Next question please.
Your next question comes from the line of Simon Leopold of Raymond James. Your line is open.
Thank you. Wanted to check on this. The CoreWeave announcement that you released last month. So obviously, it sounds like that's not the big hyperscaler opportunity. So maybe you could help us out in terms of describing the nature of that particular win and some of the color behind the use case and that opportunity. Thank you.
Yes, Simon. This is Rob. I'll take that. We're very excited to expand our partnership with CoreWeave, through both the strategic investment we made and the commercial partnership we announced, which will make Pure Storage technology available to all of their customers in the CoreWeave-dedicated cloud. As we mentioned in the release, we've been working with CoreWeave, the leading GPU cloud provider, for over the last year to stand up one of the largest-scale GPU storage deployments for a large joint customer. Building on the success of that environment, this latest partnership takes those learnings and makes that proven design more broadly available to our joint customers. We're really excited to have announced it and are already seeing early signs of interest and demand from the customer base.
And Simon, just so there's no confusion, we had announced a double-digit GPU cloud win about two quarters ago, I believe it was Q4, three quarters ago. That was CoreWeave. We were not able at that time to use their name, but that was the deal and that infrastructure is in place now.
And Simon, one last thing. Yes, CoreWeave is not the top four hyperscaler that we have been discussing today.
Thank you, Simon. Next question please.
Your next question comes from the line of Krish Sankar of TD Cowen. Your line is open.
Yes. Hi, thanks for taking the question. And Charlie and Kevan, congrats on the results on this exciting hyperscaler win. Charlie, you talked about double-digit exabytes in calendar 2026. Is this based on the agreement in place? Or is that your estimate for how big this opportunity can scale? And also on the license, is it a onetime license? Or is it annually recurring and whether it's tied to exabytes or not? Thank you.
Those are our expectations for calendar 2026 based on coordination with the customer, but most of the sales are still to be determined. The design win means we are the plan of record; to be clear, they are designing around us and we are the plan of record for the storage in their next-generation design. On the licensing, go ahead, Kevan.
Yes, the licensing for the direct flash technology and Purity software will really be tied to the capacity that we'll be shipping.
Thank you, Krish. Next question please.
Your next question comes from the line of Mehdi Hosseini of Susquehanna. Your line is open.
Yes. Charlie, in order to understand your reference to FY 2027 and the hyperscaler win that would help you with double-digit exabyte shipments. Can you help us understand what is the estimated exabyte shipment in the current fiscal year, fiscal year 2025?
No substantial shipments other than test systems this year.
Thank you, Mehdi. Next question please.
Your next question comes from the line of Param Singh of Oppenheimer. Your line is open.
Yes. Thank you for taking my question. So I really want to dive into your Evergreen//One opportunity. I know you mentioned that some of the more higher velocity sales go back to on-prem direct sales. But now that you have more hyperscale opportunities coming in addition to the one you announced today. Do you think customers might be more likely to move directly to a hyperscaler with your Purity platform and with DCF versus going for an Evergreen//One type of sale. Thank you.
Yes, I'll take that. It's an interesting question, but I think the answer is mostly no. The infrastructure we'll be selling to this hyperscaler, and potentially to other similar hyperscalers, is the underlying storage technology, not the storage services those customers offer to their users. Those services are based on their own software and are essentially the same software services they offer today. Enterprises generally move to the cloud for application deployment reasons rather than storage decisions, so those choices are largely independent of where they place their data. When customers decide to keep their applications on-prem, that's where we typically sell our standard product today. For those who keep apps on-prem, we offer the alternative of storage as a service that operates both on-prem and in the cloud. I view those as two very separate things, and I don't think our offering of storage for hyperscalers using our direct flash technology will dramatically change enterprise decisions about where to put their storage.
On the other hand, we can bring some of the best attributes of cloud storage to the enterprise. We announced that Pure Fusion is available for download this quarter, which lets companies build their own enterprise data cloud combining on-prem storage with storage they run on our software in the cloud. That enables them to operate more like hyperscalers, treating storage as a service and a cloud of data rather than as individual arrays. I see these efforts as complementary, but I don't believe our sales to hyperscalers for foundational storage will significantly influence enterprise decisions about where to place their applications.
Thank you, Param. Next question please.
Your next question comes from the line of Asiya Merchant of Citi. Your line is open.
Great. Thank you for taking my question and congrats as well on the hyperscaler announcement. The evergreen TCV, it appears that most of the customers are transitioning more to a product sale. And you mentioned this was probably a temporary phenomenon given pressures on OpEx budgets. As you look at the pipeline of opportunity ahead of you and as you look into calendar 2025, if you can help us understand what your expectations are for this Evergreen offering and Evergreen services subscription services to ramp up again to double-digit growth. Thank you.
Yes. I appreciate the question. A couple of thoughts: when we look at the dynamic of our Evergreen One conversions over $5 million, there is really no change from what we said last quarter. That was consistent with our expectations, which we walked through with our investors last quarter. The dynamic has shifted a bit on the higher-velocity Evergreen One opportunities of $5 million or less. That's where we saw a meaningful uptick in conversions to traditional sales. We believe that is not sustainable, especially given our industry-leading Evergreen One service offering. Regarding guidance for next year, I want to see how Q4 plays out and will provide an update next quarter.
Thank you, Asiya. Next question please.
James Fish of Piper Sandler. Your line is open.
Hi, guys. Just I'll echo my congrats on the cloud one as well. I did want to ask around pipeline. First on the hyperscaler cloud opportunity. You guys keep alluding the potential to get other hyperscalers based on your comments. So what are you seeing with this similar licensing potential to other clouds in terms of conversations you have going on with let's say, the top 10 cloud and AI providers. And second, maybe on the enterprise side, is the calendar 2025 setting up to be a larger refresh year for storage? Or does that evergreen model kind of mute some of the impact. Thanks, guys.
Yes, absolutely. So as we said, our discussions with multiple hyperscalers continue to progress. And I would say they progressed because of the attributes which Charlie highlighted earlier, are the reliability that's unmatched that we can deliver, the significant power savings. Certainly, the performance envelope we can deliver all driving TCO savings. I think those are the main drivers, more so than the specifics of the financial and commercial arrangement. But those discussions continue with multiple additional prospects. I do think that today's announcement with both our lead customer in the space as well as our strategic collaboration with Kioxia to go and enable a broader penetration of our flash technology into the hyperscalers will serve to accelerate that. So we're very excited about the potential ahead of us. At the same time, as we mentioned, the design win is a very, very important milestone that puts us in the plan of record with this hyperscaler, but there's clearly more work to be done. Right? And so, we'll be working very, very closely with this hyperscaler over the next year as they continue to progress through typical move to production stages, as they ramp into ultimately what we're expecting to be double-digit exabyte capacities in the fiscal 2027.
Regarding expectations for enterprise, I think your question is whether we believe demand is being pent up because of a muted IT market. I do tend to think there is some pent-up demand building, but we'll have to see what that looks like as we move into next year.
Thank you, Fish. Next question please.
Your next question comes from the line of Wamsi Mohan of Bank of America. Your line is open.
Yes. Thank you. Going back to the hyperscaler opportunity, Kevan, your comments around the revenue contribution in relation to data storage capacity purchases would be significantly lower. Is that just a lack of hardware as part of the mix? Or should we think that the pricing of the software itself is also somewhat lower on a per exabyte basis. And just to clarify, is the double-digit exabytes in calendar 2026, is that from this one hyperscaler? Or does that bake in additional wins in the future? Thank you.
Yes. Thanks, Wamsi. The double-digit exabytes in FY 2027 relate to this hyperscaler design win itself and our expectations associated with that win. And what was the first part of your question?
You're saying the revenue contribution is relatively lower compared with a storage capacity sale. Is that because the solution doesn't include hardware, or because the software is priced differently than in your traditional sales?
Yes. Thanks for the clarification. And really is going to be the primary driver of this will be the fact that hardware is excluded from our sale, and that would be really the primary driver. Now when you also think about scale in order of magnitude, pricing will play into it, but it's secondary to the fact that hardware is excluded.
Thank you, Wamsi. Next question please.
Next question comes from the line of Meta Marshall of Morgan Stanley. Your line is open.
Great. Thanks. I just wanted to ask about the volume being TBD while having a sense of what the architecture will look like. Is the TBD because you don't know how fast they will roll out this new architecture, but you do know roughly what your percentage of that data center will be? Or what are the remaining moving pieces as you narrow in on what the true volumes will be in fiscal 2027? Thanks.
Yes. Thank you, Meta. Largely, the exact schedule of the data center build-outs is a bit unknown. I would say the exact percentage we have of the data center is also somewhat uncertain, although we know that we have qualified for a very large portion of it and we're aiming for even more. So, for the most part, it's about the schedule, which is largely out of our control. They still have considerable design work on their side for other components, including ours, and for many other elements of their next-generation design.
Thank you, Meta. Next question please.
Your next question comes from the line of Eric Martinuzzi of Lake Street Capital. Your line is open.
Yes, I wanted to follow up on the change in the Kioxia relationship, given your comment about the hardware being excluded. I would think the hyperscaler would then be working directly with Kioxia. What's changed about that relationship? Because you talk about, hey, more manufacturing capacity and more cutting-edge technology, it would seem like you guys are being disintermediated.
Yes. Thanks for the question. No, we're not being disintermediated. It is still our relationship with all of the NAND vendors. We're responsible for all the qualification. We're responsible for the design. We will also be responsible for the design of the direct flash modules themselves. So we're very much in control the hyperscaler uses an integrator. It will be the integrator that purchases the hardware modules and provides it to the hyperscaler. Think of it as a meat in the channel model is probably the best way to think about it.
I would also say that this is an expansion of a long-standing relationship that goes well beyond just commercial and capacity. We've had a great collaboration with Kioxia in terms of technology road maps and technology design. That's a key part of this collaboration as well: bringing co-designed direct flash technology, as Charlie mentioned, using their bleeding-edge NAND parts to this hyperscaler customer.
Thank you, Eric. Next question please.
Your next question comes from the line of Tim Long of Barclays. Your line is open.
Thank you. Just wanted to ask on the gross margin front. I think the comment was the sequential decline in Q4 should be similar to Q3. Just if you could just clarify that. And then looking out on this line, it seems like the mix might steadily be shifting to //E and FlashArray//C Series. Is that something that we would expect on the product and the hardware side to continue? And if you can make some comments on competition now that more of the competitive set has QLC-based products in their portfolio. Is that piece of the market getting more competitive? Thank you.
Maybe, Tim, I'll hit the product gross margin and the sequential decline similar to what we saw in Q2 to Q3 and what we're expecting now to Q4 and that really is driven through our strategy of enabling our customers to transition their cost-sensitive workloads to our //E family and FlashArray//C solutions. And it's really all about the strong growth we're seeing across these solutions. Now there is less price elasticity for these solutions given the price sensitivity of the workloads. So the combination of the increased growth of these solutions and the reduced price elasticity is putting pressure on the product gross margins that you're seeing.
Thank you, Tim. Next question please.
Your next question comes from the line of David Vogt of UBS. Your line is open.
Great. Thanks guys for squeezing me in. So maybe a question, Kevin, for you on margins next year. You talked about investing in the business to support the rollout in calendar year 2026. Also mentioned gross margins would be relatively comparable in your fiscal 2026 versus fiscal 2025. Can you kind of help us understand sort of where the investment dollars are going because you've been aggressively investing for the last couple of years. Should we expect the same level of growth in OpEx? Is that kind of the framework? If I think about 2024 and into 2025, should it look similarly in 2026? Thank you.
Yes. Great question. And I'll spend some more time next quarter going into more details of our expectations. But what I did communicate is the fact that we believe our operating margin of 17%, and that's consistent with what we communicated for this year will apply for next year as well. And when we think about the main drivers of these investments, I mean, it includes focusing acceleration of our density road map with our flat DirectFlash technology and expanding our supply chain capabilities, qualifying additional NAND suppliers and manufacturing sites and really integration of our technology with the hyperscaler hardware specifications. And that's where we think the significant incremental investments will come from. And then when we think about it for this year, obviously, we're investing heavily as well from a CapEx perspective. And that's why for the full year of FY 2025, we're thinking our free cash flow margin will be slightly below 1 to 2 points below our operating margin.
Thank you, David. We're going to take our last question from Mehdi who got back in line for his second question. Thank you very much, Mehdi, for getting back into the queue. So this will be our last question.
Your last question comes from the line of Mehdi Hosseini of Susquehanna. Your line is open.
Yes. Charlie, what is the estimated exabyte shipment in FY 2025?
It's a little bit hard to predict. It's going to be very low. It's a little bit hard to say because, as I said, it's mostly test environments that we're going into preproduction environments. So maybe one, maybe a couple, so something along those lines?
Thank you, Mehdi. Before we conclude, Charlie, I think you had some final comments.
Yes. I want to thank you all for joining us on today's earnings call. The design win underscores, I think, the critical role that Pure is going to be playing in addressing the high data growth and energy demands of hyperscalers. They're facing very fierce competition for power and even turning to very unique new power sources. But these power sources are going to take a long time to come online. With power's constraints continuing to be a huge issue for them, this is 1 of the best ways for them to free up dramatic amounts of power on the order of 20% of all the power that they use today and make it available for all of the growth that they have in front of them, while improving the overall performance and capabilities of their existing storage. I want to thank once again our customers, our employees, our partners, our investors and our suppliers. We deeply appreciate all of their continued support and commitment. We look forward to speaking to you again next year. Happy holidays, everyone. End of Q&A
That concludes the Pure Storage Third Quarter Fiscal 2025 Financial Results Conference Call. Thank you for your participation. You may now disconnect your line.