Prepared remarks
Good day, and thank you for standing by. Welcome to the Rackspace First Quarter 2026 Earnings Webcast. Operator instructions. Please be advised that today's conference is being recorded. I'd now like to hand the conference over to Sagar Hebbar, Head of Investor Relations. Please go ahead.
Thank you, and welcome to Rackspace Technologies First Quarter 2026 Earnings Conference Call. I'm Sagar Hebbar, Head of Investor Relations. Joining me today are Gajen Kandiah, our Chief Executive Officer; and Mark Marino, our Chief Financial Officer. As a reminder, certain comments we make on this call will be forward-looking. These statements involve risks and uncertainties, which could cause actual results to differ. A discussion of these risks and uncertainties is included in our SEC filings. Rackspace Technology assumes no obligation to update the information presented on the call, except as required by law. In particular, our discussion today will include forward-looking statements regarding our recently announced memorandum of understanding with AMD, including statements regarding the anticipated scope, benefits, commercial potential of the collaboration, deployment timelines or financial projections, the expected execution of definitive agreements and the anticipated impact of the partnership on our business, financial results and capital structure. The MOU represents a nonbinding framework only and does not constitute a binding commitment by either party to complete any specific transaction, financing or other commercial arrangement. No definitive agreements with AMD have been reached. Discussions remain preliminary, and there can be no assurance that any such arrangements will be entered into, that the parties will reach agreement on terms or that the anticipated benefits of the collaboration will be realized. Any third-party financing required to implement the transactions contemplated by the MOU is subject to the availability of financing on acceptable terms. There can be no assurance that any such financing will be obtained. Our presentation includes certain non-GAAP financial measures and adjustments to these measures, which we believe provide useful information to our investors. In accordance with SEC rules, we have provided a reconciliation of these measures to their most directly comparable GAAP measures in the earnings press release and presentation, both of which are available on our Investor Relations website. I will now turn the call over to Gajen for an update on the business.
Thank you, Sagar. Last quarter, I said Rackspace was moving beyond being an infrastructure provider to becoming the orchestrator and operator of enterprise AI in regulated environments. We laid out three specifics: a partnership with Palantir anchored by a core build-out of forward deployed engineers, a technology stack with VMware as the control plane, Rubrik for cyber resilience, and Palantir as the data and AI platform layer, spanning infrastructure, resilience and AI and accelerating demand for Private Cloud in regulated environments. The results this quarter reinforce the strategy we've been executing against what we call where enterprise AI goes to production, governed infrastructure as the foundation, an integrated technology stack of curated partners on top of it and one accountable operator running it end-to-end. Every win this quarter sits inside that frame. We secured regulated and sovereign Private Cloud deals across health care, telecoms and financial services. We also closed our first joint Palantir deal in 41 days, a U.S.-based solar tracking manufacturer where the problem was costly and quantifiable, 16.5 days to move from a customer inquiry to a signed quote, burdened by manual intake and fragmented handoffs. Our forward deployed engineers deployed AI-enabled workflows on Palantir Foundry directly inside the customer's environment, reducing the quoting cycle by 94% and earning an expanded engagement to extend the forward deployed engineer model into EMEA. We are also deploying Palantir inside Rackspace, running end-to-end business workflows on Foundry natively. We are not just recommending Palantir to customers; we are operating our own business on it. We continue to expand our partner ecosystem. Today, I'm pleased to announce the signing of a memorandum of understanding with AMD that establishes a new category of governed enterprise AI infrastructure. We are integrating AMD Instinct GPU accelerators, AMD EPYC CPUs and the ROCm software ecosystem into a fully managed governed technology stack, purpose-built for enterprise, including health care, financial services and sovereign environments where security, compliance and accountability are nonnegotiable. The MOU establishes AMD as the launch silicon across our four integrated capabilities: Enterprise AI Cloud, our fully managed private, public and sovereign AI environment with one operator accountable across the stack; Enterprise Inference Engine, a context-aware inference runtime that retains domain knowledge, session history and enterprise-specific data context across queries with Rackspace owning the SLA; Inference as a Service, dedicated accelerated compute as a governed alternative to commodity GPU rental, launching with AMD Instinct; and Bare Metal Accelerated Compute launching with AMD Instinct for training and inference workloads requiring deterministic performance. Production inference is heterogeneous. Frontier models run on GPU, small language models, classical ML embeddings and many domain-specific workloads run more efficiently on CPU. AMD is the partner that brings both Instinct GPUs and EPYC CPUs inside one integrated architecture, which lets us route each workload to the right compute. That is what production economics requires. This puts Rackspace in a unique category. The market today is dominated by commodity GPU rental, where capacity is sold by the hour and the customer carries the burden of integration, security and accountability. We are building the opposite. AMD's leadership in open high-performance AI acceleration, combined with our operator-grade Outcomes-as-a-Service model delivers governed AI infrastructure that is accountable from silicon to outcomes. We expect the definitive agreement with AMD to be executed in the near term. Governed infrastructure is where enterprise AI either succeeds or stalls. When AI works with patient records, financial data or sovereign information, where that data sits and how access is governed determines compliance or exposure. That is why Rackspace's over 25-year history managing data centers and infrastructure is more important than ever. And this is why one of the largest EPYC environments runs on Rackspace. The second reason enterprises choose us is how we handle technical complexity. Enterprise AI cloud is not a single component problem. It takes data, compute, models, small language models, inference and governance working together in real time. If even one element in the technology stack is off, cost per token skyrocket and operational risk increases. We solve this by integrating each vendor's IT, making technologies fit together and operate as one. The third reason is accountability. In a fragmented enterprise AI cloud vendor ecosystem, nobody owns the outcome or takes responsibility when something breaks down. We solve that by being one accountable partner in the eyes of the customer, responsible for how the system performs and the outcome it delivers. That is why we are seeing momentum across the business. At our core, Rackspace is a data center and infrastructure company. We own and operate the physical infrastructure that enterprise AI runs on. That foundation, combined with our ability to take end-to-end accountability for AI in production from governed Private Cloud to AI inference and agents in production is exactly what our enterprise customers are looking for. And with that, let me get into our business performance, starting with Private Cloud. First quarter Private Cloud revenue was $235 million, with first half revenue on track with the timing of a large deal onboarding within our health care vertical, consistent with the dynamics we outlined last quarter. Segment operating margin came in at 24.7%, up 30 basis points year-over-year, driven by continued cost discipline. Our customer wins this quarter tell a consistent story. Enterprises in regulated industries are choosing Rackspace to modernize and operate environments where governance, reliability and compliance are nonnegotiable and where those environments increasingly serve as the foundation for AI adoption. For example, in financial services, we secured a long-term recommitment from a leading global online trading platform, modernizing core infrastructure through software-defined Private Cloud, improving resilience and user experience in a latency-sensitive, highly regulated environment. In health care, we signed a multiyear agreement with a major U.K. NHS Foundation Trust to migrate and operate workloads in a sovereign health care cloud with full outcome-as-a-service and security embedded from the outset. And this quarter, we expanded our relationship with AdventHealth, a long-standing customer. We already host and manage the infrastructure of their Epic EHR, one of the top five Epic systems in the world. And this quarter, we expanded our relationship to host and manage over 400 additional workloads on Rackspace Private Cloud. Health care is one of our most important verticals and one of the clearest expressions of our strategy. Epic Managed Services is proprietary Rackspace IP, purpose-built for governance, performance and uptime that clinical environments demand. As regulated health care organizations move from AI experimentation to AI in production, where data sits and how it's governed becomes the defining question. That is exactly the environment we are built to operate. This extends into sovereign markets. In Saudi Arabia, our partnership with SDAIA places us inside one of the world's most advanced national AI programs, built on in-country infrastructure, jurisdictional accountability and managed operations. In the U.K., BT recently selected Rackspace as the infrastructure foundation for BT Sovereign Cloud, positioned as the U.K.'s first full suite of sovereign services hosted and operated entirely within the U.K. with security-cleared operations teams and managed services covering migration, operations and ongoing compliance. That is the kind of public anchor that validates our sovereign thesis. These are environments where AI cannot be deployed without full control over data and infrastructure, and they are increasingly central to how sovereign and enterprise AI is deployed. What makes these environments possible at scale is VMware Cloud Foundation 9, the control plane at the center of our governed AI strategy. It unifies compute, storage, networking and security into one operating substrate with native AI workload support, data residency controls and policy enforcement that meets regulated and sovereign requirements out of the box. Our deepening partnership with Broadcom around VCF 9 is one of the most strategic commitments we are making this year because it gives our customers a single control plane that travels with the workload with elasticity to Public Cloud where it makes sense. Running on top of that foundation is where our AI platform partnerships come to life. This quarter, we expanded our relationship with Uniphore, adding agent-based workflows to our governed AI technology stack. Together, we are building context-aware inference, a capability that retains domain knowledge, session history and enterprise-specific data context across queries. So AI agents and large language models perform with the consistency and institutional memory that production environments require. Like Palantir, our engineers are trained on the Uniphore platform and embedded directly inside customer environments. We are not just orchestrating infrastructure. We are orchestrating outcomes. VCF 9 as the control plane, Dell for core infrastructure, Palantir and Uniphore for governed AI and agent workflows, Rubrik for data resilience, AMD for enterprise-ready compute. Each partner is best-in-class, but the value Rackspace delivers is making them operate as one integrated system with full accountability for how the system performs and the outcomes it delivers. Looking ahead, the next phase is already emerging. As enterprise AI evolves towards agentic workflows where machines interact with machines and processes run end-to-end without human intervention, the demands of governed infrastructure become even more acute. Training will largely sit with specialized providers, but inference, particularly context-aware inference on regulated data is where production enterprise AI lives. That is the workload we have built to operate. And as customers develop a clearer picture of their data residency requirements, more of those workloads will move into governed Private Cloud, deployed across our global data center footprint in the jurisdictions and sovereignty zones our customers require. That is why we are doubling down on VCF 9 and Broadcom this year. Our full year Private Cloud growth outlook remains on track. We have signed engagements with AdventHealth, Seattle Children's and a strategic Database-as-a-Service partner onboarding through the rest of the year. We are also seeing encouraging pipeline momentum on our Palantir and Uniphore partnerships where context-aware inference and governed agent workflows are gaining traction at deal sizes that we have not historically seen. The AMD partnership announced today adds a further layer of future optionality as governed AI compute becomes more central to how regulated enterprises operate. Together, these give us confidence in the full year Private Cloud growth profile we are reaffirming today. Now for our Public Cloud update. First quarter Public Cloud revenue was $443 million. Services revenue grew 10%, reflecting our continued shift towards higher-value engagements. Our customer wins this quarter highlight the breadth of our platform capabilities and our deepening presence in the AI space. First, we are powering a large-scale enterprise-wide multi-cloud transformation for a leading health care technology organization. Through a governance model, we are delivering program managed migrations, modern architecture, intelligent automation and measurable cost optimization, ensuring each workload is placed on the right platform for the right reasons. Second, Rackspace is serving as the implementation and managed services delivery engine for a high-growth AI-native database-as-a-service partner operating across both Public and Private Cloud environments. Our execution capabilities are a direct accelerant to our partners' client acquisition and market expansion, reflecting a high-value compounding partnership driving differentiated multi-cloud Database-as-a-Service outcomes. Our service portfolio is built for where enterprise AI is headed: production, not experimentation. We are embedding engineers directly into customer environments moving from strategy to live deployment in weeks with governance and accountability built in from day one. New partnerships expand our ability to deploy context-aware inference, governed agent workflows and forward deployed engineers inside customer environments, giving enterprises a governed path from strategy to inference workloads in production. We are complementing this with purpose-built capabilities in AIOps, identity security and data resilience, addressing the operational and security demands that become nonnegotiable once AI moves into production environments. In summary, Public Cloud is executing. As inference workloads move into production, we are increasingly positioned as the partner enterprises rely on to operate, secure and optimize their cloud environments with full accountability to match. The results this quarter confirm the thesis: governed AI infrastructure as the foundation, an integrated technology stack of curated partners running on top of it, one accountable operator responsible for the outcomes. That is what today's Rackspace delivers. With that, I will turn it over to Mark for our financial results.
Thank you, Gajen. In the first quarter, total company GAAP revenue was $678 million, up 2% year-over-year, driven by solid Public Cloud performance. Non-GAAP gross profit margin was 18.3% of GAAP revenue, down 160 basis points year-over-year, reflecting the Private Cloud revenue timing dynamics we discussed. Non-GAAP operating profit was $31 million, up 20% year-over-year, driven by continued operating expense discipline. Non-GAAP loss per share was $0.06, flat year-over-year. Cash flow from operations was $5 million and free cash flow was negative $9 million. We ended the quarter with $94 million in cash and $295 million in total liquidity, inclusive of the undrawn portion of our revolving credit facility. During the quarter, we repurchased approximately $96 million of debt, reflecting our continued commitment to disciplined capital allocation and active deleveraging. This reduces our interest burden and strengthens our overall capital structure. We are making deliberate progress on leverage reduction while continuing to invest in strategic growth. Turning to our segment results. Private Cloud GAAP revenue for the first quarter was $235 million, down 6% year-over-year, reflecting the timing of large deal onboarding within our health care vertical, consistent with the dynamics we outlined last quarter. Non-GAAP gross margin was 36%, down 110 basis points year-over-year, driven by lower fixed cost absorption on reduced revenue. Non-GAAP segment operating margin was 24.7%, an improvement of 30 basis points year-over-year, reflecting continued operating expense discipline. In our Public Cloud segment, GAAP revenue was $443 million, up 7% year-over-year with services revenue growing 10% year-over-year. Non-GAAP gross margin was 8.9%, down 60 basis points year-over-year, reflecting higher infrastructure costs. Non-GAAP segment operating margin was 4.7%, up 50 basis points year-over-year, driven by improved operating expense efficiency. Now on to our guidance. We are reaffirming our full year 2026 guidance in its entirety. Revenue, EBITDA and cash flow outlook all remain unchanged. The Q1 Private Cloud timing we described is fully reflected in our annual plan and our confidence in the full year outlook is unchanged. We continue to win larger complex engagements that carry longer deployment cycles but deliver greater revenue visibility, higher lifetime value and more durable recurring revenue streams. As they come online throughout the year, we expect Private Cloud to reflect the growth profile we committed to for 2026. With that, I'll turn it back over to Gajen.
The market is trending in line with our expectations. And this quarter, we delivered proof across every layer of that thesis. Regulated enterprises are making a deliberate decision about where their AI was, who operates it and who is accountable for outcomes. Health care is now a pillar. One of the top five Epic workloads in the world runs on Rackspace governed AI infrastructure. Epic Managed Services is proprietary Rackspace IP, decades in the making and increasingly the foundation our health care customers are choosing as AI moves into production. Sovereign is validated. BT Sovereign Cloud runs on Rackspace governed AI infrastructure. SDAIA in Saudi Arabia places us inside one of the world's most advanced national AI programs. These are anchor commitments, not pilots. The technology stack is complete. And this quarter, we extended it further. VMware Cloud Foundation 9 as the control plane running across private, public, edge and sovereign environments, Palantir for governed data and AI operations with our first joint deal closing and a growing pipeline. Uniphore, enabling agent-based workflows with context-aware inference, Rubrik for data resilience and AMD, where we are establishing a new category of governed enterprise AI infrastructure, delivering four integrated capabilities from silicon to outcomes, Enterprise AI Cloud, Enterprise Inference Engine, Inference as a Service and Bare Metal AMD Instinct, one integrated system with an investment-grade counterparty co-invested in our success and Rackspace accountable for how it performs end-to-end. We are the operator of the full enterprise AI technology stack, one accountable partner where enterprise AI goes to production. That is Rackspace. Thank you to our customers, partners and every Racker. And with that, back to Sagar.
Thank you, Gajen. Let us begin the question-and-answer session. Please go ahead.
Questions and answers
Our first question comes from Kevin McVeigh with UBS. Please go ahead.
Let me start just by congratulating you folks because obviously, there's been a lot of work to be done to get you folks to this level and a lot of patience and that needs to be recognized. And I think I just wanted to highlight that because there's a lot that's going into the results that are here today. I guess — and there was an incredible amount of detail again, but maybe talk to how AMD dovetails into Palantir? And what else — it sounds like the MOU is pretty far along. What else needs to be done to get it across the goal line? It sounds like it is, but is there anything in terms of what we should look for as that officially gets signed? Or is it officially signed? It just seems like it's pretty far along, but if you could help us with that a little bit.
Kevin, thank you, and I appreciate your comments. When we look at this, I would think about Palantir and AMD somewhat distinct from each other. Starting with the Palantir relationship, that's really about deploying and running customer workflows for the customer with forward deployed engineers, somewhat independent of what compute platform it runs on. Really think about compute as what's the most efficient place to run any given workload. The AMD piece fits into how first and foremost it gives us CPU and GPU. As we move further into inference and production workloads, being able to deliver that in an efficient manner allows us to run across the CPU and GPU stack. In terms of the partnership itself, we are certainly well along the way. We still need to get the financing locked down and tightened up, but we feel pretty confident that we are on our way to getting that done and hopefully get it announced in the near future. We feel pretty good about it.
That's super helpful. And then just again, if you could remind us the capacity in the Private Cloud versus the Public. And as these initiatives kind of scale, particularly AMD and Palantir, is that primarily across the Private Cloud as opposed to the public? Or help us understand that a little bit because obviously, there's a lot to digest and just a really, really nice outcome.
Great question, Kevin. Customer workloads will run across private and public depending on where that workload needs to land, and that's why our VCF 9 partnership with Broadcom gives us a control plane across which we can elastically drive the workload, whether in Private or Public Cloud. Capacity-wise, we have partnerships on the public side, and now we have the partnership and, hopefully soon, the compute side up and running from a GPU perspective as well, which allows us to be agnostic with the customer, focus on the outcome they want and then deliver that in the most efficient way across either a CPU or a GPU landscape, and that could be private or public. That's a ton of work that goes into figuring all this out. Our customers' challenge is thinking through whether they are building a small language model or running a large language model: where do you run the inference, how do you orchestrate that, how do you ensure it's running as efficiently and securely as possible, and how is data residency handled. Our ambition is to take that complexity off the table for them and, with our forward deployed engineers, enable, support and accelerate their journey to become more AI-enabled or operate on a full AI stack. That's the opportunity we saw, and our customers are guiding us through it. We're excited about it.
No, it's amazing. One more question — it sounds like any sense of how this starts to fan in, maybe in the back half of 2026. And is there any way to think about what type of margin this work would come in at? I know it's probably a tougher question, but any way to think about that? And then what potential capital needs you could have as you're standing some of this stuff up?
Think of it this way, Kevin. There are four distinct capability sets we are bringing to market. First, governed Private Cloud on AMD silicon: think of that as we own the entire outcome for our customer in partnership with them; that would be, if you think about margin, probably our most profitable business. Second is context-aware inference, which retains domain-specific data — that's likely the next tier down in margins. Third is the inference layer, where we provide tokens or intelligence via an API. And lastly is bare metal, which is probably a lower-margin area. As we ramp, our business will fluctuate across these four areas. Our intent is to end with fully managed governed outcomes, but there is a journey to get there and we need to work through how that plays out before we can give clear guidance on margins.
Kevin, this is Mark. I would agree with that. I also think that it's going to be largely on par, if not accretive to existing gross margin rates across our Private Cloud business. And just in terms of timing, this is not something we've materially factored into our 2026 guidance, given supply chain and delivery timing.
Our next question comes from David Paige with RBC Capital Markets. Please go ahead.
Congrats on the great results here. It seems like Rackspace is moving in the right direction not only internally as a company but also where the industry is going in terms of CPU, GPU, running SLMs, LLMs, etc. I'm curious: you seem like you're the leader, but how is the competitive environment looking? And as a follow-up, you mentioned the pipeline is strong. Should we expect more deals in the future? Could you flush that out a little bit?
Good to meet you, David, and thank you. The orientation of the business is very much about helping customers understand how they want to run AI workloads. Many regulated customer workloads already run in our Private Cloud, so guiding them to run AI workloads is a natural extension. Our partnerships on the application stack — Palantir, Uniphore — and on the compute stack give us a much more integrated view to tie all of this together and deliver it. I haven't seen anyone yet that can put all of this together in one place and then own the outcome, which matters a lot in regulated or sovereign environments. For example, 'governed' in health care means HIPAA compliance, PHI security and clinical SLAs — all of that must be integrated and delivered. There will be competitors, but having consulting, forward deployed engineers, infrastructure, compute and partnerships stitched together should give us a lead and an edge.
That was very helpful. Maybe one more: there were comments about the capital structure improving. How should we think about the capital structure evolving over the next 12 to 24 months?
David, our intent is deleveraging; that's our top priority. As we think about the deals we've announced and our capital requirements for the year, the intent is to avoid taking on more expensive debt that would add to existing maturities. Our focus is increasing operating leverage, EBITDA and additional cash flow to address the 2028 maturity. We're structuring deals to avoid creating further leverage. We decreased our net leverage from 8.6 to 8.3 quarter-over-quarter and remain focused on the out years. You'll notice in the quarter we repurchased roughly $96 million notional of debt at a significant discount, so we're looking for ways to deploy capital to reduce our debt and improve refinanceability over the next 12 to 18 months.
That concludes today's question-and-answer session. I'd like to turn the call back to Sagar Hebbar for closing remarks.
Thank you, everyone, for joining us. If you have any questions, please e-mail us at ir@rackspace.com. Have a great rest of your day. Thanks, Liz.
Thank you. This concludes today's conference call. Thank you for participating. You may now disconnect.