RXT Q3 2026 earnings call transcript, November 10, 2026

our most recent annual report on Form 10-K and subsequent quarterly reports on Form 10-Q, including the most recent quarterly report for the period ended September 30, 2026, filed on November 9, 2026. Rackspace assumes no obligation to update the information presented on the call, except as required by law. Our presentation includes certain non-GAAP financial measures, including adjusted EBITDA, non-GAAP operating profits, non-GAAP loss per share, and free cash flow. We believe these adjustments 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'll now turn the call over to Gijen for an update on the business.

Good morning, everyone, and thank you for joining us. I will begin my remarks with a review of our third quarter and our outlook as we add new layers to our public and private cloud offerings and add new partners, customers, and channels to broaden and deepen our enterprise AI capabilities. Our large existing operating footprint is a meaningful advantage in an environment challenged by access to facilities with power and cooling, and our customer and prospect base is engaging us earlier in their modernization planning as they see what we can provide. I will also highlight several new customer wins and how we are putting the business in place to drive higher margin and EBITDA as we move from commitments to deployments. Let me start with the quarter and our outlook. For the third quarter, total company revenue was $609 million, a 9% decrease versus last year. Adjusted EBITDA was $75 million, up 8.5% versus last year. We are maintaining our full year profitability outlook with non-GAAP operating profit guidance of $125 million to $135 million and adjusted EBITDA guidance of $285 million to $295 million. For the full year, our forecast at the midpoints is for operating profit to be up 3% and adjusted EBITDA to be up 5%. We are also raising our full year revenue outlook and tightening the expected range. We now expect revenue of $2.51 billion to $2.56 billion for the full year 2026, raising the midpoint by $35 million.

to $2.535 billion. At the midpoint, revenue is expected to be down 6% year over year. Current consumption is trending higher in public cloud, supporting an upward revision to our full year revenue outlook. Our private cloud business remains stable. We are modestly adjusting our revenue forecast in private cloud to reflect the timing of a customer deployment shifting from Q4 in to the first half of 2027, as well as capacity reservations for new customer opportunities as we expect to accelerate deployments under our AMD agreement. Importantly, we have made meaningful progress in contracting for both AMD and Nvidia compute under long term agreements. Pending finalization of financing and other deployment conditions, we now expect to be operating approximately 20. to 22 megawatts of cumulative high performance compute for AI applications by the end of 2027, inclusive of the Together AI deployment announced earlier today. Once fully deployed, we expect these 20 to 22 megawatts of cumulative capacity to generate annualized run rate revenues between 420 million and 460 million with EBITDA. We have a total of 60% plus, assuming deployment and utilization consistent with our current plans and expectations. As I enter my second year as CEO, we have moved from establishing a strategy and creating a framework to execute it to now beginning to translate our go-to-market efforts into commitments and deployments. Our goal is to establish Rackspace as the operator of the full and

Rackspace. Rackspace AI stack with repeatable solutions for regulated environments bolstered by domain expertise built over years of experience operating across industry sectors and partnering with leading global systems integrators and software solution providers. For us, that means bringing together the infrastructure, compute, data, and AI platforms and remaining accountable for the engineering and operating services we deliver. Our customers retain control of their data, models, technology choices, and where their AI runs. We have been very deliberate about how we participate in this market. We are not building a business around speculative GPU capacity or pursuing growth simply for the sake of adding megawatts. We believe our existing facilities gives us an advantage in time to market capital efficiency and execution risk compared with developing entirely new infrastructure. Existing infrastructure does not eliminate the need for investment, but it gives us an opportunity to direct that investment towards customer deployments rather than recreate every element of a new data center or requiring customers to move everything onto a single platform. Discipline remains central to our approach. We intend to weigh investment decisions against customer commitments, delivery readiness, and financing, and evaluate each deployment on its expected cash returns. Strengthening our balance sheet remains a priority. The opportunity in AI does not change that responsibility. It makes disciplined execution more important. Everything builds from our existing public and

private cloud businesses with decades of experience operating mission-critical enterprise environments. The customers asking us to deploy inference capacity are, in most cases, customers whose applications, data, and infrastructure we already operate. The AMD GPUs and CPUs are being deployed into our existing data centers that we operate, which have existing power and cooling infrastructure. The forward deployed engineers deploying Palantir are working inside many of these enterprise environments we already manage. This means we are extending an enterprise operating business, not starting a separate one. And new AI platforms are interested in Rackspace for this very reason. Customers are coming to us earlier and higher in the stack for modernization, data platforms, and more. And AI operating models before the managed services work begins. This morning, we announced a definitive agreement with Together AI for NVIDIA's GB300 infrastructure, with an anticipated total contract value of $380 million over several years, subject to certain conditions. Under the contract, Rackspace will provide and operate the dedicated infrastructure, power, liquid cooling, networking, and storage, with the ability to add security, governance, and compliance controls for N customers in regulated industries. We currently expect this to be in service by mid-2027. Separately, we also entered into a letter of intent with Together AI. It covers two things. Potential future capacity expansion, which we believe could represent a potential

multi-billion dollar long-term opportunity, and a proposed joint go-to-market for enterprise AI solutions, with Together AI providing inference and rack space, providing the infrastructure, platforms, and forward deployed engineering. The LOI is not a definitive agreement and is subject to future agreements and other conditions, but we are excited about the opportunity to partner with Together AI and grow our relationship. We expect to share more details as this LOI evolves or converts into definitive agreements. I am pleased with our progress under the initial 30 megawatt Strategic AI Infrastructure Agreement powered by AMD Silicon. We continue to anticipate quarterly deployments of AMD capacity in steady quarterly tranches through 2028 and are pleased with our demand pipeline. We will begin 2027 with roughly 2 megawatts deployed. We are currently seeing potential opportunities to accelerate AMD deployments, including up to 5 megawatts in Q1, to reach approximately 7.5 megawatts cumulative by March. After that, we currently anticipate adding about 3 megawatts per quarter, reaching close to 11 megawatts at Q2, 14 megawatts at Q2, 3, and approximately 17 megawatts by year end. Note that these amounts refer to AMD deployments and exclude the Together AI deployment on NVIDIA GPUs announced earlier today. This forecast is approximate, and actual timing could differ due to a variety of factors, including availability of financing, vendor delivery timing, and data center readiness.

We are currently dealing with deploying environments for a previously announced global systems integrator, and expect to begin recognizing initial revenue towards the end of the fourth quarter. Today, we also announced a definitive agreement with an anticipated total contract value of $230 million with Blacksmith, a fast-growing AI-powered developer platform focused on accelerating software delivery and improving code reliability. Under the agreement, Rackspace will provide dedicated bare metal compute powered by AMD EPYC CPUs and storage with deployment expected to begin in early 2027. I want to highlight that for more than two decades, Rackspace has worked alongside global systems integrators to build and operate customized enterprise solutions. This experience gives us a distinct advantage as GSIs look to bring enterprise AI solutions to their customers. Our partners bring industry expertise and customer relationships, while Rackspace provides the infrastructure, engineering, and ongoing operations required to bring those solutions into production and operate them at scale. We see our growing GSI pipeline as an important accelerator of our enterprise AI strategy, complementing our direct sales and strategic alliance channels. at all We also want to update you on our Palantir partnership, where we continue to progress from contracts to deployment and production. Since our last call, we have signed seven contracts across industries with five in active deployment and two in production. These engagements are addressing meaningful business challenges. For example, we are helping a regional US healthcare system use AI powered workforce analytics to manage overtime.

One of them. highest controllable costs. The solution is designed to replace manual spreadsheet driven analysis with real time insights to help leadership identify cost drivers and evaluate operational changes. We are also working with a global specialty chemical company to modernize its finance and reporting processes through AI. The customer engaged Rackspace for our forward deployed engineering expertise. Beginning with a focused boot camp to identify opportunities and establish a path towards production. Across industries, we are developing a repeatable approach combining technology, engineering expertise, and ongoing operations to move AI from proof of value into production. Supporting this progress, we have more than doubled our Palantir certifications since July. Now exceeding 800 across the organization. Strengthening our technical expertise and ability to deploy and operate AI solutions for our customers. As part of our growing partnership, Palantir is also now a Rackspace shareholder, owning approximately 2% of Rackspace's common stock. Our pod strategy is designed to make enterprise AI easier to deploy and operate at scale. These are repeatable, pre-engineered environments that bring together accelerated AI platforms, governance, and continuous operations, reducing the need to build each solution from scratch. At the foundation of this strategy is our enterprise sovereign pod, supported by NVIDIA, Dell, and Palantir architecture, which combines dedicated AI infrastructure, data and model governance, operational controls, and forward deployed engineering expertise to help enterprises deploy and operate AI.

in production while maintaining authority over their data, technology choices, and AI environments. Building on this foundation, our healthcare pod adds security and governance controls designed to support customers' compliance obligations, protect sensitive healthcare data, and integrate with existing healthcare systems. The pods also enable our integrator partners to bring Rackspace capabilities to a broader enterprise customer base, and we expect to develop additional industry-specific pods for regulated sectors with our partners over the next year. The recent elections have brought new greenfield data center opposition to the fore of the markets. In addition, the markets have begun to ask questions about the timing risk and the capital required to build gigawatt-scale data centers and its overall impact. We believe this makes Rackspace's existing data center capacity even more meaningful. Our existing data center provides capacity that we believe can be deployed efficiently without reliance on greenfield projects. Keeping these risks in mind as we consider future deployment opportunities, we expect to evaluate each potential future project against four key criteria. First, infrastructure readiness. Can we leverage existing capacity, power, and cooling to accelerate deployment? Second, customer commitment. Do we have a committed customer with the financial strength to support the investment? Third, capital structure. Can we structure the investment to optimize equipment financing, customer funding, and Rackspace capital? Fourth, what risks does Rackspace retain? Construction

permitting, equipment, customer credit? And is that risk acceptable? Over 25 years of infrastructure deployment and operating experience position us to capitalize on these opportunities while maintaining the discipline necessary to deliver sustainable profitable growth. To help support our growth initiatives, Rackspace recently entered into share purchase agreements with an investor group led by Voss Capital LP providing for an aggregate 66 million investment in Rackspace common stock. We expect to use the net proceeds for general corporate purposes, working capital, capital expenditures, and growth capital, including to fund GPU related growth initiatives related to today's Together AI announcement. We appreciate the continued support of VOS and the additional investors as we execute on our strategic plan. Note that the hardware for the Together AI deployments, including NVIDIA, GB300, NVL72 systems, is expected to be financed separately, and we are in active discussions with potential lenders. In summary, the strategy we previously articulated has begun to progress from commercial discussions in Rackspace common stock. In addition, we believe our existing infrastructure gives us a competitive advantage, and our enterprise relationships, engineering capabilities, and operating accountability give us the opportunity to create value beyond the infrastructure. Ultimately, our success will not be measured by how much capacity we announce.

We believe it will be measured by how effectively we convert that capacity into contracted deployments, execute and deliver to those commitments, and generate recurring revenue and sustainable return. With that, I'll turn it over to Mark to provide additional detail on our third quarter financial results as well as our near-term outlook.

In the third quarter, total company GAAP revenue was $609 million, down 9% year-over-year. The decline largely reflects lower public cloud revenue, partially offset by growth in private cloud. I'll walk through the drivers in our segment results. Non-GAAP gross margin was 19.5% of GAAP revenue, down 40 basis points year-over-year. Non-GAAP operating profit was $36 million, up 12% year-over-year. Cost actions and other specific initiatives taken early this year contributed to growth in non-GAAP operating profit this quarter, as payroll savings from workforce reductions, lower marketing spend, and lower bad debt expense more than offset lower gross profit. Adjusted EBITDA was $75 million, up 8.5% year-over-year. Non-GAAP loss per share was $0.06 compared to $0.05 in the prior year quarter. Higher interest expense more than offset the increase in operating profit. Cash flow from operations was positive $36 million, and free cash flow was positive $1 million. On a trailing 12-month basis, cash flow from operations was $69 million. Capital expenditures were $104 million compared to $34 million in the prior year quarter. The increase reflects private cloud customer equipment and software, including capitalized work in process on a Palantir license for internal use. It also includes preliminary hardware and equipment purchases. In support of our AMD initiatives, which we expect to ramp in the fourth quarter. Cash CapEx was $35 million. We ended the quarter with $108 million in cash and cash equivalents and $170 million in total liquidity, including the undrawn portion of our revolving credit facility. Total liquidity declined $33 million from the second quarter. The decline reflects a $24 million reduction in our receivable securitization facility. Interest in debt service.

and higher cash capex. During the quarter, we raised $12 million of net proceeds under our ATM program. Turning to segment results. Private cloud gap revenue was $260 million, up 4.2% year over year on a reported and constant currency basis. The growth came from our healthcare vertical, driven by upfront implementation revenue from the advent health migration we discussed last quarter. Customer retention also improved. with churn lower both sequentially and year over year. Non-GAAP gross margin was 31.5%, down 660 basis points year over year, primarily on the mixed impact from implementation revenue, along with increased customer license and data center costs. Sequentially, non-GAAP gross margin declined 80 basis points from 32.3%, as cost of revenue held flat on slightly lower revenue. non-GAAP... ...segment operating margin was 20.5%, down approximately 640 basis points year-over-year, and approximately 130 basis points sequentially. In public cloud, GAAP revenue was $349 million, down 17.2% year-over-year, on a reported and constant currency basis. The decline reflects the non-renewal of a large, low-margin state government contract, lower infrastructure revenue as certain larger resale accounts moved to... contracts with hyperscalers, and the start of a wind-down of certain legacy service lines. Non-GAAP gross margin was 10.5%, up approximately 130 basis points year-over-year, trending favorable due to cost actions and tightened focus on more attractive public cloud opportunities. Non-GAAP segment operating margin was 7.8%, up approximately 450 basis points year-over-year, on improved operating expense efficiency.

Including lower payroll costs and favorable bad debt. Turning to guidance. We are reaffirming our profitability outlook. We expect total non-GAAP operating profit of $125 to $135 million, up 3% at the midpoint. Adjusted EBITDA is expected to be $285 million to $295 million, up 5% at the midpoint, with non-GAAP loss per share of $0.25 to $0.30. We are also narrowing and raising our full-year GAAP revenue outlook to $2.51 billion to $2.56 billion, compared to $2.45 billion to $2.55 billion previously. At the midpoint, this is a decline of 6% year-over-year, an improvement over the 7% decline previously anticipated. From a segment perspective, we now expect private cloud revenue of $995 million to $1.015 billion, up 2% year-over-year. At the midpoint, compared to up 4% previously. The change reflects the timing of a large healthcare implementation, shifting into the first half of 2027, and reserving certain data center capacity for enterprise AI compute deployments, which we now expect to accelerate. We expect public cloud revenue of $1.515 billion to $1.545 billion, which represents a decline of 10% year-over-year at the midpoint, an improvement over the 13th. and decline previously anticipated. The change reflects stronger renewals and higher consumption from customers. Our non-GAAP tax rate is expected to be 26% and non-GAAP other expense is expected to be in the $220 million to $230 million range. Non-GAAP weighted average shares is expected to be between $265 million and $275 million shares, excluding any potential dilution from the ATM program.

I'll now turn it back over to Gijen for final remarks.

We are maintaining our full-year profitability outlook while moving our enterprise AI strategy into signed customer commitments and production deployments. We are entering the next phase of Rackspace with signed contracts with Together AI and Blacksmith, representing approximately $610 million in anticipated aggregate total contract value. Subject to successful deployment and other contractual conditions. Our combined AMD and NVIDIA deployment plans now target a cumulative total of 20 to 22 megawatts of AI compute by the end of 2027. We also have Palantir solutions already in production. Our advantage is the ability to bring the full enterprise AI stack together for customers building on trusted relationships, our existing data center footprint, and 25 years of operating critical enterprise technology. Our vendor neutral approach gives customers technology choice backed by the engineering and operating expertise to put AI to work. As we bring additional AI capacity into service during 2027, we see meaningful opportunities for incremental revenue and earnings, subject to financing, deployment timing, and execution. Our focus now is relentless execution, bringing these deployments into service, strengthening the economics of our core cloud businesses, and delivering measurable value for customers. That is how we intend to build a stronger, more profitable rack space.

Thank you to our customers.

Thank you, sir.

Yeah. In terms of our overall guidance changes, right, we basically raised our overall revenue guide in total with public cloud moving a bit higher, private cloud moving a bit softer. Total adjusted EBITDA stayed in the same range, 285 to 295 with the midpoint at 290. In EPS, there was some. But it stayed within the same, the same guided range that we had before.

Oh, great. That's helpful. And then you had, Shana, briefly in a prepared remarks, the elections, maybe some, maybe there's like a market narrative that building space is going to be a little bit tougher, possibly going forward, but you already have installed capacity. I guess. Okay. How does that translate into higher demand, maybe some pricing power, or just general, like, tailwinds for the business, just with the political environment, if it does end up changing over the next 12 months, let's call it. Thanks.

Thank you. And that concludes our Q&A. session for today. At this time, I'd like to thank everyone for attending. You may all disconnect at this time.