Cisco Systems, Inc. (CSCO) Earnings Call Transcript & Summary

Hello, everyone, and thanks for joining us this afternoon. Sami Badri here, serving as Cisco's Head of Investor Relations. With me are Chuck Robbins, our Chair and CEO, and Scott Herren, our CFO. You can find Cisco's earnings press release along with supplemental information—including reconciliations between GAAP and non-GAAP—posted on our Investor Relations website. After today's call wraps up, the recorded webcast and the slides will also be posted there for you. During our time together today, we'll cover financial results on both a GAAP and a non-GAAP basis. Product results will be discussed in terms of revenue, while geographic and customer results will be discussed in terms of product orders, unless we say otherwise. Every comparison we make will be on a year-over-year basis. I want to note that today's discussion will contain forward-looking statements, among them our guidance for the third quarter and for fiscal year 2025. Such statements carry risks and uncertainties that are described in detail in our SEC filings—in particular our most recent 10-K and 10-Q reports, which lay out significant risk factors that might lead actual results to differ materially from what those forward-looking statements contain. Regarding guidance, I'd also point you to the slides and the press release accompanying this call for additional details. Cisco won't comment on its financial guidance during the quarter unless that is done via an explicit public disclosure. With that, I'll hand things over to Chuck.

Thanks, Sami — and I appreciate all of you joining us. We turned in strong results for Q2, as revenue, margins, and EPS each came in at or above the high end of the guidance ranges we'd previously given. Driving that healthy top line was strong appetite for our technology, which yielded double-digit growth across annualized recurring revenue, remaining performance obligations, and subscription revenue — and that last figure accounted for 56% of total revenue in the quarter. On a related note, AI infrastructure orders from webscalers came in above $350 million during Q2, taking our year-to-date figure to roughly $700 million, and we remain on pace to top $1 billion in AI infrastructure orders for fiscal year '25; meanwhile, we gave back $2.8 billion of value to shareholders in Q2, which brings the year-to-date amount to $6.4 billion. We also unveiled today an increase to Cisco's dividend, and our Board gave the green light for another $15 billion in share repurchases — all of which shows how dedicated we are to delivering shareholder value via steady capital returns. Let me now share a bit of perspective on the healthy demand backdrop we encountered over the quarter. New product orders climbed 29%, or 11% higher organically once Splunk is excluded, which makes it our fourth quarter in a row of accelerating order growth. That speaks to the strength of our portfolio and the value our customers get from it. Enterprise product orders rose 27%, with double-digit expansion across every geographic segment; momentum remains very strong in Service Provider & Cloud, where product orders jumped 75%, powered by triple-digit growth in webscale. Among the top six webscalers, three each posted triple-digit order growth, while two of the six each grew by more than 50%. This illustrates how much more relevant we've become to this fast-growing customer market as they build out infrastructure for AI. Inside SP & Cloud, telco customer orders increased by more than 20%, with demand expected to continue as they put money back into their core networks to get ready for AI connectivity. Public Sector orders rose 13%, with gains in every geography, as governments worldwide look to Cisco as their trusted end-to-end partner — especially as they build and launch sovereign AI clouds to digitally transform government services; now for some detail on demand from a product standpoint. Networking product orders rose double-digits, fueled by switching, enterprise routing, webscale infrastructure, and industrial networking applications within our IoT products. Campus switching orders increased double digits, and we anticipate our campus switching portfolio, along with our WiFi 7 access points, will pick up traction as return-to-office policies become more common. We're also still seeing strong order growth in data center switching, marking our fourth straight quarter of double-digit growth. We expect this trend to continue as our 800 gig Nexus switches, built on the 51.2 terabit Silicon One chip, become available in April for AI cloud buildouts. On top of that, we are starting to see AI order growth from Enterprise customers as this large market opportunity begins to take shape and enterprises seek out partners for smooth deployment and security in production settings that leverage their proprietary data. In Q2, we finalized multiple AI System deals in which customers bought our integrated systems — including Nexus, UCS and other solutions — to run the AI applications they are putting into service now. Beyond that, we recently introduced our Cisco AI POD offering, which combines Cisco's hardware alongside software, and it's beginning to gain some early momentum. This makes selling easier for partners and simplifies AI infrastructure deployment for customers. I also want to recognize how well our Industrial Internet-of-Things offerings have performed — these are our ruggedized catalyst products that we sell to manufacturing and industrial clients. Orders increased by over 40% throughout the first half of fiscal year '25, and in Q2 we saw growth above 50%, indicating an acceleration as customers prepare to roll out AI-driven robotics along with industrial security. With the U.S. shifting toward bringing strategic infrastructure back home—things like chip-making facilities, general manufacturing, and large-scale AI power generation—we hold the leading secure networking lineup in the market, one that can link and safeguard these expensive projects at scale. Security bookings again grew by more than two times during this quarter, driven by Splunk's sophisticated data handling, analytics, and threat-spotting strengths, along with appetite for major offerings including our updated firewall; within just a year, Cisco Secure Access and XDR combined have brought in upwards of 1,000 new customers and about 1 million enterprise users for each. Similarly, Hypershield is building encouraging traction even before it hits full-scale output. In the second quarter, we secured significant platform agreements with two enterprise clients from the Fortune 100, who are utilizing Hypershield to introduce security into the network in a radically different manner. Splunk led a third platform arrangement, which incorporated various Cisco offerings and services, highlighting the effectiveness of our collaborative and cross-selling initiatives. Nearly 11 months have passed since Splunk joined Cisco, and throughout that time, we've kept merging our operations and driving synergies while maintaining our pace. Additionally, during this quarter, we brought Talos into Splunk's recently launched Enterprise Security 8.0 product and incorporated AppDynamics into Splunk's on-premises log observer. I'd also like to point out a few notable innovations we introduced in this period. Also on the Splunk front, three offerings went live: Splunk on Azure, Splunk Federated Analytics, and the AI Assistant for Splunk Observability. Then last month brought a big introduction — AI Defense, our newest security innovation, built specifically so enterprises can create, launch, and protect AI applications with confidence. Through our unmatched network visibility, AI Defense guards against AI tool misuse, data leakage, and ever more sophisticated threats. For enterprise AI application protection, we deliver it at the network layer in a way only Cisco can, providing dependable and consistent safeguards. Early access customers have responded incredibly well to AI Defense, and it reaches general availability in March. We timed the launch of AI Defense to coincide with our AI Summit. At that gathering, we brought together quite a few tech visionaries, AI trailblazers, and executives representing close to 50 of the Fortune 100, so they could discuss what AI makes possible as well as what it makes difficult, and explore key topics such as policy and ethics, advances in silicon, and breakthroughs in data science. More than 60,000 people signed up to attend online. For our customers, AI Security is still a top worry, especially since fresh AI models keep coming out. The message from all this is clear: the fight to lead in AI is going to play out everywhere on Earth, and when companies compete, that fuels new ideas, faster uptake, and bigger scale — and all of that intensifies the need for and use of AI networking gear, spending on data capacity, and top-tier AI security, and every one of those works in Cisco's favor. At Cisco Live EMEA this week, we introduced our first data center smart switch featuring built-in programmable DPUs, which runs on Silicon One. By combining networking with data processing inside a single high-capacity, multi-purpose box, this smart switch streamlines data centers and raises their efficiency, enabling them to meet the performance requirements of AI. We additionally introduced Agile Services Networking, which is a fresh architecture built upon Silicon One, and it integrates optics, assurance, and automation for service providers while they revamp their infrastructure in order to launch new offerings and deliver better connectivity to their end customers. When you consider everything, Cisco's engine for innovation has never been more powerful.

To stay on that same point, I want to restate something: when we talk about what AI makes possible for Cisco, we frame it as three pillars that stand apart but also connect to one another. The first is training infrastructure for AI at webscale customers. Among the largest webscalers, five are putting together our Cisco 8K, Silicon One, optics and optical systems in the networks they use for back-end training. The second is inference for AI plus enterprise clouds. We designed our Nexus switches, Nvidia-based AI servers, AI PODs, along with Hyperfabric and AI Defense software so that deploying AI infrastructure becomes simpler and less risky, while also delivering the strength of open, hyperscale AI networking into the enterprise. And the third is connectivity for AI networks. Our technology platforms—spanning switching, routing, security, and observability—are being used by customers to modernize, secure, and automate how their networks run, getting them ready for AI applications to be deployed everywhere. Along with mature back-end models, this will drive higher capacity demands from front-end cloud networks, both private and public. We're also continuing to weave AI into numerous products, with AI Assistants now in production on our Security and Collaboration platforms, while building out Agentic capabilities throughout the portfolio. Beyond that, advanced AI is being applied throughout Customer Experience to get the most customer value and lift productivity. During Q2, we introduced the Renewals Agent, an agentic AI-powered offering built jointly with Mistral, which makes renewal proposals more efficient. We rolled out a fresh Assistant that assists clients in making Network Change Management digital and lowering its risk, and we introduced predictive analysis plus proactive technical support into our Network Security offerings to cut resolution times substantially; these mark key steps as we work toward a Customer Experience driven by agentic AI. Prior to wrapping up, I'd like to mention something we announced earlier today. Gary Steele, who serves as our President of Go-to-Market and previously held the CEO role at Splunk, is leaving Cisco when the quarter ends to take on a CEO position elsewhere. Gary's guidance played a vital part in making sure Splunk was integrated into Cisco successfully. Like we've mentioned, things have gone really smoothly with bringing Splunk on board — the top line came in right where we thought it would, and the bottom line beat what we were looking for by a wide margin. So Gary has truly delivered outstanding work. On top of that, while serving as our President of Go-to-Market, he was instrumental in reshaping how our Sales and Go-to-Market teams operate, pushing growth and making sure we're positioned around what our customers and partners need now and down the road. I'd like to express my gratitude to Gary for the partnership, the leadership, and the steady dedication he's shown ever since we first shared our plan to acquire Splunk 18 months ago. Because of that, we've kicked off a search right away, both inside and outside the company, to find someone new to lead Go-to-Market, and Gary will remain at Cisco until Q3 wraps up so the handover goes smoothly. Let me now give you the quarter in brief. Demand for Cisco's technologies is unmistakably there and has picked up speed, something our strong results make clear. We're also moving faster on the product innovation front, weaving security deeply into our networking offerings. Thanks to how well we performed, our capital allocation model was able to deliver substantial value back to shareholders. In closing, we believe this solid performance, along with the momentum we've built, leaves us well positioned as we move into the later part of fiscal '25. Scott, I'll hand it over to you to walk through the quarter and what we're expecting ahead.

Appreciate that, Chuck. It was a solid quarter for us, with revenue and EPS landing above the top end of the ranges we had guided to, along with healthy margins and operating cash flow. Total revenue for the quarter came in at $14 billion, which is a 9% increase versus a year ago. Non-GAAP net income reached $3.8 billion, while non-GAAP earnings per share came to $0.94. Now let's dig into Q2 revenue a bit more. Product revenue totaled $10.2 billion, rising 11%, and service revenue reached $3.8 billion, rising 6%. Networking declined 3%, as gains in Wireless and Switching were offset by a drop in servers. Worth keeping in mind that Q2, '24 contained the final chunk of our elevated backlog shipments. Security jumped 117%, mainly fueled by growth in our Splunk, SASE and Network Security offerings. If you set Splunk aside, Security rose 4%. Collaboration climbed 1%, helped by gains in Contact Center, CPaaS offerings and Collaboration Devices, partly offset by a decrease in our OnPrem Webex Suite offerings. Observability rose 47%. Setting Splunk aside, Observability was up 3% in the quarter. Now for our recurring metrics. Total ARR closed the quarter at $30.1 billion, rising 22%, while product ARR grew 41%. Total subscription revenue climbed 23% to $7.9 billion, and it now makes up 56% of Cisco's total revenue. Total software revenue rose 33% to $5.5 billion, as software subscription revenue grew 39%. Total RPO stood at $41.3 billion, up 16%. Product RPO rose 25%, while total short-term RPO climbed 17% to $21 billion. Q2 product orders rose 29% versus a year ago. Setting Splunk aside, organic product orders were up 11% year-over-year. Turning to product orders by geographic segment, the Americas rose 30%, EMEA rose 24% and APJC rose 35%. Within our customer markets, Service Provider & Cloud rose 75%, Enterprise rose 27% and Public Sector rose 13%. Total non-GAAP gross margin landed at 68.7%, up 200 basis points from a year ago, right in line with the range we guided to. Product gross margin was 67.7%, up 250 basis points, thanks to Splunk and productivity gains, partly offset by pricing. Services gross margin reached 71.6%, up 110 basis points; we stayed focused on profitability and financial discipline, and non-GAAP operating margin came in above the top end of our guidance range at 34.7%. Moving over to the balance sheet, we closed Q2 with total cash, cash-equivalents, and investments of $16.9 billion. Operating cash flow was $2.2 billion, up 177%, mostly because of higher tax payments in the prior year. On capital allocation, we gave back $2.8 billion to shareholders in the quarter, made up of $1.6 billion for our quarterly cash dividend and $1.2 billion of share repurchases. Because of the confidence we have in our business now and going forward, our Board has approved an additional $15 billion for share repurchases, which brings the total outstanding authorization to roughly $17 billion. We're also lifting our dividend by $0.01 to $0.41 per quarter, marking our 14th consecutive annual increase. We keep making investments in our innovation pipeline, both through organic means and by acquisition. In Q2, we completed the purchase of Deeper Insights, which broadens the technology footprint and engineering talent of our Customer Experience team, speeding up innovation and momentum behind Cisco's CX AI capabilities and services. This move fits hand in hand with our internal R&D and is consistent with our strategy of bolstering our AI position through targeted strategic M&A. In short, it was a strong quarter, with results on both the top and bottom lines beating what we expected, powered by healthy order growth and margins. Our attention stays on strategic investments in innovation throughout our business, so we can make the most of the sizable growth opportunities we see ahead, all supported by disciplined spend management. That potent mix keeps driving our robust cash flow generation and our capacity to give substantial value back to shareholders. Ahead of the outlook, I'd like to offer some big-picture context about the U.S. tariffs that were recently proposed; our guidance reflects the extra cost from the higher tariffs on China, along with the proposed tariffs on Mexico and Canada. This is a fast-changing situation, but we've devoted considerable time to planning for it, drawing on the strength of our best-in-class global supply chain team and the flexibility and agility we've embedded in our operations over the past few years. We're ready to act to lessen the impact whenever tariffs take effect. Now to our financial guidance. For fiscal Q3, here's what we expect: revenue between $13.9 billion and $14.1 billion. We expect non-GAAP gross margin between 67% and 68%. Non-GAAP operating margin should land between 33% and 34%. Non-GAAP earnings per share should come in between $0.90 and $0.92. For fiscal year '25, our guidance calls for revenue between $56 billion and $56.5 billion. Non-GAAP earnings per share should fall between $3.68 and $3.74. We're assuming a non-GAAP effective tax rate of roughly 19%. Sami, let's go ahead and open it up to Q&A.

Thanks, Scott. Ahead of kicking off the Q&A part of this call, I want to remind the analysts that you get one question plus a single follow-up. Michelle, could you bring in the first analyst waiting in the queue?

I appreciate you fielding this one — it may be tough to address, and I'm grateful for what you said about tariffs. Still, perhaps the larger concern when it comes to the federal government is all the commotion surrounding DOGE. So it would be great if you could go deeper on your exposure to the federal space, plus what scenarios you have in mind should there be significant job cuts in U.S. government employment. Thank you. I also have a brief follow-up.

All right. Thank you, Simon, for asking. Well, clearly we've been tracking this quite closely. I believe it's key to recognize that this push is essentially aimed at achieving efficiency, along with productivity, within government, something that in numerous instances is in fact enabled through technology. Thus, we believe there may be certain benefits for us. It matters to note that the federal — U.S. federal business accounts for under 10% of our overall business, to begin with. And a further point to remember is that the DoD provides 75% of that U.S. federal business of ours; moreover, the bulk of this activity is in fact taking place on the civilian portion of the business. So we had a — we still had a decent quarter, I would say, when the executive orders first came out. We witnessed a burst of confusion, I'd say. After that, we've observed many large deals we'd been pursuing kind of return to their normal path, and all appears to be moving forward the way we anticipated. When you look ahead to Q3 and Q4, we aren't building into our models any significantly stronger results, nor any significantly weaker results in federal, compared with what we saw in the prior quarter.

Thanks, Simon. Please proceed, Simon.

If I could ask one more question. Right, so chatter has been growing regarding a technology called co-packaged optics, and considering Cisco's, shall we say, components — including optics, vertical integration, switching, routing — I'm just curious what your thinking is when it comes to your strategy over the longer term? Thanks.

Simon, my sense is that the way we approach this is, the way we currently go to market with the web scalers, is that we provide them systems that have optics integrated. We provide them standalone optics, optical systems, or silicon. So regardless of how the market develops, we'll keep developing what we offer, and keep developing our portfolio, simply to meet them wherever they are and however they wish to buy our products.

Thank you, Simon. Michelle, can we move to the next analyst?

Great. Thank you, Chuck. Perhaps to kick things off, could you share what you're hearing from customers regarding their willingness to spend? Clearly this was a very strong quarter. But as we've moved into the year, with somewhat more clarity on administration but perhaps less on directionality, just what are customers telling you? Then secondly, you're obviously seeing considerable strength in the data center area, and you mentioned seeing some of that on the campus side; but how is that strength extending into the campus portfolio? Thanks.

Yes, I believe, if you look. Thanks, Meta. I believe, if you look broadly, demand was fairly robust across geographies, across the portfolio, and aside from the U.S. Federal matter we discussed earlier, mostly across segments too. We observed public sector strength outside the U.S. in Europe, where it rose by high single-digits, I believe. And in Asia, performance was solid too. So we saw it in those regions, and we've seen it in campus. We mentioned double-digit demand for campus switching, saw the same in enterprise routing, saw healthy demand for wireless. And then for the fourth quarter in a row, double-digit demand growth in our Nexus portfolio, our data center switching, which mainly goes into private cloud infrastructure, into the enterprise. And then clearly the web scale results we saw with five of the — or, sorry, three of the six growing over 100% year-over-year, two of the six growing over 50% year-over-year, so it was quite balanced. And so far, I think that in spite of geopolitical risk, in spite of the uncertainty happening in the U.S. and in the marketplace globally, I think one thing our customers grasp is that their need to keep spending on technology simply persists; and there's so much change occurring right now from a technology standpoint that there's both enthusiasm about the opportunity. And frankly, there's a bit of fear of slowing down excessively and letting your competition get too far ahead of you. So we saw solid demand.

Thanks, Meta. Also, a quick reminder for the analysts: please ask one question along with a follow-up at the same time. Michelle, let's go to the next analyst.

Yes. Hello, and many thanks, everyone. Congratulations on the quarter. It's nice to see all of you having that momentum. Chuck, quite recently we received many inquiries regarding a newly released switch from you, the 9,300 Nexus. And within this programmable switch market, strong prospects appear abundant. I'm just curious whether you might elaborate a bit further on that offering and perhaps on the reason AMD was selected for it and what that signifies for your — compared with your Nvidia offerings? Then my follow-up concerns only the broader remark about general demand, which sounds far better than it did several quarters back. I simply wanted some — your perspective on whether you still believe people are preparing for AI? Or perhaps I didn't hear that from you this time — during the prior two quarters, people were preparing for AI and it accelerated. And this simply came across as somewhat more solid, yet I was just seeking a bit more detail on that. With luck, that didn't go on too long, Sami, but I'll hand it back to you.

Thank you, Ben. No, I caught both. Look, above all, I want to thank the team within our product organization, since the innovation they have delivered across the last 12 months to 18 months has occurred at a pace unlike anything we've witnessed in a very long time. We've seen plenty in the collab space, in the security space, a great deal in the silicon space and then within the web scale infrastructure. And now I believe what you mentioned, this switch containing the DPU, is an excellent illustration of the innovation we intend to drive within our core networking portfolio. And so, it's — I cannot address precisely why the team picked the AMD, DPU, except that I'm certain they conducted an analysis and concluded it was the optimal solution for us. But this is a genuine — it's a breakthrough product, since it enables us to, frankly, remove security from those appliances in which it arrives and place it where it properly belongs, which is everywhere. Security must be deployed everywhere, and this truly enables us to embed security deeply within the network and at the network's speed. So what this enables us to do is — you can operate network services or security services on these DPUs within the network at the speed of ASICs rather than shifting this traffic to a central location where you'd be constrained by the processing power of a single box. So the initial application, as you know, is Hypershield, which we rolled out, and during the previous quarter we secured a few significant Fortune deals, even though we're still finishing up the last features for Hypershield; therefore, that will be the first thing deployed. When you consider that scenario, you could place one of these switches positioned between zones inside a data center, pushing all traffic through it at line speed while performing security checks through Hypershield to safeguard various zones against East West threats. So we believe it holds substantial promise, and this is merely the initial service or first scenario the teams have devised. Regarding your next question, we firmly think much of this demand stems from customers getting ready for AI. We also suspect that a great deal of the telco demand we're seeing involves telecom companies expanding their infrastructure so they can be ready to offer AI as a service or services powered by AI. Thus, though I didn't mention it specifically, it's among the pillars we discuss concerning our part in AI, and we do believe there's a surge among businesses simply attempting to prepare for what they'll require. It's evident that genetic AI work streams will add more capacity to the network.

Thank you, Ben. Michelle, can we move to the next analyst?

Hey, everyone. I really appreciate you taking my question. Perhaps I can build on what was asked just before. You mentioned that enterprises are putting somewhat more money into AI. I'm curious, now that they've entered the new calendar, what are you observing and hearing, particularly regarding AI, from your enterprise customers? And how much does it actually matter for your business right now? Then, as a second question, perhaps this one's for Scott. This past quarter, you posted roughly 68.7% on the gross margin line. The guidance, as I read it, suggests a decline to somewhere between 67% and 68%. I'd like to know what's behind that expectation. Many thanks.

On the enterprise side, with respect to AI, here's how I'd put it: it's still — customers remain in the earliest stages, and every one of them recognizes they must determine precisely what their use cases will be. That said, we are beginning to observe some spending on particular AI-driven infrastructure. And our thinking is that once we get AI pods deployed, we've got HyperFabric on the way, we've got AI Defense on the way. Hypershield is already in the market; and we've got this new DPU switch, all of these will form part of the infrastructure needed to support these AI applications; so we're starting to see this take shape, but I also think it's genuinely crucial to grasp that when enterprises take advantage of their private data, their proprietary data, and they conduct some training on it and afterward run inference, clearly, using that data; our belief is that this opportunity is ten times bigger than what training has delivered for us so far; we will — we're going to keep innovating and developing capabilities that better position us to be a genuine beneficiary as this keeps speeding up. But as things stand today, we feel we're in a fairly solid position.

Matthew, regarding what you asked about gross margin: for Q3 the guidance is 67% to 68%, and incidentally, my sense is it lands within that band across the entire year. What sits inside that figure, and what causes the decline relative to Q2, is — and I mentioned this earlier in my prepared remarks — that we have included the expense of the proposed tariffs observed thus far. That means the extra 10% on China, plus 25% on Canada and Mexico. For Canada, the effect on us is negligible; as for steel and aluminum, we have again quantified those, presuming the March 1 date. My view is the environment is so fluid at present that saying what will truly occur is very hard. Even so, I wanted the guide protected, and to make sure we incorporated the 25% that has been proposed. Let me say that our supply chain group, over the past several years, has developed considerable capability concerning the tariff we faced in China and how we might navigate around it; and between when that was first imposed and now, our mitigation efforts have cut our exposure there by roughly 80%. Thus we have mapped out multiple scenarios and actions we could implement based on whatever actually takes effect, and we will begin implementing those actions to lessen the tariffs' impact. However, what I have placed into cost of goods sold, which is where tariffs will land, is the complete effect of 25% in Mexico and Canada plus an additional 10% in China, with no mitigation at this point — but no mitigation included at this time.

Thank you, Matt. Michelle, we can move to the next analyst.

Appreciate you fielding these. I suppose I'll begin by picking up where that tariff discussion left off. You've said you're factoring the tariffs into your costs, yet plenty of people are also asking how we should be thinking about demand. So Chuck or Scott — do you have any perspective on how enterprise buyers have historically reacted, from a demand standpoint, to earlier rounds of tariffs? And what does that tell you about what we ought to anticipate on the demand side? Also, are you noticing customers moving orders up to get ahead of some of these tariffs? Then for my second question, it's really about hyperscalers and the AI momentum you're seeing. Orders are showing strong momentum, but could you perhaps share some insight into deployments? We can all see the robust CapEx figures or forecasts coming from them, but is deployment activity speeding up, or is it roughly where you'd expected it to be? Thanks.

Yes, Samik, regarding your first question, we aren't seeing anything that suggests customers are accelerating purchases. The tariff situation is extremely fluid at the moment. It's a highly dynamic set of circumstances. Everyone is paying attention, but I think until it becomes clear what will actually take effect, our customers will find it very hard to determine their course of action. Another point — as we consider the steps available to us to lessen the demand impact, there are quite a few moves we could make to counter the cost; price is clearly one lever we'd have, but there are many factors we must work through before reaching that point. At present, I don't see that happening. What we'd want to do initially is implement certain adjustments within the supply chain to reduce the effect of whichever tariff framework ultimately is enacted. So, no sign of any demand being pulled in.

Yes. So Samik, regarding what you asked about webscale demand compared with their deployments, I'd say the key takeaway is this: I've sat down with every one of these customers, and they've told me directly, build more and we'll purchase more. A few of them came in with initial targets for 2025 unit volumes, and that figure has already increased by 50%. So deployments are progressing quite quickly, they remain on track, and we keep meeting all of their requirements. And honestly, the adjustments they make as their strategy shifts — it's highly dynamic. This market is highly dynamic.

Thank you, Samik. Michelle, we can move to the next analyst.

Hello. Fantastic quarter, though I want to ask about what might be the softer areas. Assuming my calculations are correct, and I can't be certain they are. Therefore, I need some clarification. Splunk declined 11% during this quarter compared to the same period a year ago, though perhaps my figures are completely off; so could you clarify what's going on with Splunk, and provide an update regarding the trends? My second question concerns security and observability. Those are expected to speed up — they did speed up during the previous quarter. Could you discuss what's happening this quarter and sort of the positive aspects along with the negative aspects of that growth? Thank you.

All right. I'll handle the first one, and after that, Chuck, the two of us can jump in together for the second, if that works for you. Regarding the — I don't know which calculation you're using, Tal. One factor that isn't apparent to you given the calculation sitting in front of you involves the close of Splunk's quarter, their fiscal — incidentally, this would have been the conclusion of January, which was the conclusion of their earlier Q4. Thus, just like any enterprise software company, naturally, Q4 represents the largest quarter of the year. Moreover, it actually becomes back-end loaded, meaning week 13 of that Q4 ends up being the largest week within the quarter. Given how the calendar landed this year, our fiscal quarter concluded in January 2025. Consequently, that final week of January, which historically was Splunk's Q4, in fact isn't included in our Q2 results; it will move into our Q3 results. That could be the reason your calculation is coming out wrong. Splunk continues to grow at a double-digit rate. It's genuinely right in line with what we anticipated, slightly above what we anticipated on the top line, and is profitable well before we had projected it would become profitable. So Splunk is doing quite well for us.

Yeah. Regarding security, Tal, here's my take: we've consistently told everyone that the pickup would arrive during this year's back half. Over the past 4 quarters, we've recorded growth in organic security where demand is concerned, and that's an encouraging indicator. What we've brought to market is a whole set of fresh innovations, and ramping those simply requires time — though I can share a few figures. Take just 2 of the offerings, Cisco Security and XDR. Combined, those 2 have now been deployed at upwards of 1,000 customers, and separately, each one supports more than 1 million enterprise users. So the traction is there; it just takes time to ramp. We've also rolled out Hypershield. As I mentioned before, during the quarter just ended, we landed 2 Fortune 100 enterprise accounts, and enthusiasm is substantial. Turning to AI defense and the AI Summit we held not long ago — I believe roughly 20-something customers are currently keen on pursuing proof of concept with us on that; close to 50% of the Fortune 100 attended that event. So my feelings about our position are positive. As we keep scaling these offerings, it will convert into stronger demand.

Thank you, Tal. Michelle, can move to the next analyst.

Okay, thank you for letting me ask something. I have a brief thing about the 700 million in AI orders that have come in so far this year. Would you say those are mostly optical wins where you're switching business? Or are you also landing wins with other products? And I'm also wondering — since those AI orders look quite strong — do you think any of that begins turning into revenue during this fiscal year, or does it land more in fiscal 2026? Thanks.