Netskope, Inc. (NTSK) Earnings Call Transcript & Summary
September 2, 2026
Earnings Call Speaker Segments
Michelle Spolver
executiveGood afternoon, everyone, and thank you for joining us. With me today are Netskope's Chief Executive Officer and Co-Founder, Sanjay Beri, as well as our Chief Financial Officer, Drew Del Matto. We issued a press release earlier today detailing our financial results for the second quarter of fiscal 2027; you can find it on our Investor Relations site, investors.netskope.com, along with the supplemental presentation. Before we begin, please note that certain remarks made during this call will be forward-looking in nature — this includes statements about our guidance for the third quarter and for the entire 2027 fiscal year, the opportunity in the market, prospects for growth, sales ramping, where we stand competitively, how AI is affecting us, and demand for AI security. This call includes forward-looking statements, which are subject to both identified and unidentified risks and uncertainties. Because of those factors, what actually happens may end up substantially different from what we've projected here. In addition, these remarks reflect conditions only as of the current date, and we are not committing to revise them later on. If you'd like a fuller picture of the risks and uncertainties involved, we encourage you to review the documents we file with the SEC, along with the press release we issued about our earnings. Before we go further, a quick note on the numbers: aside from revenue, everything financial we bring up today reflects adjusted, non-GAAP figures, unless we say otherwise. You can find reconciliations comparing those non-GAAP measures to their closest GAAP equivalents in the earnings press release we issued. With that, I'll hand things to Sanjay, who will walk through how the business is performing and the key points from our second-quarter results. Sanjay?
Sanjay Beri
executiveThanks, Michelle. We turned in a robust second quarter, and those results show that demand for Netskope's platform — which is highly differentiated — remains durable. With AI now part of the picture, protecting systems and modernizing networks can't be separated anymore, and organizations simply cannot keep sacrificing security in order to gain performance. As they adopt AI along with cloud, enterprises require a modern architecture that grasps the context and intent behind today's internet, cloud, and AI environments. Such an architecture has to safeguard enormous volumes of transactions and data that stretch across thousands of cloud and private applications and data stores, billions of websites and other destinations, plus a huge array of commercial and open-weight AI apps and models; it also has to inspect and control traffic in real time, at high speed, and with data sovereignty. The demand for this architecture grows even sharper as the volume and velocity of transactions and data, along with the count of humans and AI agents generating those transactions and working on data, expands exponentially. Through combining our Netskope One platform with the NewEdge private cloud network, Netskope is alone in providing this modern, scalable, resilient, and sovereign real-time architecture; that's the reason customers pick us as the trusted partner who helps them say yes to AI. What they want is to seize the enormous potential of one of the most defining technologies of our lifetime, all without giving up security, performance, or control. That puts us in an exceptionally strong position to go after a massive $170 billion greenfield opportunity in AI security inside our $336 billion total addressable market. I'll return to that shortly. But first, some highlights from the quarter. Q2 closed with ARR of $899 million, which is up 27% year-over-year, and we delivered net new ARR of $54 million. Revenue rose 29% year-over-year to $221 million, beating our guidance, while our net retention rate, or NRR, climbed to 114%. That outperformance carried through to the bottom line, as our operating margin improved 11 percentage points year-over-year to — 9%, well above the range we had guided. Demand stays strong for our Netskope One platform of 25+ security networking, analytics, and AI products, since enterprises keep modernizing their infrastructure for the AI era. We felt particularly encouraged by the traction coming from our AI security suite, which we announced recently. Although it's still early, we're seeing strong customer engagement and rapid pipeline generation for these products, with a number of deals closed and many more now sitting in the proof of concept, or PoC, stage. Around the world, enterprises remain strategically focused on modernizing their security and infrastructure. We are dealing with technology sprawl, protecting sensitive transactions and data, ensuring data sovereignty, and, naturally, using AI safely. One hot topic in my daily conversations with CXOs is that AI generates exponentially more transactions and data along with a far more complex attack surface, and how modernization and AI adoption move together. The Mythos moment last spring made clear just how fast the landscape is shifting. Innovation is moving quickly in frontier models, and open-weight models keep getting more capable, which speeds up how fast we can find weaknesses and harden defenses using AI. Yet that speed cuts both directions. Such models likewise make things easier for those who would do harm, shrinking how long it takes to go from spotting a flaw to abusing it, and widening what can be attacked — not just by people with bad intentions, but more and more by autonomous AI agents running at machine speed; we already witness this happening as AI agents break out of contained settings, take advantage of flaws, gain higher privileges, move sideways through systems, and swipe credentials, even though nobody told them to launch an attack. They had merely been asked to finish a cybersecurity benchmark, and once the expected path turned out to be too hard, they located a different route to the answer. This may be the strongest illustration so far of cyber danger going past human attackers who use AI. Having malicious intent is no longer what defines the security line. An agent doesn't need a prompt telling it to behave badly. Rather, a task that's been stated, plus sufficient autonomy inside a setting that can be worked around, may be all it takes to cause harm. Everything here underscores that the threat of a rogue agent isn't some one-off event or a fluke. It's a genuine, growing control risk that CISOs contend with right now, and it widens the security challenge substantially; that's the reason AI security has become such a vital priority for enterprises. What's hard is no longer just figuring out how to protect AI models or stop staff from adopting AI. What's hard is how to provide enterprises with the visibility, control, protection, and performance they require while AI grows deeply woven into their people, applications, data, and increasingly autonomous agents. Over this quarter I've held more than 100 conversations with customers, and in nearly all of them, some CISO or CIO returns to one question: how can we go faster with AI while keeping control? Security, IT, and infrastructure leaders have no wish to reject AI use. Their wish is to embrace it, yet do so in a safe manner. Allow me to walk through several challenges they're up against and how Netskope helps address them. To begin, let's take visibility. Most customers with whom I discuss AI security have no idea what AI models and applications get used or how, which corporate data is being fed to them, where agents sit inside their organization, what those agents can reach, and what they're up to. Netskope's platform fixes this by letting companies respond to the question of which AI I'm using, which includes delivering complete visibility into agentic and MCP traffic through our Agentic Broker; it closes the distance between a human or agent and LLM interactions, or machine-to-machine workflows. For customers protecting AI, Agentic Broker has served as a natural place to begin and a game changer as they deal with unsanctioned and unmonitored AI usage inside their organizations. Beyond that, with the AI Command Center we just brought out, clients get one consolidated, live, uninterrupted, and linked picture of their AI risk exposure, plus policy and remediation guidance that assists them in taking action. The next challenge customers face involves building an appropriate protective layer surrounding AI so that threats unique to AI, such as prompt injection and jailbreaking, are blocked, and so models stay compliant with corporate rules, which stops misuse or undesirable outputs. Picture a situation in which an attacker tries, across multiple turns, to bypass system instructions so they can steal data. Our AI Guardrails offering was built specifically to assist customers with exactly that. Furthermore, our AI Gateway protects API traffic flowing among private applications, autonomous agents, and LLMs, and it can run either on premises or in the cloud. Number three, customers require a network with strong performance capable of managing the enormous quantity of AI transactions and data, along with the expanding volume of highly interactive, latency-sensitive agentic communications, all while making sure they remain compliant with tight regulations, data sovereignty included. More than 120 data centers worldwide make up our NewEdge private cloud, and at every one of those sites we run all our products on our unified platform, which produces clear advantages in both performance and sovereignty. NewEdge additionally lets customers set geo-based policies so they can dictate precisely where processing for their AI security and networking takes place, granting them sovereignty over their transactions and data no matter where it resides or travels. Customers now recognize that the AI landscape of today has turned into a defining moment for security.
Sanjay Beri
executiveOur AI security pipeline is expanding quickly, and deals are advancing into proof-of-concept stages. Actually, we figure that roughly one-third of that pipeline has already reached or is about to reach the crucial PoC stage. Typically, enterprises go through their established cycles for budgeting, validation, executive sign-off, and procurement, and that process usually spans six to 12 months. Let me walk through a handful of early AI security wins that closed in the second quarter, along with the use cases we address for customers. To start, a global electronics manufacturer in EMEA required visibility into agentic or MCP traffic, plus a method to grasp the related risk, turn on governance, and establish a control point for enforcing policy; during Q2, they broadened their current Netskope deployment through a wide-ranging AI security upsell that covered AI Guardrails, our DLP AISecOps Agent, Agentic Broker, and AI Gateway; they are implementing those controls so they can safely adopt agentic AI instead of having to slow it down or switch it off. Take another example: a large auto insurer received a leadership mandate to push AI adoption throughout the company. Yet they understood that they could not sustain that speed absent the proper security and governance foundation. In Q2, they extended their Netskope One platform with AI Guardrails, AI Gateway, Agentic Broker with DLP, and Red Teaming, which provided the visibility, controls, and guardrails needed to proceed with AI confidently. These wins demonstrate how we help customers say yes to AI right now, allowing them to use AI safely rather than block it, and to move more quickly with it, while trusting that their transactions and data remain protected, their AI usage stays governed, and their agents stay within suitable boundaries. Netskope provides that AI runtime security, along with the guardrails and high-performance network that customers require to embrace AI widely without sacrificing security or user experience. As security leaders work through an ever more complicated AI landscape — spanning open-weight models to closed frontier models, copilots to autonomous agents, and traditional AI applications to MCP-based interactions — they require a platform through which they can see, understand, and govern everything. That is precisely what Netskope One was designed to accomplish. Right from the beginning, our AI-native platform was created to give customers granular visibility along with real-time context and control over all their transactions, users, agents, tool calls, data, and beyond. This involves dynamically and intelligently comprehending the nature, intent, and risk of those transactions, paired with a high-performance private cloud network that supplies both security and performance while enabling real-time policy and security enforcement. That foundation is growing even more vital now in the AI era. We are thrilled to witness this significant validation of our product market fit and strategy striking a chord with our current customers, along with a robust AI pipeline of opportunity among new customers. Beyond our early achievements in AI security, our platform selling motion keeps generating momentum throughout our SSE and SASE business, as customers adopt a growing number of products across our Netskope One portfolio. In Q2, we saw a 23% increase compared to the same period last year in the count of clients who spend over $100,000 on ARR. Meanwhile, 59% of those we serve now utilize at least four Netskope One offerings, which is up from 51% during the prior year. Throughout the second quarter, we achieved excellent additions of new logos and expansions spanning various regions and major sectors such as financial services, manufacturing, healthcare, telecom, and government. Allow me to highlight a handful that demonstrate the primary challenges we address for clients across main scenarios. To start, clients keep choosing our Netskope One platform to update their operations for the cloud and AI. As I've noted before, this kind of modernization is a crucial step toward safely implementing AI. Consequently, clients are picking up our SSE and SASE solutions as the groundwork and base upon which they can then incorporate our AI security products; we observed this in an outstanding new logo acquisition with a U.S. financial services firm where cloud modernization and enabling AI are central efforts propelling their upcoming expansion and growth. Our platform's distinct advantages in networking and security allowed us to secure a competitive agreement where they acquired 8 offerings from our SASE portfolio; furthermore, they also adopted our AI Guardrails and Agentic Broker AI security solutions. In a similar vein, we secured another deal involving cloud modernization and AI with a top technology firm that required better insight into SaaS and cloud environments and data safeguarding, along with protecting and managing AI utilization, such as shadow AI, and overseeing and regulating MCP traffic. Once more, our lone integrated platform and exceptionally swift NewEdge network served as the distinguishing elements versus rivals in this transaction. We keep witnessing clients swap out outdated setups for our contemporary SASE design engineered for scalability. For instance, a Fortune 500 healthcare organization picked Netskope to update its security, substitute disjointed old systems, and reduce vendor proliferation through a single platform for a worldwide dispersed staff. Achieving this meant safeguarding extremely sensitive intellectual property and other information for regulatory adherence and securely facilitating and managing heightened generative AI use. Our finely detailed contextual controls, integrated data safeguarding, and NewEdge's superior speed were crucial in securing this seven-figure, multi-product agreement away from a competing incumbent. Lastly, data sovereignty is growing in significance for clients in heavily regulated sectors and public institutions. For instance, we broadened our work with a European government body that selected us due to our data sovereignty strengths and acquired our digital experience management to complement our in-country NewEdge network for enhanced user experience. Another instance involves a fresh victory with a financial services firm where data sovereignty is essential for meeting regulatory requirements; beyond accomplishing this via our NewEdge data planes, which operate all our offerings at the sovereign edge, they likewise streamlined and updated their outdated network security instruments using Netskope's integrated SSE platform. As these successes show, clients are drawn to Netskope to update their networking, establish a fundamental security platform for the cloud and AI age, and remove the compromise between security and network speed. Changing topics for a moment — for a long time now, our philosophy has centered on transparency, working together as an industry, and connecting things throughout our ecosystem. Back at the start of this year, we became part of Anthropic's Project Glasswing as well as OpenAI's debate programs, and we put out integrations alongside those companies and other cloud and AI partners. What these alliances show is how significant a part Netskope plays in the wider AI and security space. During Q2, we kept growing and strengthening those relationships, and on top of that, we revealed major new partnerships. It was a pleasure to become part of NVIDIA's Open Secure AI Alliance, a group of industry leaders dedicated to creating open frontier AI tools that defenders are able to inspect, adjust, and rely on. For genuinely positive and impactful results to reach the world, both open and closed frontier models are needed, coordinated proactively and carefully throughout the whole AI ecosystem. Building alongside NVIDIA has been wonderful, and through this effort we speed up our dedication to creating open AI tools, making certain that our enterprise customers are able to trust, adapt, and safely deploy advanced AI throughout their environments. We were delighted as well to become part of CrowdStrike's Project QuiltWorks, feeding real-time data originating from Netskope into Falcon's next-gen SIEM, which delivers essential visibility across users, applications, and data, and assists defenders in automatically correlating risk and acting with greater speed. Our work with Anthropic also kept expanding, as we integrated our industry-recognized DLP and threat scanning into Claude Enterprise.
Sanjay Beri
executiveAlso, we told everyone about a new integration with Amazon Bedrock AgentCore that brings Netskope AI Guardrails into agentic workflows for AWS customers. Organizations can now put AI agents into production knowing that what an agent is permitted to do and what it truly does match. And on the go-to-market partnership side, we rolled out the Netskope Catalyst Managed Service Provider program, which makes it easier to deliver managed services built on Netskope solutions. We've said before that our relationships with MSPs worldwide matter to us, and we see those relationships as a key vehicle and lever for growth in the mid-market. On the prior quarter's call, I talked about how Netskope is transforming its operations and how artificial intelligence is accelerating how quickly we ship products. Throughout Q2, we kept up that unrelenting speed of innovation. Let me take you through a few of those developments. Four weeks ago, we introduced Netskope One DataSec Command Center, a single, unified control plane that locates, understands, and protects sensitive data no matter where it lives — across the cloud, the network, on premises, on endpoints, in email, and within AI applications. It addresses an issue I hear raised by CISOs constantly. They still lack one centralized view of where their sensitive data sits. That lack of cohesion creates a sizable, underserved market opportunity for a platform that can bring it all together. The way DataSec Command Center achieves this is by tying together signals coming out of DLP, DSPM, CASB, SWG, and other sources, which lets teams go from spotting a risk to resolving it with just a few clicks, instead of spending days investigating across tools that don't talk to each other. Sitting next to DataSec Command Center is the DLP AISecOps Agent that we rolled out during AgentSkope last quarter. As their use of AI grows, customers are paying more and more attention to getting the best cost and performance out of every workload, and that's fueling demand for network optimization built specifically for AI. Back in July, we shared the real-world results from our AI Fast Path technology. On the AI side, our Fast Path capability improves the route taken between where users and agents sit and the AI endpoints they reach, which speeds up inference and shortens the wait for the first token, helping agentic AI workflows run faster. When we tested it in real conditions across our NewEdge network, latency dropped by up to 90%. Every day, NewEdge evaluates tens of millions of routes, weighing latency along with jitter, packet loss, and additional variables, and in the end executes tens of thousands of route adjustments so AI traffic travels the quickest, most dependable path. Outside of AI, innovation continues throughout our Netskope One platform — in Q2, for instance, we rolled out improvements to both our enterprise browser and our zero-trust access offerings. Lastly, we upgraded our platform in response to quantum computing breakthroughs that are compressing the Q-day window — the point at which certain encryption methods underpinning internet security could be broken. Advanced adversaries are carrying out a "harvest now, decrypt later" approach, stockpiling encrypted data today so they can unlock it once sufficiently powerful quantum machines exist. Because of this, governments worldwide have established hard deadlines for deploying post-quantum cryptography. To meet that challenge, Netskope developed post-quantum cryptographic algorithms approved by NIST and integrated them directly into its SASE platform throughout its global network of more than 120 data centers in Q2, bringing quantum-resistant encryption to every corner of the globe. That makes it easier for our customers to shift into a quantum-safe setup and satisfy regulatory requirements covering their sensitive communications with SaaS and AI services everywhere. We hold a strong conviction that we possess — and keep strengthening — the ideal platform for this particular moment; during Q2, we took pride in earning meaningful third-party confirmation of our leadership in important markets. Once more, Netskope earned a leader position in the well-regarded Gartner Magic Quadrant, for SSE for the fifth year running and for SASE for the third consecutive year. Along the same lines, in Gartner's accompanying Critical Capabilities report covering SSE, Netskope placed among the two top-scoring vendors across all four category use cases: essential SSE, advanced SSE, private application access, and secure SaaS and AI enablement; in the matching Critical Capabilities report on SASE, Netskope stood alone as the vendor with the highest score for three key use cases: foundational SASE platform, zero-trust SASE platform, and sovereign SASE. On top of that, IDC's worldwide SASE MarketScape report, released last month, acknowledged Netskope for SASE leadership, highlighting our single policy engine, common data model, and NewEdge's distributed enforcement as major differentiators; that also underscores something essential to grasp. Even though AI dominates the discussion, contemporary cloud and network security forms the bedrock for corporate AI adoption that stays safe without sacrificing performance. Indeed, as agentic infrastructure grows, customers increasingly see that speed amounts to a clear-cut competitive edge, and that Netskope provides the best route for inference. I've talked before about Netskope's AI-native philosophy — not just in the way we build our market-leading platform, but in the way we run our business. Nowadays, AI speeds up how we function throughout the company, helping us innovate and grow our Netskope One platform quicker than at any point before, while also speeding up sales rep and SE training, making customer support smoother, streamlining recruiting and talent development, and automating additional processes. Our teams have embraced this new era, which lets us move quicker, run more efficiently, and scale with more leverage. To wrap up, Netskope occupies the crossroads of cloud, AI, networking, and security, which puts us in position to pursue an enormous market opportunity whose capture is still at an early stage. We're expanding our route to market in a way that works, so we can grab that opening, and at the same time we keep rolling out new ideas fast and making our position with clients and allies even stronger strategically. What we're after is to turn into the must-have adaptive fabric for the AI-driven enterprise of today, and we have every belief that our unique design, our lead in technology, and a growing roster of customers add up to a durable structural advantage that will build on itself over time. When it came to Q2, we came in above what was expected across every metric that matters, and that makes me incredibly pleased. The Netskopers on our team fill me with pride daily, since they continue to demonstrate the courage, determination, honesty, and grit that sit at the core of who we are and what we believe in. On top of that, I want to thank the thousands of customers who put their faith in Netskope to guide them through cloud and AI — two of the biggest technology revolutions we will ever see. With that said, I'll hand things over now to Drew.
Drew Del Matto
executiveThanks, Sanjay. You've just heard it: the appetite for what we offer is robust. Momentum keeps building behind our platform-led sales approach, and innovation is happening at a fast clip. Before I go further into our Q2 numbers, a quick reminder — unless otherwise noted, every financial comparison is presented year-over-year and on a non-GAAP basis. Now, on to Q2. ARR climbed 27%, reaching $899 million. Net new ARR came in at $54 million, up 9%. Revenue rose 29% to $220.5 million, surpassing the range we had guided to. Across every region we serve, demand remains durable. EMEA revenue was up 37%, APJ was up 31%, and the Americas was up 25%. Customer expansion and retention rates also mirror the strength of these results. NRR moved up to 114%, while our gross retention rate, GRR, reached yet another all-time high, rising once more in Q2. Remaining performance obligations, RPO, increased 36% year-over-year, hitting $1.35 billion. Now for customer metrics: as Sanjay pointed out, customers producing over $100,000 in ARR rose 23% year-over-year in Q2, totaling 1,686; those customers account for 87% of our total ARR. More and more, customers are adopting our Netskope One platform. By the close of Q2, 59% of them used four or more products, compared with 51% a year earlier, while 41% used five or more, up from 35% a year earlier. Our platform expansion keeps gathering pace as we bring additional products into Netskope One, which now spans over 25 products. This ongoing innovation widens our market opportunity to $336 billion and lengthens our growth runway. Shifting to the remainder of the income statement: our spending stays disciplined; we're showing the operating leverage that results from building our platform and infrastructure to scale. Gross margin stood at 77%, an increase of roughly 2 percentage points year-over-year. The scale advantages of our NewEdge architecture are behind this gain, as we keep advancing toward our long-term goal of 80% gross margin. Q2 operating margin was — 9%, a striking 11 percentage point gain versus Q2 of last year, and well ahead of what we had guided. Operating leverage throughout the P&L, as revenue grows, drove this improvement. R&D was the largest contributor, improving roughly eight points as a percent of revenue versus last year. The shared platform architecture of Netskope One delivers the rapid product velocity Sanjay referenced earlier, even as we scale efficiently. Our AI investments are speeding up that velocity. Sales and marketing expenses stayed roughly flat year-over-year as a percent of revenue, as we continue ramping our existing sales force and putting money into quota-carrying sales reps to go after the enormous market opportunity in front of us. G&A expenses likewise improved about one point as a percent of revenue versus Q2 of last year, reflecting leverage throughout our infrastructure. Net loss per share came to $0.03 on 405 million weighted average shares, beating our guidance. On a fully diluted basis, applying the treasury stock method, we had roughly 511 million shares outstanding as of July 31st, 2026. Free cash flow came in negative at $29.8 million, which was a bit better than what we had anticipated; the upside came from beating our expectations on both the top and bottom line. Keep in mind that contracted future billings rose 75%, which mirrors the move we're making to annual billings. As Sanjay mentioned, the results AI is producing are already impressive. Part of this transformation involves moving certain investments into the places where we see the biggest opportunity and demand. So we shifted spending toward our AI infrastructure and tokens within R&D and G&A. We took the difficult step of cutting roughly 5% of our workforce as we keep pushing AI nativeness across the entire company. Lastly, our balance sheet remains strong, and we closed the second quarter with $1.1 billion in cash equivalents, and marketable securities. Now for some modeling points and assumptions that sit behind our Q3 and fiscal year 2027 outlook. Starting with ARR. We still anticipate that net new ARR will increase year-over-year during the second half of our fiscal year. That's consistent with our usual second half quarterly cadence, which features a seasonally stronger fourth quarter. Turning to billings, just a reminder that we're moving customers to annual billings, and that process is moving along faster than we had expected. We anticipate finishing that transition by the middle of next fiscal year; this change temporarily pushes out cash collections, but it provides us with strong forward visibility into cash flows and customer commitments. On cash flow, we're looking for somewhere between $10 million and $20 million of free cash flow in Q3; for the entire year, we now anticipate capital expenditures of about 4%-5% of revenue tied to the ongoing infrastructure investments in our NewEdge network. We've laid out these modeling points in the appendix of our investor presentation. Now I'll walk through our guidance, which captures the strong underlying demand, early traction with our AI security products, and continued progress in sales reps ramping. Just a reminder, all of these numbers are non-GAAP unless we say otherwise. For Q3 fiscal 2027, we anticipate revenue in the range of $227 million-$229 million, which represents growth of approximately 24%. Operating margin of approximately — 8% and net loss per share of $0.03-$0.04, using approximately 415 million weighted average common shares outstanding. For the full fiscal year, we're raising our guidance; we now anticipate revenue in the range of $888 million-$892 million, which represents growth of approximately 26%. We're happy to raise our full-year revenue guidance by more than the amount we beat in Q2. That speaks to our momentum and our confidence in how durable demand is. Gross margin of approximately 77%, operating margin of approximately — 9%, net loss per share of $0.15 using approximately 415 million weighted average common shares outstanding, and positive free cash flow margin of approximately 2%. To wrap up, demand for Netskope solutions is strong. Our platform momentum keeps building, and the rapid pace of our innovation puts us center stage for the age of AI. Thanks, Operator. At this point, we're ready to take questions from the line.
Executive
executiveHi there. Emeran here, filling in for Matt Hedberg. I appreciate the question, and nice work on the quarter. To start with me, when we look at ARR and how it decelerated compared with the prior quarter, might you expand a bit on the key building blocks that might suggest ARR speeding up going forward?
Sanjay Beri
executiveAppreciate that. Thanks for asking. To start, we're clearly quite pleased with how Q2 turned out and with the expansion we're seeing in our pipeline when it comes to AI security and other areas too. As we've said, one thing that matters a lot to us is that we can see that AI security pipeline, and a portion of it did close during Q2. That said, the bulk of it we expect to land in the second half, particularly as we get closer to Q4. And since our reps are ramping, that's clearly essential for our growth. Roughly 50% of our reps are in that ramping stage; and a crucial focus for us isn't just innovating on the product side, but also keeping that number of ramping reps increasing. That's going to take place during the latter part of the year, starting with EMEA and APJ, since those are the regions where we first really began building up our fully ramped reps, and after that in North America.
Executive
executiveThanks, that's useful. Now, if we zoom in on the AI component, can you put a bit more of a number on it and assist us in gauging how big this contribution is? And then, more broadly, what makes you well placed for the age of AI?
Sanjay Beri
executiveYeah, it is a great question.
Executive
executiveLonger term.
Sanjay Beri
executiveAbsolutely. I'm not able to sit down and talk with the people who buy from us or might buy from us. I was hosting an AI Fast Path event. These are gatherings we put on in cities all over the globe. I was up in New York with a big global media CIO sharing the stage with me, and more than that; we're holding these events in many places. The reason we're in a strong spot is, first, if you look at our platform, it's genuinely a wide-ranging platform for more than a third now of the Fortune 100, and we serve as their inline processing point; their traffic flows through us; we've got a public site, ai-index.netskope.com, where you can literally observe which AI traffic is moving through enterprises right now. We're handling trillions of AI connections. The reason our footprint is so good is that a great deal of this traffic is already visible to us. It already passes through NewEdge, the quickest route for inference on AI transactions; now those new offerings we brought out, Agentic Broker earlier in the year, Guardrails, our AI Command Center last quarter, they truly illuminate that traffic, asking, "Hold on, what is that traffic? Does it come from agents? Does it come from users? Is it prompts, responses? What type of data?" We hand them the visibility they're after with, honestly, very little deployment effort. Getting that visibility is quite simple, and after that we let them enforce policy in real time. For us, it's really a mix of the things we consistently emphasize, the reality that we operate one of the world's largest private cloud networks; we've launched AI Fast Path. Put that together with how finely we can interpret the language of the internet for cloud and AI plus our data protection, and you end up with an almost ideal traffic point for governing AI. That's the first thing. The second question you raised alongside it had to do with the AI growth pipeline and similar topics. We began rolling out many of our AI security products in Q1. We shipped additional ones in Q2, and we've in fact already unveiled one tied to that, our DataSec Command Center, in Q3, this quarter. For us, those are entering PoC. I believe we said close to one-third of them are now in proof of concept. Enterprises stick to their usual cycle. PoC, correct? Oh, okay. Once they finish the PoC, they secure budget, then they move through procurement, and so forth; we truly expect that standard six to 12 month cycle, and that's the reason we pointed to the back half of the year, where we expect some of those deals to close, on top of the ones that will close this quarter. Still, good question.
Executive
executiveGreat. Thanks, guys. Congrats again.
Jonathan Ho
analystHi, good afternoon.
Sanjay Beri
executiveHello.
Jonathan Ho
analystHi there.
Analyst
analystExcellent. I appreciate you fielding my question. Bob here, filling in for Brad at the moment. Sanjay, I'd like to stay on the same topic regarding your AI security suite. It's encouraging to witness such solid demand for the offering. Could you perhaps share additional specifics about which offerings within the suite are landing best with buyers as they begin these PoCs? And on a wider level, do you find these fresh capabilities are opening doors to more potential customers that previously might have been tougher to reach?
Sanjay Beri
executiveYeah, that's a fair question. So regarding the offerings that are landing. I keep returning to use cases. When you're a CISO or CIO, and you examine things now, you're aware that a big chunk of AI activity inside your organization is driven by business units or happens in the shadows. In lots of enterprises, 90% of AI activity fits that description. The initial thing on your mind is, hold on a second. Show me what I'm actually using. Which AI is present within my organization? Which agents, which rogue agents, which sanctioned agents, which MCP servers? Provide me with that visibility. After that, naturally, enforce my policy; it's likely no shock that what's landing is, first, our Agentic Broker. What's its function? It illuminates which agentic traffic and agents exist inside your organization, correct? It informs you what those agents are reaching. Are they heading to my corporate Office 365? Are they touching private data? What exactly are they up to? So that's one, Agentic Broker. That's how you comprehend agentic usage, and afterward enforce policy, since we aren't a visibility platform. We're a real-time policy enforcement tool and platform too. The second one is Guardrails. Guardrails examine each prompt and each response, and they ask, hold on, how do I ensure that beyond whatever Netskope provides me on granular context, such as telling me that's a corporate version of Claude Code, a personal version, and allowing me to make governance decisions about which kind of AI can be utilized. How do I also, once it's utilized, ensure it isn't emitting data or content that falls outside my acceptable use policy? I don't want it emitting weapons content or content that isn't applicable to what I want my organization to hear. Guardrails is that second component. Those were in fact the first two that we began releasing; then we actually released our AI Command Center last quarter; that's a central governance point where you can view all agentic usage within your organization. We anticipate over time that it'll become another major driver for us.
Analyst
analystOkay, maybe just a single follow-up question for Drew, regarding how net new ARR moved up on a quarter-over-quarter basis compared with the prior quarter. Could you perhaps walk through what drove that increase and whether any particular item stands out?
Drew Del Matto
executiveNo, I'd say it simply reflects robust demand across the business. A few AI deals were in there, but I'd characterize it as demand within the business overall. The AI pipeline continues to be solid. That's something we anticipate, I think, will fuel us across the long run, along with solid staying power of demand over the longer term.
Analyst
analystThanks for taking the question.
Drew Del Matto
executiveYou're welcome.
Drew Del Matto
executiveThat's a solid question.
Jonathan Ho
analystHello, I'm here again. With any luck, this time I'll be able to address what you asked. I was hoping to go a bit deeper on what you said regarding quantum proof cryptography. Could you speak a bit about that capability, and does it either create fresh opportunities for you or boost your capacity to observe win rates? I'd just like a bit more detail on that.
Sanjay Beri
executiveYes, excellent question. Consider what we've now deployed globally — it genuinely lets our customers, whether that's someone as an end user, or someone working on a manufacturing floor, or someone on any system, communicate with Netskope through quantum resilient encryption. We rely on lattice-based encryption, which is essentially the standard NIST puts forward. Because of that, that essential piece, sending traffic out to the internet, gets protected. So what does that unlock for us? Truly, it lets us stay well ahead of the schedule for when people are advised to roll out quantum resilient encryption. For our company and the people we serve, that means, hold on, if I run a financial services company, if I run a healthcare organization, I understand that with no action at all, truly, if you're a Netskope customer, you don't genuinely need to lift a finger right now to turn on quantum resilient encryption. You've got Netskope, you've got the platform, and we embedded that capability for you. Truly it lets us accomplish that. And later, as more websites and AI applications adopt it, and they back quantum, Netskope stands ready. That site, once it provides support, or that app, Netskope will be capable of communicating with it through quantum resilient encryption. So it's future-proofing, it's currently letting organizations satisfy their regulatory environments, and it's letting organizations take advantage of it right now. That's genuinely, for us, a crucial thing; we always aim to skate to where the puck is heading, and that's a fine illustration of it. That will assist us clearly in proof of concepts, sustain our strong win rates, which exceed 80% once we reach a PoC. That's likely a decent way to view it.
Jonathan Ho
analystGreat. Could you provide a bit of an update regarding the federal government sector and a few of the chances you have involving either FedRAMP High or some of your sponsoring agencies, plus some fresh possibilities that are emerging, particularly around zero trust initiatives on the government side? Thank you.
Sanjay Beri
executiveYes. Wonderful; we believe our position for the federal market is very strong. As you're aware, we achieved FedRAMP certification, FedRAMP High certification. We began assembling our federal team. This year we added our federal leader for the U.S. federal market, and since then we've really been growing our sales team and developing them; for us, federal represents a small portion, a small slice of our business, yet a highly important expanding one where our platform is a great fit; we feel quite confident about that and about being able to serve both the commercial and beyond side of the federal for many years ahead.
Richard Poland
analystHi folks. Appreciate you fielding this. Sanjay, I'm simply wondering — the broader AI narrative, the way it's advancing, came across as quite positive. I believe you said a third of the AI security pipeline has already reached the PoC stage, with a usual sales cycle of six to 12 months. So starting from the PoC stage, is there generally a rough estimate of where within that six-to-12-month window we tend to be? Any insight you can offer regarding what the uplifts have looked like thus far for those that have closed, or just some framing around the monetization aspect — though it's likely not that significant yet — any early signals on that front. Thank you.
Sanjay Beri
executiveYes. That's an excellent question. AI security wins are happening for us. We touched on a few during my opening remarks. Be it in financial services, be it in tech companies and other areas. We're experiencing solid momentum with people genuinely embracing our AI security offerings. Looking backward, we truly launched our AI security products, beginning the rollout in Q1 of this year. We brought out additional ones in Q2, such as AI Command Center, and then we shipped another related offering, our DataSec Command Center, this quarter, merely three, four weeks back. When you consider what a standard enterprise cycle looks like, what they do first is assess. They examine, "Alright, what's my issue? Alright, that's discovering AI, comprehending it. Let me assess a solution." They proceed to PoC. What they do next is, quite often with AI, they go request funding. It might come from a stream, where a committee convenes each quarter, and they say, "Alright, I need to secure budget for this." Once approved, you then advance. That's simply the ordinary enterprise sales cycle. That genuinely sits within that six-to-12-month sales cycle process; on our end, we converted, and have converted, several of our earlier beta customers since they got an early look at it; we observe a portion of that pipeline that has reached PoC really in the later portion of the year. More toward the year's end, where some of that is converting; we anticipate converting several of our more beta customers and beyond in Q3. We see that pipeline growing, we see the PoCs growing, and consequently, we see the ARR growing. That's the way I'd view it. The final point I'd add is AI security, as far as we're concerned, isn't a product. It's a component of our platform, and it's actually made up of multiple products. Customers too, in time, will take chunks of it. For us, we're genuinely constructing a major pillar of Netskope where you'll have numerous products over time, and as your AI security sophistication increases, you'll expand alongside Netskope. Naturally, we want to get that feature and the new product out there well before you actually require it, so that's a helpful way to frame it in your mind.
Analyst
analystGood. Ryan here, filling in for Meta, and I appreciate you fielding the question. Could you share any further specifics regarding the divestiture of the AI product portfolio and the way that's affecting sales cycles? Are you observing them shorten as customers seek to advance their security stack far more rapidly, or lengthen because the cycle might be more complicated? Any extra specifics on that would be useful.
Sanjay Beri
executiveYes.
Analyst
analystThanks.
Sanjay Beri
executiveYes, of course. When we think about AI security, we have two groups: the folks already with us, plus clearly brand-new ones, and we pursue both. For someone already with us to bring in AI Guardrails and Agentic Broker, one great thing is that our platform was built in-house, correct? We developed it from scratch; we ship features only once they're genuinely connected, correct? We don't merely slap them on a price list. We genuinely connect them through one shared GUI, shared policies, a single data protection engine, a single threat protection engine. That in-house method of truly constructing a platform the right way lets customers bring these offerings on board and roll them out technically with great ease. So with Agentic Broker and Guardrails, you can switch that on. Suppose you've deployed our next-gen SWG product, for instance — you simply turn it on, and you're able to test it; that's the great part of having it. It's one shared GUI plus DLP. That's the first thing. Then there's a different group of offerings, such as the AI Gateway, which is designed for east-west coverage of your AI traffic, perhaps inside your public cloud or your data center. Clearly, you'll install that, correct? You'll deploy it. What we aim for is simply to make it as simple as possible, to ensure we cover north, south, east, west, and all through one GUI with shared policies; so I believe that helps customers deploy technically with less friction; they still go through their sales cycle, though, correct? Apart from that. But what we aim for is to make the PoC portion as simple as possible.
Analyst
analystHello, thanks for giving me the chance to ask; John here, speaking for Brian. I'd like to focus on the CapEx piece. You said full-year CapEx is now anticipated to land at roughly 45% of revenue, connected to NewEdge. My question is whether this increase comes mainly from demand-driven capacity, or whether it's a pre-build in front of the agentic traffic you expect? I'm wondering, as that traffic expands, ought we to expect CapEx to grow alongside it? Or would the architecture eventually take in that volume at a reduced incremental cost? Thanks.
Drew Del Matto
executiveNo, that's a good question, John. Well, this is ongoing infrastructure spending. Once more, we anticipate solid demand ahead. We've consistently indicated it would be low single digits. I believe we mentioned a range of 3% to 5%. Now we're saying 4% and 5%, between 4% and 5%. I'd say we're fairly aligned with what we've stated throughout. Frankly, we've experienced some growth, we're doing slightly better than expected, so perhaps it ramps up somewhat because of that. ARR arrives ahead of revenue. Just keep that in mind in that regard. I'd say we're comfortably inside the usual expectations, what we had. The remaining considerations truly aren't a factor at this point. Naturally, we'll share more on that topic down the road.
Analyst
analystHi everyone, Zach here, filling in for Shrenik. Appreciate you letting us ask. It's really encouraging to watch NRR climb back to 114%, with 59% of customers at four or more products and 29% at six or more. Your portfolio still includes over 25 offerings. So my question: how do we frame the realistic ceiling on NRR, particularly once AI security, data security, SD-WAN, and the other modules reach maturity? With a portfolio that wide, is there a route to expansion that stays sustainably elevated? Or, as enterprise scale grows, does that naturally hold NRR down even when dollar expansion is stronger? Thank you.
Sanjay Beri
executiveYeah. From where I sit on NRR, as we've said in the past, it can move around from one quarter to the next. That band we're sitting in, correct? We've got 113%, 114%, 115%. That band, which we've observed across prior quarters, matches what we've historically seen. We don't provide guidance on NRR, but qualitatively, I'd say that since the typical customer runs four or five products, we retain substantial room to upsell across many, many, many years. Customers and enterprises, as you're aware, with a broad platform such as ours, frequently begin with one or two core use cases, then expand in the following year and the year after that. So what we've built through the platform and the count of products feels like a durable platform that will scale alongside them. AI security will certainly contribute to that NRR, on differing timelines across customers and verticals. For us, though, it's a major pillar, and we feel quite good about where we stand there, and that will help us push expansion.
Aidan Perry
analystHey, this is Aidan sitting in for Rob O., and I appreciate you taking my question. I know it might still be early days, but could you share how customers are reacting to the transaction-based pricing on new AI products, now that deployments are beginning to scale, and are usage levels backing up bigger commitments?
Sanjay Beri
executiveYeah. Regarding the transaction-based pricing piece, if you consider the way we set prices, let's use the Agentic Broker as an illustration. An agent isn't a user. We aim to price in a manner that aligns with what we're truly doing. If you look at the Agentic Broker, it's overseeing agent transactions, so our pricing is per transaction. I believe much of what we'll encounter on the internet will be non-human, and because of that, maybe user pricing, even purely on a subjective level, wouldn't be sensible. From our standpoint, I think people understand it. They grasp that transaction-based pricing is logical. That's how agents operate. They're accustomed to, say, pricing tokens. Transactions are essentially a prompt plus a response. What we've been working to do is ensure they have visibility into this, so they can observe how many transactions are taking place. Consequently, they get a feeling of, all right, what will that look like when I make a purchase? I think the crucial thing with your customers is simply to ensure you're transparent, you're providing a means to view it, and then set prices based on what's logical from a usage standpoint. I think the transaction model has been well received. You've seen a few of our other offerings, such as AgentSkope, which is our AI agents, distinct from AI security. We began rolling out some of our AI agents, like our DLP AISecOps Agent. That one leans more toward outcome-based pricing; it depends on, for instance, how many cases out of the thousands or millions of DLP incidents do we generate and locate that needle in the haystack for you, and that's the outcome you're after. As for us, we're dedicated to the models of transaction-based and, for AgentSkope, outcome-based pricing.
Eshaan Shetty
analystHi there, Eshaan here, filling in for Eric Heath. Thanks once more for the question, and sorry ahead of time about the noise behind me. Sanjay, what's your take on how the SASE competitive landscape looks right now? Specifically, does demand at present seem strong enough that several scaled vendors can coexist over the long haul? And then maybe one quick follow-up: when competitive takeoffs happen, what are the main reasons customers pick Netskope instead of rivals? On the flip side, where do competitors manage to beat Netskope? Thanks again for the question, and congratulations on the quarter.
Sanjay Beri
executiveYeah, thank you. If you examine SASE, you'll find we offer 25 press products. Those span anything from managing databases in the cloud and on-prem, to cloud firewalls, to digital experience management, to enterprise browsers. I mean, the definition of SASE just keeps getting broader. An increasing number of things are being lumped into it. As a matter of fact, what was once around 20 or 30 separate vendors is now getting rolled up into SASE. So honestly, how I usually frame it is this: data and network security represented one of the largest markets within security and networking. In the past, you'd buy appliances, boxes, different kinds of data protection tools, assorted VPNs, and edge firewalls. All of those things are now being brought together and streamlined for simplicity, updated from a security standpoint, into SASE. So the answer is definitely yes. SASE, in my view, is a lasting market over the long term, and it can accommodate several vendors, particularly because what you're really doing is merging so many different pieces. Beyond that, consider what we've been talking about regarding what's known as AI security. Now, is that simply a subset of SASE? Or does it represent an entirely separate market? We view this as a foundational element that actually sits beyond SASE. In our case, we recognize that SASE represents a highly resilient, extended, attractive CAGR opportunity. We hold a leading position there. That was evident across every analyst report. The scope of SASE continues to expand, meaning our TAM increases as SASE progressively absorbs additional markets. On top of that, we've moved into AI security, which opens up an entirely new and enormous TAM. And then there's AgentSkope, our AI agent. So honestly, we have no shortage of TAM for a very long time. To be honest, as we've always said, when it comes to security and networking, the majority of CIOs we speak with aren't looking for a single platform covering everything in those areas. That's just not what they want. What they actually want is a handful of open platforms, and that's exactly our commitment — building an open platform that brings many different systems together while still integrating with the rest, such as your EDR, as we demonstrated with the CrowdStrike announcement, including some fresh integrations and more to come. So both currently and going forward, we have a lot of confidence about that.
Nolan Jenevein
analystHello, Nolan Jenevein here, filling in for Ittai Kidron. I appreciate you letting me ask. I'd like to dig a bit deeper into what was mentioned regarding the sales force ramping. I believe you mentioned before that around 50% were ramped at this stage, and you expect that figure to get better as the year goes on. Could you perhaps verify whether that represents a sequential rise when we consider the proportion of ramped reps? Any additional detail would be helpful. Thank you.
Sanjay Beri
executiveDuring the back half of the year — if you look back at last year and near its conclusion — we obviously began putting money into new reps and getting them ramped. That effort kicked off in EMEA and APJ, and you've witnessed the growth, clearly, in those areas. Afterward in NAM, since obviously we were completing some of the leadership pieces in NAM for the next level of scale last year; that is going to lead to a greater count of fully ramped reps. That's likely a superior way to frame it. As time passes, just keep expanding our capacity. Beyond that, we also made an announcement, for instance, on the earnings call, a moment ago, about the Catalyst program for managed services; that's simply one more indication of where we're persisting and building our partnerships too. Yes, we're adding more feet on the street along with more reps and more SEs, and we're ramping them, and that capacity becomes active later in the year and next year, yet we're likewise broadening our partnerships, and that matters greatly to us. The AI partnerships, Anthropic, as an example. We discussed the Amazon partnership. We discussed the NVIDIA Open Secure AI Alliance. We discussed too our partnerships with MSPs, SPs, SIs. Every one of those is also a major component of our strategy and plan as we ramp and grow our go-to-market team; given such a strong win rate, it's only natural that we do so.
Michelle Spolver
executiveThanks, and I appreciate all of you being here with us today. Our second-quarter performance has us feeling good, and we like the momentum building throughout the business. Our attention stays on assisting enterprises as they move through cloud and AI transformation, on keeping up steady innovation across our strong platform, and on delivering against a major opportunity that lies in front of us. We're grateful for your ongoing backing, and we hope to talk with many of you in the weeks and months ahead. That's all from us, so we'll end the call now. Thanks once more.
Operator
operatorThat concludes today's conference call. Thank you for your participation. You may now disconnect.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Netskope, Inc. transcript — plus 132,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.
Get the API View API docs →This call discussed
For developers and AI pipelines
Programmatic access to Netskope, Inc. earnings transcripts and 132,000+ others is available through the
EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments,
full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.
More tools for stock research
OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in Netskope, Inc. or any ticker on your watchlist.
StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.