管理層發言
Good day, everyone, and welcome to Palo Alto Networks Fiscal Third Quarter 2025 Earnings Conference Call. I am Hamza Fodderwala, Senior Vice President of Investor Relations and Strategic Finance. Please note that this call is being recorded today, Tuesday, May 20, 2025, at 01:30 PM Pacific Time. With me on today's call to discuss our fiscal third quarter results are Nikesh Arora, our Chairman and Chief Executive Officer, and Dipak Golechha, our Chief Financial Officer. Following our prepared remarks, Lee Klarich, our Chief Product Officer, will join us for the question-and-answer portion. You can find the press release and other information to supplement today's discussion on our website at investors.paloaltonetworks.com. While there, please click on the link for quarterly results to find the Q3 '25 supplemental information and Q3 '25 earnings presentation. During the course of today's call, we'll be making forward-looking statements and projections regarding the company's business operations and financial performance.
These statements made today are subject to a number of risks and uncertainties that could cause our actual results to differ from these forward-looking statements. Please review our press release and recent SEC filings for a description of these risks and uncertainties. We assume no obligation to update any forward-looking statements made in the presentation today. This presentation contains non-GAAP financial measures and key metrics relating to the company's past and expected future performance. Non-GAAP financial measures should not be considered a substitute for financial measures prepared in accordance with GAAP. The most directly comparable GAAP financial measures and reconciliations are in the press release and the appendix of the investor presentation. Unless specifically noted otherwise, all results and comparisons are on a fiscal year-over-year basis. We also note that management is scheduled to participate in the Bank of America Technology Conference this quarter. I will now turn the call over to Nikesh.
Thank you, Hamza. Good afternoon, everyone, and thank you for joining us for our earnings call. I'm delighted with our Q3 results. We continue to make progress on our platformization strategy while releasing a number of unique innovations in Q3 that set the pace for our industry. It is becoming increasingly clear that as organizations aspire to simplify and modernize their security architectures in the age of AI with data at the center, our strategy is resonating, resulting in larger deals. Most notably, we crossed an important milestone of $5 billion in next-generation security ARR, up 34% year-over-year, delivering industry-leading growth at scale. Our net new ARR growth also showed positive momentum in the third quarter. We believe we've reached an inflection point in our next-generation security story, as a growing majority of our incremental growth this year is derived from our AI-powered XSIAM, SASE, and software firewalls.
These offerings with large TAMs should help underpin your confidence in the sustainability of our NGS ARR growth as we march towards our $15 billion ARR target for FY '30. On the profitability front, Q3 continues to show the leverage in our business model. As we drive efficiencies from our scale and benefit from the economies of larger multi-product deals in addition to continuing to drive AI efficiency benefits slowly and steadily. We also generated healthy free cash flow while continuing to manage our transition from billing focus to RPO. As such, we remain confident in achieving our adjusted free cash flow margin targets over the next few years. Let's be clear, you can't walk around a street corner or conference without hearing the words AI. The urgency to adopt AI is omnipresent in all of our customers. It no longer seems to be a choice. It's becoming a strategic imperative for every customer as the risk of inaction is too high.
During every conference, every customer conversation, the topic of AI transformation is more and more frequent, and now the conversation is shifting to agentic AI. What's fascinating is this is actually creating a higher sense of urgency amongst our customers to undertake their technology transformation. Transformations that require fundamental change in their infrastructure. Traditional IT architectures weren't built for the scale, speed or complexity of AI. To truly capitalize on AI's potential, enterprises need modern, cloud-delivered platforms that can ingest vast amounts of data and operate in real time at scale. We've seen customers who were previously delaying their cloud migrations are now reaccelerating their investment. This is good news for cybersecurity. And as AI becomes more deeply integrated into our customers' businesses, the need to protect the underlying data, models, and infrastructure will become paramount.
Over the next year, an estimated more than $300 billion will be spent on AI infrastructure alone. That kind of spend doesn't just power models. You saw the video we opened with Glean and we're hard at work enabling secure adoption of this next-wave of AI-powered applications. This is precisely why industry must change the paradigm, shifting away from today's fragmented security landscape and towards consolidation. The cause of fragmentation is friction. Friction causes latency. Latency is the enemy of real-time cybersecurity. Now more than ever, bringing data together into a unified platform is critical. At its core, security is a data problem. We believe our depth and breadth of data is amongst the largest in the industry and is something that point solutions simply cannot access. This leads to superior threat detection efficacy, reduced false positives, and faster incident response times.
All critical metrics for our customers are key drivers of our market leadership. The volume and complexity of threats are not slowing down either. Bad actors are using AI to move faster than ever. Recently, our Unit 42 team was able to simulate an entire ransomware attack in under 25 minutes using AI at every stage of the attack chain. That's a staggering increase in speed, powered entirely by AI. Over a year ago, we doubled down on our platformization strategy. We're pleased with the large deal momentum we've had since and the endorsement of our strategy broadly across the industry. Our platform approach is working well with customers slowly and steadily. Our approach, which favors better and speedier security outcomes and lower cost of ownership, is being adopted by more and more of our customers. In Q3, we delivered over 90 net new platformization deals and now have a total of approximately 1,250 platformizations within our top 5,000 customers.
Digging deeper, the number of customers with multiple platformizations grew nearly 70% year-on-year in Q3. In particular, the number of customers platformized on Cortex was up nearly three times, reflecting strong momentum with XSIAM. The overall growth in largest customers also reinforces our success. We had 130 customers with over $5 million in NGS ARR in Q3, up over 40% year-over-year, and 44 customers with over $10 million in NGS ARR, up over 60%. To give more color on what these platformizations look like, I want to take a look at a few examples from Q3. Of particular note, beyond the size and scope of our deals, is the customer's ability to consolidate a significant number of products with Palo Alto Networks. A leading global consulting firm signed a transaction worth over $90 million in Q3. This customer platformized on Cortex for XSIAM, replacing a legacy incumbent SIEM provider.
Our differentiated value was centered in our ability to materially reduce mean time to respond. We also reduced cost by consolidating a total of four products in this deal. As a result, our NGS ARR of this customer nearly doubled year-over-year. A leading financial services company signed a $46 million transaction with us. The customer recognized the value of XSIAM and consolidated four products with us, including the displacement of their well-established EDR and SIEM vendors. Beyond XSIAM, the customer also expanded and extended their platformization on our network security platform in the quarter. A U.S. financial services firm signed a $33 million transaction with us. This customer platformized on network security and consolidated cloud security vendors, driven by a company mandate to consolidate their security tools and reduce complexity in their cybersecurity stack. In this deal, they consolidated four products.
Now, moving on to an update on Cortex. As you can tell from my earlier comments, I'm particularly excited about the momentum we're seeing with XSIAM, which saw accelerating growth in Q3. XSIAM is not only our fastest-growing product ever, it is now more impactful to our overall growth rate. I believe that from a strategic perspective, XSIAM has the potential of being the game changer for both the industry and Palo Alto Networks. In the first innings of baseball, not cricket, of transforming the cybersecurity industry with XSIAM. By consolidating security data into a single AI-driven SOC platform, XSIAM is modernizing and disrupting the traditional SIEM market or continuing to see amazing milestones, including customers' mean time to respond from weeks to minutes. As security teams face growing complexity and talent shortages, we believe XSIAM is well positioned to be an operating system for modern SecOps.
The numbers speak for themselves. We now have approximately 270 customers in XSIAM, and the average ARR per customer is over $1 million. This already makes it one of the most successful products in the history of cybersecurity. What's even more remarkable is that we've reached this level of adoption and impact just 30 months after XSIAM was made generally available to customers. XSIAM ARR grew over 200% year-over-year in Q3, nearly twice as fast as our closest next-generation SIEM competitor. On a trailing 12-month basis, XSIAM bookings are now approaching $1 billion. About three years into our XSIAM journey, our sustained strong momentum bolsters our confidence in a long growth runway as we increasingly tap into this estimated $40 billion SecOps TAM. Last quarter, we unveiled Cortex Cloud, our breakthrough in unifying cloud posture and SOC operations. Over the last three months, we have seen strong early customer interest in Cortex Cloud, with the nine-figure pipeline, spanning hundreds of customers.
This quarter, we also announced two products that will enhance our ability to further expand XSIAM and its capability. XSIAM, once deployed, has become the foundational security data platform for our customers. We now understand the data we're capturing in this is the data you actually need for a whole variety of use cases. In April, we launched advanced email security to help stop threats before they reach the inbox and our exposure management capability was launched as well designed to cut through the noise and focus security on the risks that truly matter. But that's not all. Think of XSIAM as our data-to-market engine. Every byte on nearly 12 petabytes of telemetry we ingest daily around cloud identity endpoints in email and more acts as high-octane fuel. This massive data stream isn't just powering XSIAM, it's igniting our ability to identify and accelerate our entry into entirely new markets, unlocking additional TAM in the tens of billions we are now uniquely positioned to address.
Through our comprehensive understanding of data sources and broad data ingestion capabilities, we're beginning to deliver solutions using multiple content capabilities of XSIAM. Every piece of telemetry we ingest makes our platform smarter. The more data we put in, the further our engine can go, and the smarter it gets, the faster we can build entirely new capabilities on top of it. We're encouraged by the early customer feedback and look forward to continuing to discuss this more in the future. Now, shifting our focus to network security. We continue to lead the market in network security and gained share across all three of our best-to-breed form factors. As enterprises look to securely and increasingly secure hybrid workforces and IT environments spanning headquarters, branch offices, data centers on the cloud, we're uniquely positioned with a consistent security architecture. In Q3, our product revenue grew 16% year-over-year.
This growth was broad-based. With software continuing to increase in the overall mix, we also saw stable demand in the appliance market. Software firewall ARR grew approximately 20% year-over-year in Q3 with public cloud deployments continuing to be the primary driver. AI is accelerating cloud adoption, and we believe this trend will expand the long-term need for software firewalls that scale modern workloads. Shifting to SASE, which continues to be our fastest-growing form factor in network security and a strong contributor to our overall growth. As customers transform their networks to keep pace with delivering first-class security capabilities for remote users and branch offices, we continue to see robust growth for SASE. Many SASE projects are large and comprehensive, which is well-suited to our rich offering and enterprise-focused sales expertise. In Q3, our SASE ARR grew 36% year-over-year, more than twice as fast as the overall market and ahead of our key SASE competitors.
Furthermore, 40% of new SASE customers were net new to Palo Alto Networks in Q3. We now have approximately 6,000 SASE customers, up 22% year-over-year. Meanwhile, the drivers of our SASE momentum are broadening. This quarter, we saw particular strength in Prisma Access Browser, which again accounted for a third of our Prisma Access seats sold in the quarter. In just 18 months since our talent acquisition, we have now sold approximately 3 million license seats on Prisma Access Browser, up more than 10x from a year ago, and we have a healthy nine-figure pipeline. As AI drives more data and applications to the cloud, the browser is becoming the primary interface to accessing these resources, acting as the application runtime environment. The operating system in this scenario becomes less about local resources and more about securely connecting to and managing cloud-based services. And as more and more critical applications of data reside within the browser environment, it naturally becomes the target of cyberattacks.
Prisma Access Browser's native controls and real-time visibility are designed to help ensure that sensitive data remains safeguarded during browsing sessions regardless of the user's location or the application they're accessing. And we believe Prisma Access Browser is strategically positioned to be the future OS in enabling secure and productive work in an evolving AI-driven world. Now, shifting to our newly launched Prisma AIRS, or AI Runtime Security. As I mentioned earlier, it's more important than ever to bring data together in order to leverage AI, enabling customers to stay ahead of the attackers. We're also seeing customers demanding that we help them secure their AI transformation journey. In this mad rush for AI in the industry, many of our customers are experimenting with AI. At Palo Alto Networks itself, our teams are leveraging over 35 models across multiple products, each of which, and the AI artifacts need to be discovered, scanned, constantly tested, and protected against.
Prisma AIRS allows for just that. It helps enterprises discover, scan, and test all the AI artifacts to ensure they're safe. It allows for a world-class data security posture deployment, and once in production, it ensures that applications using AI are constantly monitored and any security flaws are both protected against as well as remediated across the enterprise. Prisma AIRS extends our existing capability in posture management and runtime security and will add security for AI agents in the future. And we announced the intent to acquire Protect.ai, an early innovative leader in security for AI providing AI model scanning in the red teaming to further bolster our capabilities. Customer interest has been strong. We currently are in conversation with hundreds of prospects and already have an eight-figure pipeline since making the announcement last month. In summary, we see strong momentum heading into our fiscal year-end, driven by continued transformation, and we look for our first north of $4 billion quarter.
We see strong desire for consolidation and the desire to implement AI securely, including a robust Q4 pipeline. We continue to take share across multiple security categories driving strong growth in NGS ARR at an industry-leading scale. Our platformization strategy translates into tangible business benefit for customers, including a strong security posture and improved operational efficiency through vendor consolidation. As our relentless focus on innovation, we believe Palo Alto Networks is the ideal partner to help organizations achieve and secure their AI transformation goals. I'm particularly proud of our teams for driving phenomenal success in Q3, which was fraught with geopolitical discussions, tariff discussions yet our teams kept our heads down and continued to execute, setting us up for what we hope will be a great Q4.
Thank you, Nikesh, and good afternoon, everyone. To maximize our time spent on Q&A, I will provide you with highlights of Q3. You can review the detailed results in our press release and the supplemental financial information on our website. In Q3, total revenue was $2.29 billion and grew 15% at the high end of our guided range. Within total revenue, product revenue grew 16%, while total services revenue grew 15%. Within total services, subscription revenue grew 18% and support revenue rose 10%. On a trailing 12-month basis, the proportion of our product revenue from software is approaching 40%, driven by our growth in our virtual form factors and SD-WAN. We continue to see stable demand for firewall appliances with market growth in the 0% to 5% range as we have discussed previously. Moving on to geographies, we saw double-digit growth across all theaters with the Americas growing 12%, EMEA up 20%, and JAPAC growing 23%.
Our remaining performance obligation, or RPO, grew 19% to $13.5 billion. Our current RPO was $6.2 billion, growing 16% year-on-year. The average duration of new contracts remained at approximately three years. Contract duration decreased slightly on both year-over-year and quarter-over-quarter basis. Customers continue to make significant commitments to Palo Alto Networks through our platformization deals, particularly when adopting XSIAM to transform their security operations center. We continue to see increasing demand for annual payments, particularly deals over $1 million, but we are absorbing this transition while maintaining the high end of our fiscal year '25 annual adjusted free cash flow margin guidance, as well as reiterating confidence in our adjusted free cash flow margin targets of 37%-plus in fiscal year '26 and '27. In line with what we talked about earlier in the year, we saw a year-over-year increase in bookings that went into annual billings in Q3 and a decrease in deals leveraging PAN-FS with a neutral impact on cash flow.
Turning to next-generation security ARR. As Nikesh highlighted, we surpassed the $5 billion mark in Q3 and ended the quarter at $5.09 billion in NGS ARR, a growth of 34%. Within NGS ARR, we continue to see significant momentum around our Cortex platform and our AI ARR is now approximately $400 million in Q3, up over 2.5 times year-over-year. I'm particularly excited about trends driving the NGS ARR, and I wanted to provide some additional insights around the evolution of our net new NGS ARR. We've made a number of significant investments over the last several years to both continue to lead the network security market as well as build leadership positions in new markets. You've seen the results of this effort in our NGS ARR. Over the last several years, in network security, we invested in advanced cloud-delivered versions of our subscriptions that attach to our appliances. We saw strong adoption of these advanced subscriptions as customers saw the value of adding these to their existing network security deployments, and this has driven meaningful NGS ARR growth.
In addition, we have NGS product offerings beyond those advanced subscriptions in new markets across network security, cloud security, and security operations. We refer to those as our new market offerings as they also fueled our NGS ARR growth. We continue to see momentum from our advanced subscriptions in fiscal year '25, driving a healthy and relatively consistent level of net new ARR compared to prior years. At the same time, we've seen our net new NGS ARR from new market offerings grow significantly and these are becoming a larger proportion of our total net new NGS ARR dollars. This is a result of our strengthened position in these markets. Our large sales team becoming more adept at selling these offerings and platformization taking hold with our customers and in the industry. As we look forward, we expect to see the new market business to be the stronger driver of net new ARR dollars.
It is this dynamic that gives us confidence in our long-term targets. Moving down the income statement. Total gross margin was 76%. Product gross margin was 78.4% in the quarter. As a reminder, we have been transitioning to a contract manufacturing facility in Texas as our primary manufacturing and fulfillment center to benefit from scale and innovation, as well as to take advantage of a foreign trade zone that can help us mitigate tariffs in products that we ship to international destinations. As we said on our previous call, we continue to believe that we differentiate ourselves by being the only pure-play cybersecurity firm at scale to assemble all of our hardware in the USA. As a result, tariff impact to our business has been immaterial. We expect product gross margins to remain in the high 70% or low 80% margins in Q4. Our total services gross margin was 75.4%. We are excited to see continued strong adoption of our SaaS offerings.
We continue to execute on cloud cost efficiencies, including engaging with our key cloud service providers to negotiate favorable procurement arrangements as the scale of our cloud-hosted products continues to increase. Encompassing those gross margin dynamics, we continue to focus on executing our operating margin targets, which delivered year-over-year improvements in operating margin. As I have often said, our business scales well across every single line item of the P&L. On an operating expense as a percentage of revenue basis, we saw 340 basis points of year-over-year leverage this quarter as we drove scale and efficiencies across sales and marketing, R&D and G&A. We delivered $0.80 of diluted non-GAAP EPS and diluted GAAP EPS of $0.37, our 12th consecutive quarter of positive GAAP EPS. We generated $578 million in adjusted free cash flow in Q3. Turning to the balance sheet, you will see that our debt balance came down by $151 million as we continue to see early conversion of our convertible debt, which occurred at the discretion of the debt holders and were settled by us in cash and equity.
As a reminder, our convertible notes reached final maturity in June and our convertible notes can no longer be early converted. We will settle the remaining convertible debt in cash and equity in Q4. As Nikesh mentioned, we announced our intention to acquire Protect AI for a total consideration of $700 million in cash and replacement equity awards. We expect the transaction to close by our first quarter of fiscal year 2026. We did not repurchase any shares in Q3 and our buyback strategy remains opportunistic. We have $1 billion in authorization remaining through December 2025. With that, let me turn to guidance. For the fiscal year 2025, we expect NGS ARR to be in the range of $5.52 billion to $5.57 billion, an increase of 31% to 32%; remaining performance obligation of $15.2 billion to $15.3 billion, an increase of 19% to 20%; revenue to be in the range of $9.17 billion to $9.19 billion, an increase of 14%; operating margins to be in the range of 28.2% to 28.5%; our diluted non-GAAP EPS to be in the range of $3.26 to $3.28 per share, an increase of 15%; adjusted free cash flow margin in the range of 37.5% to 38%.
As we noted last quarter, we do expect a higher Q4 contribution to our annual free cash flow. And for Q4, specifically, 80% of the collections are from deals that have already been booked. Our annual cash flow seasonality is more second half and Q4 weighted this year, influenced by the timing of deferred payments from customers that signed deals in prior periods and the timing of bookings within the year. For the fourth fiscal quarter of 2025, we expect NGS ARR to be in the range of $5.52 billion to $5.57 billion, an increase of 31% to 32%; remaining performance obligation of $15.2 billion to $15.3 billion, an increase of 19% to 20%; revenue to be in the range of $2.49 billion to $2.51 billion, an increase of 14% to 15%; and diluted non-GAAP EPS to be in the range of $0.87 to $0.89, an increase of 16% to 19%. We've included our typical modeling points in the presentation for your review.
分析師問答
Okay. Great. Thank you. To allow for broad participation, I would ask that each analyst ask only one question. The first question will come from Saket Kalia followed by Brian Essex of JPMorgan.
Okay. Great. Hey, guys. Thanks for taking my questions here, and congrats Hamza on the move over. Nikesh, maybe for you. Lots of things.
There's life after, you can see.
Absolutely. Nikesh, maybe for you. Lots of things to be excited about with XSIAM. I wanted to dig into one part of that opportunity, in particular, which is the QRadar on-premise customer base. Clearly, a big base there that you can upgrade. Maybe the question is how are customers thinking about that upgrade? And how big of an ARR opportunity could that be for Palo Alto Networks on that path to $15 billion?
Well, Saket, first of all, thank you for the question. Look, I think we had this debate last quarter, how big is XSIAM going to be versus Prisma Access Browser, because we believe they are both new trends in the industry, primarily driven by the AI wave we're seeing around us. And whilst we started the XSIAM journey, and then we did the deal with IBM, where the first set of conversion we saw over from QRadar where people who already adapted SaaS, already in the sort of cloud-delivered SIEM market, we saw that conversion. But we've had a phenomenal partnership with IBM where we have been able to go with them to many of their large customers and work them through the transformation of going from an on-premise SOC to what is effectively now a cloud-delivered SOC. So, the QRadar on-prem transition is not just moving QRadar to XSIAM; it's also moving from an on-prem delivered solution to a cloud-based solution.
And clearly, you've seen the ARR uplifts. We announced the large deal, one of the largest deals, a $90 million deal, which had XSIAM component to it. The second largest deal we had also had XSIAM component. So, clearly, large deals on XSIAM are possible. I mean, think about it, you've been covering security for a long time, which is the last product that came out where the average ARR was $1 million a year. Right? There are companies out there in security who started and claim they have $10 million customers or you say every customer in XSIAM is $1 million ARR customer. So, I think the opportunity is huge. If you go back historically, and I've said this in cybersecurity, certain swim lanes reach inflection points where the next set of products are so much better that everybody has to be shaken out of their stupor and their old solutions to go replace. I mean, you saw that happen in the endpoint market where we had players who had to be replaced over time.
You saw that historically when we came out of the next-generation firewall, and I think this is the moment of the SIEM market. It's a $40 billion TAM. I think in the next three to five years, it will get replaced. It will be replaced by new age players. The legacy players will drive their hardest to hang on to it, but the architectures are fundamentally different and the architectures of yesteryears. And it's not their fault because we designed a product 17 years ago. It was designed where data was expensive to store. Latency was high. Things were done offline. Today, we live in a world with low latency. As you see, hyperscalers announcing real-time translation this morning, so you can see that we can process data at immense speeds and deliver results immensely. If you concatenate that with what I said, we were able to regenerate a SYN ransomware attack in 25 minutes, right? If that's the pace of the bad guys, the pace of the good guys has to be faster.
So, there's no way to get to the other side from an incident response, incident management perspective if you don't transform what is fundamentally a legacy technology. So, I think XSIAM has a huge potential, one, and I apologize for taking longer than I should for your answer, but I said this in my prepared remarks, I want to make sure I emphasize is that what we've discovered is once we ingest all the data in an enterprise, we can actually go and make peacetime products better. So, our launch of email is actually a multi-context launch. Email security products are traditionally swim lane products. I look at your email, I protect you. We're able to look past email in the SIEM and say, 'Wait, when somebody clicked on that email link, what happened?' So, we have the entire organizational context on the data that we have that allows us to go back and make the email product spectacularly better.
So, we think there are many, many new swim lanes which will get discovered as we deploy XSIAM with our customers. We've seen early examples of exposure management and email management. We think those are the first two use cases. But I think there's many use cases behind it. Eventually, we think something like XSIAM becomes the underlying security fabric of enterprises.
Very helpful. Thanks.
All right. Next, we have Brian Essex from JPMorgan, followed by Keith Weiss, Morgan Stanley.
Great. Thank you. Thank you, Hamza, and congrats on the move from me as well. Thank you for taking the question. I was wondering if you could unpack, Nikesh, the details behind the product revenue growth, really strong quarter of growth there. And I heard Dipak's comment that approaching 40% software in terms of mix. But I was wondering if maybe you can shed some light on how much of that is share shift, how much of that is pricing increase, and how much of that might be refresh.
Brian, our product revenue growth stems from both hardware and software, not just hardware alone. The core hardware business continues to grow at a steady rate of 5% to 8%, which I've consistently stated. We pointed out software firewalls in our remarks for a reason, as we believe this segment is poised for significant growth. In the last six months, there's been a noticeable shift related to AI; even those hesitant to adopt it are starting to explore their options. To be at the forefront of AI innovation, organizations need to leverage cloud solutions. On-premise AI implementations lag behind by around six months because there aren't equivalent offerings like Gemini, OpenAI, or Lambda for on-premises. While we can deploy Lambda in an on-prem environment, it requires advanced technical expertise to implement effectively. As a result, even the most cautious organizations must transition to the cloud to access the rapidly evolving AI models. This necessity to move to the cloud brings about the need for real-time security. Currently, we offer the best multi-cloud technology for securing cloud network traffic through our software firewalls. This shift is driving the growth of our software firewall business, supporting both the transition from hardware to software and enhancing our overall product revenue at Palo Alto Networks.
Next question will be Keith Weiss from Morgan Stanley, followed by Joe Gallo, Jefferies.
Excellent. Thank you...
Keith, why did he pick your third?
I know, that's a good question. That being said, Hamza, if you ever realize you made the wrong decision, there's always room for you back at Morgan Stanley. It's never too late. But congratulations, guys, on a really solid quarter. I wanted to kind of expand on that last question what you were talking about in terms of the AI imperative and what that opportunity means for Palo Alto Networks. When you're talking to customers that are looking to secure these new AI infrastructures, is this just about AIRS or is there a wider opportunity? What is it like pull-through from the Palo Alto portfolio when people are looking to secure this new surface area? And to what degree does this hope give you guys confidence in the growth in like the next-generation ARR into next year? That's where I hear the most concerns from investors that they're too aggressive for next year. Is it AI? Is it platformization? Like, what are the elements that give you confidence in that out-year forecast?
That's a great question, Keith. If we consider the software firewall example, our discussions now begin with the necessity to secure AI implementations. During a Board meeting last week, our InfoSec team shared the extent of AI usage within Palo Alto. I was surprised to discover that we currently have 35 models under evaluation and deployment. It's not just one model; we have 35, as they serve various purposes. When we mentioned this number, our Board became concerned about how these models would be secured, as they could come from anywhere. To secure these models, we effectively need to envelop them with a runtime firewall or AI firewall, which extends our software firewall capabilities. The AI firewall will enhance the software firewall, leading customers to lose the distinction between AI traffic and traditional cloud traffic. As a result, we are witnessing an accelerated adoption of cloud firewalls driven by the AI trend, which is influencing the transition from hardware to software. This shift, which has been gradually occurring over the past five or six years, is becoming more rapid, leading to a transition in our NGS ARR from hardware to software. This evolution will reduce our traditional ARR while increasing our NGS ARR, generating favorable conditions for NGS ARR growth from a software firewall perspective.
Excellent. Thank you, guys.
Thank you, Keith. Next up, we have Joe Gallo from Jefferies followed by Gabriela Borges from Goldman Sachs.
Hey, guys. Thanks for the question, and congrats, Hamza, on the new role. Nikesh, you briefly alluded to executing through geopolitical volatility. You're the first cyber name to report. Can you just elaborate on your conversations with CIOs, CSOs? Is it back to business as usual after the first couple of weeks of April, or is there still a lot of uncertainty? Maybe just kind of talk through the pipeline conversations and what's embedded in the 4Q guide.
That's a great question, Joe. The reason I didn't focus on it in our prepared remarks is that April was an unusual month. I believe we are returning to normal, but there were recent discussions about tariffs globally and various anticipated supply chain disruptions. This led some of our customers to feel uncertain about their shipping costs and overall operations. You could see this uncertainty in the market during the last month of our quarter. I'm particularly pleased that our teams were able to remain focused and execute effectively, as it was a challenging quarter. Without the tariff discussions and geopolitical tensions, it would have been easier to sell through the quarter. We learned valuable lessons from the pandemic and the supply chain crisis, which helped us to adapt our strategies. We're now aligned in the same pacing as we approach Q4, staying ahead of the curve. While many customers haven’t changed their transformation plans, there was a brief pause as they assessed the market direction. Fortunately, it seems we've moved past that uncertainty as a global economy, and there is now a little more stability in the business climate compared to early to mid-April. I'm very proud of how our teams executed during that time.
Great. Next up, we have Gabriela Borges from Goldman Sachs, followed by Matt Hedberg from RBC.
Hey. Good afternoon. Thank you. Nikesh, I wanted to follow-up on your comment on learning your lesson from cloud. Maybe just a little bit more on how you think about the AI product portfolio evolving from here. How do you think about what this looks like a couple of years out, the mix between organic and inorganic? And on your point of being sideswiped, how do you think about insulating yourself from being leapfrogged in AI technology specifically given how quickly the technology is evolving? Thanks.
Thank you for the question, Gabriela. Looking back five or six years ago, when the cloud security evolution began, there was a significant emphasis on cloud posture. Companies were focused on identifying issues, understanding misconfigurations, and learning how to set up platforms like GCP and AWS. Many companies emerged during that time, including Dome9, Evident, RedLock, and Twistlock, which were eventually integrated into our core cloud offerings. However, there was comparatively less focus on runtime security as many were still experimenting with cloud deployments. We may have underestimated the importance of runtime scenarios, which have now gained traction as the market has matured. Currently, we observe more activity in cloud detection and response and cloud SIEMs than in cloud posture, which is becoming commoditized as every vendor offers some form of cloud posture module, resulting in rationalized pricing across the industry.
In AI, we anticipate an accelerated version of this trend as the technology evolves faster than any previous transformation. We were fortunate to have implemented native firewalls with all major public cloud providers, which allows us to monitor AI transactions in real time, as most AI applications are hosted on cloud services rather than on-premises. Consequently, it’s crucial to analyze this bidirectional traffic because AI systems will operate with increasing autonomy, potentially leading to unpredictable behavior. This makes AI runtime security a critical aspect of our strategy. Our team has made significant progress on our AI firewall, but we also recognized the need to assess the reliability of AI models actively. Customers are asking for model scanning and ongoing red teaming, and we realized that developing this capability could take months. Instead, we sought out the best solutions in the market and discovered Protect, whose founders aligned well with our team’s vision. We are moving forward with the acquisition, and that team will lead our AI security business.
Excellent. Thank you.
Thank you, Hamza, for your question. Congratulations on the results, everyone. I have a question for Lee. While at RSA, it increasingly felt like it was focused on AI rather than cybersecurity. You mentioned $400 million in AI annual recurring revenue, which is impressive. Can you explain why Palo Alto is well-positioned to assist customers in navigating this new paradigm shift? Additionally, what foundational elements do you believe are strong from an agentic perspective?
Great question. Two years ago, RSA focused on AI, last year it was AI security, and this year it's agentic AI. We're observing a progression in the different forms of AI. With agentic AI, I see it as a transition from AI serving as a helper to gaining a level of autonomy, where it can take actions independently. Over time, as trust is established, these actions will be regulated and gradually expanded. There are essential building blocks that we need to consider; it's not about allowing AI to operate without limits. There will need to be guardrails and constraints from both an enterprise and cybersecurity viewpoint. Our strength lies in our extensive capabilities related to automation. Our XSOAR platform boasts over a thousand integrations, demonstrating our expertise in connecting with various enterprise systems in a reliable manner. By integrating an AI engine, which embodies the agentic aspect, we can create, evolve, and improve over time.
No one else possesses the same level of integration and automation that can effectively work alongside AI. Additionally, a major concern I've heard from customers is about the permission infrastructure necessary for these agentic systems to operate within defined limits. Our understanding of how different enterprise applications interconnect positions us well to determine appropriate permissions and understand the associated security implications. We've shared our initiatives in two main areas: what we're doing on Prisma AIRS to secure agentic systems and a preview of agentics, outlining our strategy for developing agentic platforms for our customers. I'm excited about both initiatives.
Thanks, guys.
Thank you, Matt. Next up, we have Shaul Eyal from Cowen, followed by Jonathan Ho from William Blair.
Thank you. Good afternoon, guys. Congrats on solid results. Congrats to Hamza on leading his first call. Nikesh, back to QRadar and Talon, and I get these are two different products at their core. Both seem to be exceeding their targets internally, both should continue to show great results. From where you sit, would you rank them as Palo Alto's probably better acquisitions over the course of the past four or five years? Where is the bigger TAM opportunity longer term?
I think if you step back and you think five years out, I think five years out, security will be delivered on top of a large security data lake by some version of machine learning and AI. I think the traditional approach to security of writing policies and writing rules and writing human sort of controls at the edge is going to fail, because AI will be incessantly banging at those edges trying to find the loopholes and the misconfiguration in the way humans have not secured the perimeter. So, if you believe that and you say that the way to make sure that your perimeter is secured is look at all the data, analyze it, pattern recognize it, and have your agents, the same agents that Matt was asking about. Those agents should go back and say, 'I found a misconfiguration at the edge. Let me go fix it,' right? Why should that not happen? Why should security in the future be some security analyst sitting and writing a policy pane saying do this for traffic and do this for HTTPS, et cetera?
So, by definition, it means you should have some version of agentic AI securing your perimeter five years from now. That requires that agentic AI to have the access to all the data across all your endpoints, right? It doesn't mean you have to have the same endpoint. It doesn't mean you have to have Prisma Access Browser or SASE or Palo Alto SASE. You can have anything. You can have Zscaler, Netskope, Palo Alto, whoever you want. But at the end of the day, the data has to get to one place. Now what's happened is we've been collecting data recently in a new paradigm for the incident response use case. Well, guess what? That same data can be used for the policy remediation use case or protect your perimeter use case. So, I think the evolution of security will be people will go towards harmonized security data lakes, which will be used for multiple activities. And I think we're seeing the beginning of that.
And I think three to five years from now, you look back and say, that was obvious. It's not obvious today because people are still going to sell best-of-breed products at the perimeter to protect it. But how can you protect the perimeter if you only have a singular lens on the problem? If you only see what you see with your product, how can you protect the perimeter? You have to see the entire enterprise to understand where the relative opportunities, where the relative mistakes are. And I think that's where security goes in two to five years. So, in that context, I think Prisma Access Browser is a great edge device, which will actually be very interesting and useful because you have full visibility of the interactions of the endpoint or the human perhaps or the agent perhaps in that use case, but you're going to need something in the back to try and make sense of that data and be able to respond in real time. And I think that's where the underpinning of the next-generation SIEMs of the world is going to be.
All right. Thank you, Shaul. Next up, we have Jonathan Ho from William Blair.
Hey, guys. Can you talk a little bit about the enterprise browser and why Talon has seen the inflection that it has? And does this perhaps help your platform approach, particularly around the AI opportunity? Thank you.
Sure. The key aspect to consider regarding the secure browser is that it serves a specific use case when it operates independently. A user browsing the Internet can have their activity secured. However, once we integrate it within our platform, it evolves from a standalone tool to a versatile solution. We can secure users' Internet traffic, their connections to SaaS, and private applications. With our recent VDI integration, we can even secure non-browser based connections. What I described is made possible by offering it as part of a comprehensive platform rather than in isolation. We can provide standalone solutions, and some customers are beginning with that as a first step towards broader adoption, which showcases the flexibility of our platforms. Ultimately, this solution is driving the inflection point we are observing. Regarding the role of AI, through the secure browser, we are able to present a unique solution for securely adopting AI that enhances both security and user experience. By embedding many security controls directly in the browser, we simplify the adoption process for users.
Sorry about that, Jonathan. Next up, we have Joel Fishbein from Truist, followed by Andy Nowinski from Wells Fargo.
Thanks for the question, and Hamza, congrats. Nikesh, I guess, a follow-up to Shaul's question or your answer to Shaul's question, pretty powerful statement about the data lake security. I'm curious about how you view threat intelligence and the importance of sharing threat intelligence in this AI era and protecting against the bad guys. Love to hear your thoughts on that.
We believe that sharing threat intelligence is generally important, though there are exceptions. We've been open about this and are founding members of the Cyber Threat Alliance, which promotes sharing among security vendors, including our competitors. The rise of AI, especially in attacks, will likely change this landscape. Much of threat intelligence has focused on specific attacks, malware, or indicators of compromise to build protections. However, as AI evolves, we can expect more novel attacks that are different from past ones, making AI-based protections crucial. While these AI protections will still be informed by threat intelligence, the relationship will shift. It will become vital for shared threat intelligence to focus more on attack techniques and methodologies, necessitating changes in the threat intelligence field. It will be interesting to observe how this transformation unfolds.
Thank you.
Okay. Thank you for squeezing me in, and I'll make it a good one here. You guys had solid growth in platformization customers. I think you now have 1,250 that have deployed a platform. That's a small percentage of your total installed base over 70,000 customers. So, I'm wondering, is there any way to look at the percentage of ARR that those platform customers account for? Because I would guess that they account for a much higher percentage of your ARR than the 2% they account for in your installed base. I'm really trying to understand the importance of the growth in these platform customers relative to reaching your $15 billion ARR target. Thank you.
So, let's recap, Andy. I think we had said we want to get to between 2,500 to 3,500 platformizations. And we believe when we get there, at the current trajectory of ARR per platform deals, we're going to get close to our $15 billion target. And that assumes somewhere between 60% to 70% of our NGS ARR is made up of these platform customers. Does that make sense?
Yeah. Certainly. So, okay.
That's our math. So, you can do the math now yourself and say, what does that mean? If 70% of $15 billion is not hard to compute. That's it. And then you can divide that by 2,500 or 3,500, whichever you like and see the average ARR per deal. And then, we'll tell you if you work that back 5%, 6% a year over the next five years, you'll say, wow, that must be the average platform deal that Palo Alto have, give or take.
Okay. Great. With that, we will conclude the Q&A portion of this call. I will now turn it back to Nikesh for his closing remarks.
Again, thank you everyone again for joining us for our earnings call. We look forward to seeing many of you at upcoming investor events. I also want to thank our customers, partners, and as I said, our employees for powering through what was a tumultuous April, but we think now it's business as usual. All of us are heads down trying to execute on our big $4-plus-billion quarter in Q4. Thanks again.