F5 (FFIV) 2026财年第四季度业绩电话会：产品收入增长23%，预计2027财年增长8%-9%lg...

核心要点

• F5公布2026财年第四季度营收同比增长13%，主要得益于产品营收增长23%的推动。全年营收达到34亿美元，同比增长11%，其中产品营收增长19%。
• 第四季度产品营收为5.10亿美元。系统营收增长29%至2.39亿美元，软件营收增长19%至2.71亿美元。
• 2026财年直接AI应用场景订单额突破1.00亿美元，同比增长近300%。活跃直接AI客户数量翻了两番，超过200家。
• 管理层预计2027财年营收增长为8%至9%，产品营收将连续一年实现两位数增长。预计软件业务将保持稳健的两位数增长。
• F5预计2027财年Non-GAAP营业利润率为33%至35%，Non-GAAP每股收益（EPS）为18.00美元至18.30美元。
• 公司批准新增15亿美元的股票回购额度，截至第四季度末，可用授权回购总额约为18亿美元。

关键财务业绩

会议纪要包含互相矛盾的第四季度总营收数据：CEO弗朗索瓦·洛科-多努（François Locoh-Donou）引用的数字为9.17亿美元，而CFO库珀·维纳（Cooper Werner）随后引用的数字为9.70亿美元。披露的产品和服务营收数据分别为5.10亿美元和4.08亿美元。

2026财年，软件营收增长8%至8.70亿美元。定期订阅增长14%至5.78亿美元，SaaS及托管服务营收增长3%至1.82亿美元，永久软件许可下降8%至1.11亿美元。

安全业务总营收增长10%至约13亿美元。产品安全营收增长15%至7.46亿美元。SaaS及托管服务的年度经常性营收（ARR）达到2.11亿美元，增长14%，其中核心SaaS及托管服务ARR增长24%。

业务与运营表现

F5已连续第九个季度录得产品营收两位数增长。管理层将这一需求归因于混合多云的应用采纳、扩展中的应用安全威胁态势，以及日益增长的AI推理工作负载。

分区域看，欧洲、中东及非洲（EMEA）地区领跑增长，营收增长31%，占总营收的30%。亚太地区（APAC）营收增长14%，占比为17%，美洲地区增长5%，贡献了53%。

企业客户贡献了第四季度产品订单额的69%。政府客户占18%（其中6%来自美国联邦政府客户），服务提供商贡献了13%。

平台采纳率继续扩大。在F5的前1000家客户中，目前有29%使用至少三个F5产品系列，而2025年这一比例为22%。管理层强调，客户正在将应用交付、Web应用安全、API保护和云服务整合到一个统一的策略与管理框架中。

安全需求依然是主要驱动力。超过300家客户在F5 AI驱动的WAF（Web应用防火墙）推出后六个月内完成了部署，且其中超过75%的部署运行在阻断模式下。今年3月推出的F5 Insight已拥有超过200家客户使用其集群可见性、升级编排和生命周期管理功能。

F5分布式云遥测记录显示，WAF请求量同比增长84%，API安全事件增长103%。管理层表示，客户正从零散的、基于特征码的安全产品转型为能够跨应用、API及AI工作负载实施运行时保护的统一平台。

AI相关需求在三大直接应用场景加速增长：AI数据交付、AI运行时安全和AI工厂负载均衡。2026财年直接AI订单额突破1.00亿美元，增长近300%。第四季度部署中包含29.2 Tbps由F5驱动的AI数据交付能力，同比增长近250%。

管理层预估，到2030年，F5的可寻址AI市场规模可能达到约110亿美元，其中包括90亿美元的AI安全市场、10亿美元的AI数据交付市场和10亿美元的AI工厂负载均衡市场。作为对比，管理层认为2026财年的该市场规模约为10亿美元。

管理层业绩指引

管理层预计，随着当前更新换代周期的持续，2027财年上半年的系统业务增长将强于下半年。受定期订阅续约时间节点的影响，预计软件业务增长将集中在下半年。

F5计划继续加大对销售能力、AI安全专家团队和市场营销的投资。2027财年，专家型销售人员数量和AI安全营销投资预计将增加一倍以上。不过，管理层预计全年运营支出占营收的比重将有所下降。

风险与关注焦点

• 管理层预计，零部件成本上升将导致2027财年Non-GAAP毛利率较2026财年第四季度水平有所下降。
• 随着硬件更新换代周期中的替换需求放缓，预计下半年系统业务增长将有所放缓。
• F5正在增加库存和供应商预付款，以保障零部件（特别是内存和存储部件）的供应。管理层表示，供应链风险尚未完全解除。
• 库存投资以及2026年11月支付的2026财年员工奖金预计将对第一季度的经营活动现金流产生影响。
• 短期运营杠杆受到销售、营销以及AI安全领域投资增加的限制。
• 管理层将AI安全和AI工厂负载均衡描述为处于早期阶段的业务机会，因此未来采纳的时间节点和规模存在不确定性。

分析师问答环节要点

管理层表示，2026财年大部分直接AI订单来自AI数据交付，其中存储到GPU的流量需求正成为应用交付控制器采购决策中日益重要的考量因素。AI安全尚处于早期发展阶段，但F5报告称已出现对模型护栏、红队测试、敏感数据保护和影子AI发现的初期需求。

在系统业务方面，管理层表示最近约三分之二的业务来自于“更新换代+”周期，三分之一来自于新客户或扩容机会。随着2027财年下半年替换订单的减少，业务组合预计将向新部署和扩容方向倾斜。

软件业务的增长得到了订阅拓展和利用率提升的支撑。管理层还预计，随着完全退出部分传统SaaS产品，SaaS及托管服务将对2027财年的营收增长做出更有实质意义的贡献。

关于供应保障，F5表示已采购保障的内存和存储库存能够支持其业绩指引并提供潜在的上行空间。由于交付周期较长，公司还在为其下一代硬件设备采购零部件。

管理层表示，价格调整通常将继续按年度进行评估，但如果零部件成本上涨过快，仍有可能进行周期外的非定期调整。

业绩电话会议完整文字记录

---

完整财报电话会议逐字稿

管理层陈述

Operator

Good afternoon, and welcome to the F5 Inc. Fourth Quarter Fiscal 2026 Financial Results Conference Call. [Operator Instructions] Also, today's conference is being recorded. If anyone has any objections, please disconnect at this time.

I'll now turn the call over to Ms. Suzanne DuLong. Ma'am, you may begin.

Suzanne DuLong

Hello, and welcome. I'm Suzanne DuLong, F5's Vice President of Investor Relations. We are here to discuss our fourth quarter and fiscal year 2026 financial results. Francois Locoh-Donou, F5's Chairman, President and CEO; and Cooper Werner, F5's Executive Vice President and CFO, will be making prepared remarks on today's call. Other members of the F5 executive team are also here to answer questions during the Q&A session.

Today's press release is available on our website at f5.com, where an archived version of today's audio will be available through January 26, 2027. We will post the slide deck accompanying today's webcast to our IR site following this call. For additional information or follow-up questions, please reach out to me directly at s.dulong@f5.com. Our discussion today will contain forward-looking statements, which include words such as believe, anticipate, expect and target. These forward-looking statements involve uncertainties and risks that may cause our actual results to differ materially from those expressed or implied by these statements. We've summarized factors that may affect our results in the press release announcing our financial results and in detail in our SEC filings.

In addition, we will reference non-GAAP metrics during today's discussion. Please see our full GAAP to non-GAAP reconciliation in today's press release and in the appendix of our earnings slide deck. Please note that F5 has no duty to update any information presented in this call.

I will now turn the call over to Francois.

François Locoh-Donou

Thank you, Suzanne, and hello, everyone. Fiscal 2026 was an exceptional year for F5, and Q4 closed it out with strong execution and accelerating strategic momentum. We delivered Q4 revenue of $917 million, up 13% year-over-year and above the top end of our guidance range. Importantly, that top line performance was powered by product revenue growth of 23%. With Q4, we have now delivered 9 consecutive quarters of double-digit product revenue growth, a clear signal that our customers are choosing F5 consistently and at scale.

For the full year, total revenue reached $3.4 billion, representing 11% annual growth, including 19% product revenue growth. These results reflect 3 megatrends we have consistently highlighted, hybrid multi-cloud adoption, the expanding threat landscape and accelerating AI inference. We call these trends early and we built for them. Our FY '26 results demonstrate how these trends are actively reshaping customer behavior right now. They also frame our FY '27 revenue outlook. Against an FY '26 baseline that finished well above our main investor event expectations, we are guiding to FY '27 growth of 8% to 9%. We expect to deliver a third consecutive year of double-digit product revenue growth in FY '27, providing a solid foundation to our outlook. Cooper will elaborate on additional guidance details in his remarks.

Before I highlight some customer wins from the quarter, I want to speak to another factor driving confidence in our outlook. The pace and quality of our innovation and the structural position it is reinforcing. The fundamental truth of modern enterprise security is that you cannot enforce consistent runtime protection without being in the data path. For 3 decades, F5 has built and defended exactly that position in the flow of traffic for many of the world's largest enterprise application environments. That footprint is the foundation. What we are building on it is the advantage. In the last 6 months, that advantage has become more pronounced because the threat environment has fundamentally changed.

Frontier AI has accelerated vulnerability discovery to machine speed, expanding attack surface through API's agents and AI models while compressing the time available to respond. More than 70,000 vulnerabilities were discovered and published through September this year. This is up roughly 50% over all of 2025. NIST reported a 208% year-over-year surge in vulnerabilities between July and September alone. Meantime to exploit has collapsed to negative 7 days. A typical 43-day patching cycle is now a liability. Customers know it, and security is increasingly leading our customer conversations and investment is moving away from fragmented point products towards converged platforms that enforce consistent policy in real time across every environment where applications and AI run. That is precisely where F5 is differentiating.

In less than 6 months since launching in March, we ended Q4 with more than 300 customers deploying our AI-powered WAF capabilities, and more than 75% of these deployments are operating in blocking mode. That figure matters. It means customers are not just evaluating AI-assisted security, they are trusting it to stop malicious activity in production. Let me make that real with something that happened just last week. On Wednesday, 2 critical vulnerabilities were disclosed in PHP, the software behind roughly 70% of the world's websites. When a disclosure like that hits, a race starts, attackers begin weaponizing exploits within hours. Defenders typically wait for vendors to write, test and ship a new signature. Last week, we all saw websites being taken down, so as not to risk attack.

Our AI-powered WAF customers did not have to wait. After the disclosure, our researchers hit an F5 protected application with multiple attack variations. Our AI-powered WAF stopped every one of them in blocking mode while letting legitimate traffic through. Our customers did not have to write a new rule, and they stayed protected while they passed. That is the shift we are betting on. Security that adapts at the speed of the attacker, not the patch cycle. We are applying that same AI-based detection capabilities to API security moving protection beyond static signatures and rules to models that identify and stop malicious API requests in real time.

As AI agents drive more machine-to-machine activity APIs have become the primary pathways agents use to reach models, data and critical business systems. Protecting those interactions at run time is becoming essential and F5 is already doing it. Protecting customers at one time is one dimension of our advantage. Strengthening the software itself is another. We are leveraging frontier open source and open weight AI models to identify and remediate vulnerabilities in our own code base faster than was ever previously possible. We are continuously scanning more than 140 million lines of code with F5-built AI harnesses and leveraging AI to remediate thousands of vulnerabilities. As a result, in just the last 3 months, we have implemented engineering fixes that previously would have taken 7 years of work.

In Q4, as part of an accelerated security cadence, we delivered 2 software updates focused entirely on remediating vulnerabilities before they can be exploited, and we are pairing that accelerated cadence with F5 Insight, our solution for fleet-wide visibility, upgrade orchestration and life cycle management across big IP environments. Launched in March, F5 Insight has gained the trust of more than 200 customers with automated fleet management already dramatically reducing upgrade times and operational risk. Together, accelerated critical releases and F5 Insights automated deployment capabilities create a faster path from vulnerability to protection, enabling customers to find, fix, release and deploy at greater speed and scale.

F5's differentiation in this fundamental threat environment is showing up in our pipeline and our wins. It's also being reflected by the analyst community. In July, F5 was named a leader in SecureIQLab's cloud WAAP cyber risk validation comparative report. In September, IDC named F5 a leader in its market scape for worldwide WAAP platforms. And also in September, just 4 months after we launched F5 AI security platform, Gartner recognized F5 as 1 of only 2 market shapers in AI application security. 3 recognitions in 3 months. We believe this progression reflects both the pace of our innovation and strength of our market position. Our application delivery and security platform is one platform, one policy framework, one management plane. We deliver consistent security and delivery policy from data center to cloud for every app, every API and every AI workload the enterprise ones. No point product vendor can replicate that position, and we are investing and innovating to extend it.

Before I pass the call to Cooper, let me bring our differentiation to life with a few customer wins from the quarter. H1 shows why the megatrends reshaping IT are translating directly into demand for F5. I will start with wins driven by hybrid multi-cloud adoption, which is driving demand for F5 across data center buildout, digital sovereignty, competitive displacement and platform consolidation. Customers operating across on-premises, colocation and cloud environments cannot meet data residency and compliance requirements with disconnected point tools. They need consistent policy enforced everywhere and that requires a converged platform. That fact showing up directly in our business. 29% of our top 1,000 customers now deploy 3 or more F5 product families, up from 22% in 2025, and we are seeing strong platform traction across all theaters.

In an 8-figure win from Q4, a large regional North American financial institution selected F5 as its unified application delivery and security platform across its on-premises, colocation and cloud environments. The customer had accumulated a fragmented architecture over time, including 2 incumbent ADC providers and 2 separate cloud security and CDN tool sets. This approach created policy inconsistency and operational complexity that solvency and compliance requirements increasingly prohibit. Following a competitive evaluation, consolidated the entire estate, displacing all 4 incumbent vendors. BIG-IP anchors high-performance application delivery across on-premises and colocation environments. Distributed Cloud WAAP, data intelligence and CDN extend consistent protection into the cloud. F5 Insight centralizes visibility and management across their full environment, one platform, one policy framework from on-premises to cloud.

In another win from the quarter, a Central Bank technology authority in Asia selected F5 as the unified platform for a national financial services cloud builds to keep critical banking infrastructure under domestic control. Following a rigorous public sector tender on technical merit and cost, F5 displaced 2 incumbent application delivery competitors. BIG-IP powers high-performance traffic management and high availability delivery across the customers' own data centers. Distributed Cloud WAAP, data intelligence and CDN extend consistent application security and API protection into the cloud, all managed through a single F5 management plane. The expanding threat landscape continues to accelerate demand for F5 and the data from our own solutions tell the story.

In core, 5 distributed cloud telemetry recorded an 84% year-over-year increase in WAF request and a 103% surge in API security events. Frontier AI models have escalated the threat level for organizations globally. As a result, the volume and velocity of attacks at the application layer is compounding and organizations are taking action to defend themselves. During Q4, a leading European financial institution determined that signature-based tools could no longer keep pace with the speed and scale of AI-driven threats. Following a competitive evaluation, the customers standardize on F5 as their single application delivery and security vendor across EMEA, APAC and the Americas. They deployed our AI-powered WAF and bot defense on Distributed Cloud alongside advanced WAF on BIG-IP. This approach established proactive signature-less virtual patching across their global digital banking infrastructure.

The ability to consolidate onto a single vendor, while maintaining consistent protection across every region and every environment was the decisive factor. This win reflects both the urgency customers feel around post-Mythos threat velocity and the growing man for platform standardization at global scale. The expanding threat landscape is also driving demand for purpose-built API security, a central bank and financial regulator in our APAC region selected Distributed Cloud API security to protect a rapidly expanding API ecosystem, underpinning the nation's digital payment and banking infrastructure. The customer had invested in an API management platform, expecting it to provide sufficient security, but shadow APIs were going undetected and critical banking APIs were exposed to operational and reputational risk. Opening regulatory requirements made the status quo untenable.

Following a competitive proof of concept, F5 displaced 3 security incumbents Distributed Cloud API security now delivers automated discovery across public, external and internal APIs. F5 brings real-time OLAP API top 10 threat mitigation that integrates seamlessly with the customer's existing BIG-IP infrastructure, deepening F5's platform footprint in a strategically important regulated account. Finally, the AI inference inflection is driving multiple vectors of growth. The center of gravity in enterprise AI has shifted decisively from training models to operationalizing inference. Industry analysts estimate that industry spending on AI inference has reached $23 billion, surpassing training-related spend of $19 billion.

Within our customer base, 78% of organizations now run their own inferencing and 77% say inference, not inning is their main AI activity. The AI inference inflection is benefiting F5 across 2 dimensions through our direct AI use cases and directly across our business. During Q4, we saw accelerating demand across our 3 direct AI use cases, AI data delivery, runtime security and AI factory load balancing. Direct AI use case bookings for the year more than doubled from the first half to well over $100 million in FY '26. This is up nearly 300% year-over-year. In addition, in FY '26, our active direct AI use case customer count grew 4x to more than 200. Those numbers reflect strong momentum today, and we see significant runway ahead as enterprises move from AI experimentation into production at scale.

In AI data delivery, we are accelerating secure AI data pipelines, ensuring high throughput storage, networking and data delivery that keeps AI clusters fully utilized across hybrid and multi-cloud deployments. Command is scaling. Customers deployed 29.2 terabits of F5-powered AI data delivery capacity in Q4 alone. This is up nearly 250% year-over-year with our banking, financial services and health care verticals leading adoption. In an AI data delivery win from Q4, a large conglomerate spanning energy, retail and digital services in our APAC region put a planned BIG-IP capacity expansion out to competitive evaluation. F5 successfully defended and grew the account against 3 challengers. BIG-IP is now the single high throughput control point for data moving between storage and GPU compute at scale.

In AI runtime security, we are safeguarding AI applications, APIs and models from abuse data leaks and attacks like prompt injection. We are delivering real-time threat defense, red seating and robust guardrails and we are seeing a noticeable uptick in momentum that we believe reflects AI moving from pilots into production. In addition, open with models now account for 34% of enterprise token volume. Because safety and guardrail behavior is very widely across open weight models, enterprises cannot rely on the models for security. Protection must be enforced in the inference path. Customers are choosing F5 because best-class security for AI models requires runtime protection that hyperscalers and frontier model providers cannot deliver.

Runtime protection at scale and with specificity is precisely where F5 is differentiated. In a milestone win, a global AI infrastructure leader selected F5 AI guardrails as the production safety layer for up to 500 AI models hosted on its flagship model platform. The customers' requirements were exacting out-of-the-box protection against prompt injection and sensitive data leakage, custom guardrails builds to their own framework and self-hosted deployment at scale. With the EU AI Act mandating in production monitoring of frontier AI models, the urgency was regulatory as well as operational. This customer also selected F5 as a launch partner in its Zero Trust AI agent security architecture, embedding F5 at the center of how they think about AI safety at scale.

AI is moving so fast that new threats emerge almost monthly and customers are turning existing F5 capabilities against them. Recently, one of the world's largest financial institutions used its F5 bot defense to solve a problem that barely existed 6 months ago, personal AI agents logging into customer accounts. The bank decided the risk to its clients and to itself was too high. F5 bot defense now identifies and blocks those agents at the front door. And this may is not alone, several other large enterprises are turning on the same advanced detection and protection. This is what we mean when we say AI is changing everything. Across our bot defense deployments, traffic from self-identifying AI agents more than doubled in a single quarter. Agents are becoming a new class of user, and we believe we are at the very beginning of that curve.

Since F5 sits in the data path, we can respond in real time, often with capabilities customers already have. In AI factory load balancing, we are optimizing traffic and GPU utilization both across and within AI factories, increasing token throughput, reducing time to first token and lowering per token cost. While this remains our most nascent direct AI opportunity, we saw a notable pickup in activity in Q4. Earlier this month, we announced that Indosat or Ooredoo Hutchison one of Southeast Asia's largest telecommunications providers selected F5 to power the Indosat intelligence suite. This is an enterprise AI platform purpose built to solve a challenge we are hearing consistently across regulated markets. Enterprises are ready to move AI from pilot to production but cannot do so without solvency, governance and cost predictability in place. The Indosat Intelligence suite addresses all 3 giving businesses govern access to more than 300 open and closed AI models through a single interface.

BIG-IP Next for Kubernetes provides the infrastructure layer, delivering tenant isolation container security and GPU aware load balancing to optimize latency and token throughput at scale. For organizations with data residency requirements, workloads can be processed in market with policy governing what can be routed to global models. A similar sovereign AI dynamic is playing out in Europe. Last week, we announced that Telenor AI factory selected F5 BIG-IP Next for Kubernetes as the intelligent service plan for its sovereign AI infrastructure. The Telenor platform is purposeful to give Norwegian and European enterprises access to trusted high-performance AI services under full domestic control. where Indosat is solving the model access and governance challenge, Telenor is solving the infrastructure governance challenge, including how to operate shared sovereign GPU capacity securely across multiple tenants while meeting European data residency and compliance requirements.

BIG-IP Next for Kubernetes on NVIDIA BlueField I DPUs deliver secure multi-tenancy policy-based traffic control and real-time observability directly in the AI service path, preserving host CPU capacity for AI workloads. Together, Indosat and Telenor illustrate that as AI factories transition from infrastructure projects into commercial service platforms, service providers need a governed, policy in force to deliver layer and F5's AI factory load balancing offers a proven solution. I noted, we also continue to benefit from indirect AI demand. While it's challenging to quantify as AI adoption accelerates broadly, it is generating exponentially more traffic, driving higher security requirements and accelerating digital solvency investments, all of which are showing up in systems demand and the rapid adoption of solutions like our AI-powered WAF.

Now I will turn the call over to Cooper who will walk through our Q4 results and our outlook. Cooper?

Cooper Werner

Thank you, Francois, and good afternoon, everyone. Q4 marked another quarter of solid execution with both revenue and earnings well above the top end of our guidance. We continue to see strong demand with customers' hybrid multi-cloud adoption, driving expanding opportunity for F5. I will review our Q4 results and close with our guidance for FY '27 and Q1. We delivered a strong Q4, growing revenue 13% to $970 million, with a mix of 56% product revenue and 44% services revenue.

Product revenue totaled $510 million, increasing 23% year-over-year, while services revenue of $408 million grew 3% year-over-year. Systems revenue totaled $239 million, up 29% over Q4 fiscal year '25. Our software revenue of $271 million grew 19% year-over-year. Subscription-based software revenue totaled $229 million, up 16% year-on-year, representing 85% of our Q4 software revenue. Perpetual license software totaled $42 million, up 37% year-over-year. Revenue from recurring sources contributed 68% of our Q4 revenue.

Shifting to revenue distribution by region. Our teams drove strong growth across all theaters. Revenue from the Americas grew 5% year-over-year, representing 53% of total revenue. EMEA delivered another robust quarter with 31% revenue growth, representing 30% of revenue. And APAC grew 14% year-over-year, representing 17% of total revenue. Looking at our major verticals. Enterprise customers contributed 69% of Q4's product bookings. Government customers represented 18% of product bookings, including 6% from U.S. Federal. Finally, service providers contributed 13% of Q4 product bookings.

Our strong top line performance in the quarter carried through the rest of our operating model, driving earnings outperformance in Q4. App gross margin was 82.2%. Non-GAAP gross margin was 84%. Our GAAP operating expenses were $543 million. Our non-GAAP operating expenses were $462 million. Our GAAP operating margin was 23.1%. Our non-GAAP operating margin was 33.7%. Our GAAP effective tax rate for the quarter was 18.9%. Our non-GAAP effective tax rate was 20.9%. Our GAAP net income for the quarter was $186 million or $3.23 per share. Our non-GAAP net income was $255 million or $4.42 per share.

I will now turn to cash flow and balance sheet metrics. We generated $118 million in cash flow from operations in Q4 and free cash flow of $91 million. These results reflect deliberate and strategic working capital decisions to mitigate supply chain risk, including inventory builds and supplier prepayments executed to secure supply for FY '27 revenue. CapEx was $27 million, reflecting our ongoing investment in our global pop infrastructure to support growing demand for distributed cloud services. DSO for the quarter was 44 days. Cash and investments totaled $1.62 billion at quarter end. Deferred revenue was $2.2 billion, up 10% from the year ago period.

We remain committed to returning cash to shareholders. In Q4, we repurchased $100 million worth of F5 shares at an average price of $422 per share, bringing our total share repurchase for FY '26 to $600 million. Finally, we ended the quarter with approximately 6,650 employees. I will now briefly recap our full year software and security revenue results before moving on to guidance. Software grew 8% year-over-year, totaling $870 million, with software subscriptions representing 87% of FY '26 as software revenue. Our software revenue spans 3 categories. Perpetual licenses, term-based subscriptions and SaaS and managed services. We provide a breakdown of each category's contribution annually.

Perpetual software licenses contributed $111 million in software revenue, down 8% year-over-year. term-based subscriptions contributed $578 million, up 14% year-over-year driven by strong renewal rates and continued expansion within our multiyear agreements. And SaaS and Managed Services contributed $182 million, up 3% year-over-year. We also disclosed annualized recurring revenue, or ARR, for our SaaS and Managed Services offerings on an annual basis. Over the past 3 years, we have been executing a planned transition away from certain legacy offerings. In FY '26, we completed this transition, ending the year with $211 million in total SaaS and Managed Services ARR, up 14% from FY '25.

Our core SaaS and Managed Services ARR grew 24% year-over-year. With the legacy transitions completed, our SaaS and Managed Services offerings are now positioned to be a growth driver for our software business in FY '27 and beyond. Finally, we also break out our security-related revenue annually. This year, our total security revenue, which includes standalone security attached security and maintenance revenue related to security grew 10% year-over-year to approximately $1.3 billion. Our product security revenue, which includes both our stand-alone and attached security revenue, totaled $746 million, up 15% year-over-year.

I will now speak to our outlook for FY '27. We FY '26 was an exceptional year. We exceeded expectations on systems demand, delivered healthy software subscription expansion. We enter FY '27 with momentum. The 3 secular megatrends Francois described hybrid multi-cloud adoption, the expanding threat landscape and AI inference are accelerating, and F5 is positioned at their intersection, creating healthy demand across each of our revenue categories. That positioning, combined with the strength of our pipeline as we exit FY '26, supports our FY '27 revenue growth outlook of 8% to 9%.

Let me address each revenue category in turn. On systems, the strong refresh cycle we have described throughout FY '26 persists into the first half of FY '27. Beyond refresh, systems demand continues to be driven by competitive takeouts digital sovereignty and AI readiness investments and capacity expansion as AI-enabled workloads drive higher throughput requirements. Taken together, these dynamics support our expectation of upper single-digit systems growth for the full year, an improvement over our mid- to upper single-digit growth target outlined at our May investor event. Given the cyclical nature of our ongoing systems refresh, we expect growth in the systems business will be stronger in the first half of the year.

For software, we exit FY '26 with strong momentum, and our multiyear subscriptions give us visibility into the renewal and expansion timing and activity in FY '27. We continue to see strong growth in utilization rates across our software subscription base. In addition, with the legacy offering transition now complete, our SaaS and Managed Services business is positioned to accelerate in FY '27. Putting each of these factors together, we expect to deliver solid double-digit software growth for the year. And given the timing of term renewals, we expect this growth will be stronger in the second half of the year.

Finally, for services, we continue to expect low to mid-single-digit growth for FY '27. On gross margin, we expect FY '27 non-GAAP gross margin in the range of 80.5% to 82.5%, reflecting the higher component costs we discussed previously. We will continue to balance operating efficiency with investment in innovation and go-to-market capacity to drive our revenue growth opportunity in FY '27 and beyond. We expect non-GAAP operating margin in the range of 33% to 35%. Operating margins will be lowest in our fiscal Q2, with a similar seasonal sequential decline to prior years. This reflects payroll tax resets in January, investments to expand sales capacity and costs associated with our annual customer event in March. We expect our FY '27 non-GAAP effective tax rate will be in the range of 19% to 21%. Finally, we expect FY '27 non-GAAP EPS in a range of $18 to $18.30.

Turning to a few items that will be relevant to cash flow in FY '27. First, on inventory. As I discussed earlier, we are strategically leveraging our cash generation to meaningfully increase our inventory position. And we expect to continue to increase inventory in FY '27 with the strongest increases in the first half of the fiscal year. This is a deliberate action to increase delivery capacity to support systems demand. Second, a note on bonus timing. In FY '26, we transitioned our employees bonus plan from a quarterly payment cadence to an annual cadence to better align incentives to full year performance. As such, full year bonuses earned across FY '26 will be paid out in November 2026. These 2 items will have an effect on operating cash flow in Q1. We expect to return to normalized patterns for the remainder of the year.

Finally, on capital allocation. Our philosophy remains unchanged. We are committed to a balanced approach that returns cash to shareholders while preserving flexibility for strategic investments. As announced in our earnings release, our Board has authorized an additional $1.5 billion for share repurchases. We had $322 million remaining on our authorized share repurchase program as of the end of Q4, bringing the total authorized to $1.8 billion. Consistent with our practice over the last several years, we expect full year share repurchases in FY '27 to represent at least 50% of our annual free cash flow.

I will now wrap up with our outlook for Q1. We expect Q1 revenue in the range of $875 million to $895 million. We expect Q1 non-GAAP gross margin of approximately 81.5% to 82.5%. We estimate Q1 non-GAAP operating expenses of $420 million to $432 million. We expect Q1 share-based compensation expense of approximately $70 million to $72 million. We anticipate Q1 non-GAAP EPS in the range of $4.25 to $4.37 per share.

With that, I will now pass the call back to Francois. Francois?

François Locoh-Donou

Thank you, Cooper. Let me close with this. When I step back and think about what we have built and where the market is going, I keep coming back to the same conclusion. The world is deploying more applications, more AI, more agents across more environments under more regulatory scrutiny against the faster and more sophisticated threat landscape than at any point in our history. Every one of those dynamics is a reason for our customers to have F5 in their architectures. That is not a coincidence. It is the result of decisions we made years ago about where to build and what to build for. We are meeting this moment and our opportunities are growing as a result. .

Operator, please open the call to questions.

Operator

[Operator Instructions] Your first question is from the line of Matt Hedberg with RBC Capital Markets.

分析师问答

Matthew Hedberg

Great. Congrats really on a very strong year and a really great outlook. A lot of things to parse through here. I wanted to start out with the direct AI use cases. Francois, you talked about it, I think, reaching over $100 million in bookings, I think that grew 300% year-on-year. Can you kind of help us just step back and think about what is the addressable TAM that you guys are focused on from an AI perspective. And how should we think about the scaling of that bookings trajectory and its potential to accelerate overall growth? .

François Locoh-Donou

Matt, thank you. So let me just start with your question about TAM. I think we shared at our Investor Day that we expected our addressable market by 2030 to be around $11 billion in AI, of which $9 billion was AI security, about $1 billion in AI bidder delivery and the $1 billion in AI factory load balancing. $11 billion in 2030, up from what we believe is about $1 billion this year. So it would be a 10x increase in the addressable market over the next 4 years. So that's in terms of the TAM.

In terms of the opportunity we're seeing, let me break this down into those 3 categories because what we've seen here is very promising. So I'll talk with AI data delivery, which represents the majority of the -- we said the bookings have grown to over $100 million for the year, which is up 300% year-on-year. The majority of that is actually AI data delivery. And what we're seeing there is increasingly the AI data delivery features are important criteria in selection of ADC technologies by our customers. So in our core ADC market, AI data delivery is playing an important role. We are already differentiated, and we've accelerated the differentiation with new features for data delivery, specifically support for S3 protocol, support for MCP protocol, our partnerships with some of the key storage players. And so we see that market continuing to grow. And we feel very, very good about the opportunities that we are we are seeing in data delivery this year and will continue to grow next year.

The next use case that we've talked about is AI security. And this one is -- I think it's still early days. We launched our AI security platform earlier this year, just about 3 or 4 months ago. But we are already seeing very significant traction in this area. If you look at the last 6 months in cyber security, one of the things that has become painfully clear is that model providers, frontier model providers cannot provide appropriate guardrails and security for their models or agents, number one. And number two, the dynamic we're seeing with our customers is that all of them are really deploying AI in an environment that are hybrid multi-cloud and also multimodal, meaning they're deploying open weight model, open source model and frontier models for cost reasons, for compliance reasons, protecting their intellectual property. And all of that calls for a player that can secure all of these models, whether in the cloud or on-premise, and that's where our hybrid multi-cloud advantage, we think, is going to be pretty significant. And we're already seeing a number of use cases on that.

One of the ones I mentioned in the -- in my prepared remarks, is that one of the largest companies in the world that's at the heart of the AI revolution is using F5 to protect 500 of their models that are being accessed by developers external developers. And they're pretty -- their requirements are pretty stringent, but they've chosen F5 because we have this programmability. We have the ability to customize models for the most demanding environments. And that's one of the reasons that we also have had this recognition from Gartner as 1 of the 2 market shapers in AI security. So we feel really good about AI security, but we feel this market is just going to grow substantially over the next several years.

And then in the third area, Matt, AI factory load balancing, we saw a pickup in activity this quarter you saw the announcement of the 2 sovereign AI implementations we have in Asia and in Europe, I shared earlier that the neo clouds, which will be a target market and is growing in terms of the infrastructure spend, initially, we have been very focused on offering GPU as a service basically bare metal GPU as a service primarily for hyperscalers. That's not a use case where there is a need -- a significant need for F5. But as we see either the neo clouds or sovereign AI players start to offer token as a service for enterprises, F5 benefits become very clear. around tenant isolation, helping with dominance, helping with security. And those are the initial wins that we've had in this market, but I expect more to come in 2027. When you take all that together, we are looking at a TAM that is going to grow very significantly over the next 4, 5 years. And in all 3 of these areas, we have initial wins and traction that we believe can accelerate in '27.

Operator

The next question is from the line of Joseph Cardoso of JPMorgan.

Joseph Cardoso

Congrats on the results. Maybe just wanted to touch on the operating leverage. It was a bit lighter this quarter and the outlook for 1Q and next year, I guess, in aggregate, kind of suggests more of the same. Can you maybe just walk through what drove that this quarter? It looks like OpEx investments stepped up meaningfully. And then as we kind of look ahead, it seems like the guide is also implying a little bit of limited flow-through of revenue. And obviously, there's a bunch of moving pieces there around margins and OpEx, et cetera. But maybe just talk about the key assumptions that's driving the outlook here? And maybe more importantly, what needs to go right for you to kind of flip the script here on that trend and maybe see a better flow-through than what's implied in the outlook?

Cooper Werner

Sure. Thanks, Joe. This is Cooper. So yes, I can talk to both Q4 and Q1. But essentially, the biggest driver on the OpEx are investments that we're making to drive future growth. So we've been increasing our sales capacity. We started in Q4, and we've been aggressively ramping additional capacity, both kind of broad-based capacity as well as specialists for the AI security opportunity. And really, what you're seeing is a reflection of our belief that that's a growth opportunity that we think extends beyond FY '27. We're being a little bit earlier in the cycle in terms of making those investments to ensure that we've got that capacity fully ramp as that opportunity continues to mature.

We're also making other go-to-market investments across marketing, to support our sellers and raise the visibility. We think that we're really positioned to win in the AI security space. And so we just want to make sure that we're capitalizing on that opportunity. And that's largely what you see in the Q1 guide with OpEx. And then over the course of the year, we do expect that you'll see OpEx as a percentage of revenue come down. For the full year, it should be down year-over-year as a percentage of revenue. as we start to see the growth continue to flow through in the year. And then just from Q4 this year, the other dynamic was we -- at the end of the year, just with the outperformance that we had, there is an impact to accelerators, both on the commission and bonus perspective across the employee base.

Operator

The next question is from Tim Long of Barclays.

Timothy Long

Yes, I'll start with one on the system side, and then I'll come back with a follow-up on software. Thanks to the color on the linearity into the year, it sounds like a strong first half, so more front-end loaded. Just curious when looking at kind of a medium, longer-term trend here, and knowing that you guys have seen a lot of these refreshes, which has obviously been a very good one. So when you look at the kind of one half to second half deceleration, do you -- is this just timing of the impact of the refresh kind of running its course? Is it -- you've had enough -- you had the bulk of the share wins, competitive takeouts? Or is there something else at play? And then I had a software follow-up.

Cooper Werner

Thanks, Tim. Yes, it's entirely around the cyclical part of the refresh motion. And you've heard us refer to it as a refresh plus. Where we see a deceleration is just on the replaced part of that motion. And we pass those end of software support dates, we expect that the revenue we're getting just from replacing legacy installed base, you're going to see that decline over the back half of the year. But I think what's more interesting is that the growth that we're seeing in new use cases and expansion continues to be very strong. In fact, it accelerated over the back half of the year. And then the competitive opportunity, we don't think that, that's slowing down. We actually think that that's going to grow over time. But the real story around the growth over the course of the year is going to be the AI data delivery opportunity that Francois already referenced. We think that's really early. Continued expansion and then, of course, the digital sovereignty concerns that we're seeing in international theaters, those are all still relatively early as well. So we think on balance, you'll see growth through the year, that growth will be slower in the second half just from the cyclical part of the refresh. But on balance, the underlying kind of secular drivers will continue to see -- drive a lot of opportunity in our systems business.

Timothy Long

Okay. And then just the software follow-up. I think you beat Q4 pretty handy on the software number. Curious if that means that more likely to be term? And just would love any color on whether it's true-ups or whatever you call them now or new activity that drove that? Just curious how that would carry into next year, the upside in Q4.

Cooper Werner

Yes. It's really just new use cases and expansion that we're driving that. I mean a lot of that does come through in the form of these renewals but where we saw the upside growth, it was a higher rate of expansion. We also won some perpetual use case opportunities that we largely had visibility going into the quarter. But broadly, I think the trend that we're excited about is that we've had a long history of driving healthy expansion, the adoption we're seeing across the full portfolio of customers that are leveraging multiple product families continues to increase. And just given that we've got a large base that comes up for renewal next year, it's important that we see those strong utilization rates, it gives us that visibility on the strong growth opportunity.

And then the second one, which we talked a little bit about is our SaaS and Managed Service business, which historically, we hadn't seen growth because we were undertaking a transition from some legacy offerings. You saw that the ARR on that grew pretty substantially in FY '26. That's the ratable part of our software business. So that growth that we saw in the ARR in FY '26 will come through in the model is revenue growth and up at '27. So it's for the first time, it's an additional leg to our software growth opportunity, and we expect that growth to continue beyond FY '27.

François Locoh-Donou

Tim, I'll just add -- I want to come back to the competitive dynamics that you mentioned in your question because part of it is taking out kind of more traditional ADC competitor directly. That continues. In fact, I think this quarter was one of our strongest quarters of the last several years in terms of direct competitive takeouts for ADC. We expect that to continue to grow because in a lot of cases, what we've taken out is just an initial implementation of a very large install base in a customer. So we expect to continue to grow share in existing customers where we have had an initial pick-out win. But the other phenomenon that we're seeing from a competition perspective is consolidation with a lot of our customers as they -- 93% of enterprises are now deploying this hybrid multi-cloud architectures and to try and remove the complexity of hybrid multi-cloud, many of our customers really want to consolidate across private cloud and public cloud environment and have a single platform providers for both.

In fact, in our prior remarks, we mentioned one such win this quarter, which was an 8-figure deal with a financial institution where we took out not 1 or 2 competitors, but 4 competitors to replace both ADC players and CDN players and cloud players into the single platform of F5 on-prem hardware software plus F5 distributed cloud. And we -- that ability for us to consolidate for large enterprise customers in hybrid multi-cloud environment. is core to our strategy. It is a part of our differentiation that is accelerating and where we're going to see more wins in the future as our platform develops. This quarter, we just introduced a capability called F5 Insight that gives our customers basically one policy, one place to look at all their environments and monitor the performance of their applications and get a lot of AI-powered insights. And that just further accelerates disability to consolidate environments and simplify operations for customers. So we'll come back to that more as our competitive differentiation develops, but that is -- that's core to what we're doing and to our growth.

Operator

The next question is from the line of George Notter of Wolfe Research.

George Notter

I wanted to ask about sort of the sales and marketing push behind the AI Guardrail product. I hear the commentary about increasing expenses in sales and marketing, you got to get that. But I'm just wondering if you're putting enough of a push behind guardrails, is there any way you could quantify how much you're investing there from a selling and marketing perspective. Maybe talk about anything you're doing in terms of incentivizing the sales force head count I'd just be really interested to learn more about that because it seems like it's a terrific opportunity, but at the same time, there are a lot of competitors with products that trip upon some of the functionality that you guys have. And obviously, I think a lot of this is going to be a sales and marketing game. So anything you can tell us there would be great. .

François Locoh-Donou

Well, thank you. There may be a lot of competitors. But if you look at the initial recognitions that we have received and the perception of the market, we are clearly one of the top companies in this area already, and we are investing significantly. First is, of course, our platform because it isn't just about guardrails, it is the ability to discover vulnerabilities in models through red seating, turn these vulnerabities into mitigations, custom mitigations for customers, the ability to discover or shadow AI and have usage control, all of these are capabilities that are converging, and we're converging all of that in a single platform, and we're going to expand from that. Of course, sales and marketing investments are absolutely critical to our success. We have been ramping our sales and marketing resources throughout the year in this area. We will be doubling our headcount. The specialist seller head count that we have will be doubling in 2027.

Our marketing investment in this area will be more than doubling in 2027. And in addition to the specialist organization that we have put in place for that, we have put also incentives for all of our field force to engage customers in AI security conversations because our belief is that every single one of our large enterprise customers is looking to solve the problem of securing AI. So all of our field sellers will have incentives to engage in AI security in terms of their compensation structure.

Operator

The next question is from the line of James Fish of Piper Sandler.

James Fish

Cooper, do you feel like you have enough apply today in terms of visibility to meet the guidance and then some? Or will you need to find more, and that's why we have the continued meaningful increase in inventory. Just kind of help us bridge inventory to kind of what you're seeing on the systems side. and just overall how the transition towards the latest memory is going.

Cooper Werner

Yes, thanks. No, we've been investing pretty aggressively, as you've seen. We've talked about that over the course of the year. And there are long lead times on these parts, which is why we got -- started making these investments early and what you're seeing is they're starting to flow through onto our balance sheet. So we feel very good, particularly around memory and storage, which is the area that we felt like would be at most at risk. And just based on the inventory, we've been able to secure we're very well positioned not just for our guidance, but for upside. Now it doesn't mean to say that the supply chain has been completely derisked. I mean there are always different areas of components where there can be risk from time to time, and it's something that we aggressively monitor our inventory strategy is not just around raw materials, but it's also being having an upside build forecast that we put out to our manufacturers early to make sure that we're driving the full set of components.

The other thing that we're doing is we're also investing in the supply for our next generation of appliances, just given the nature of lead times, we're getting in front of that as well. And so that's really kind of what's behind the continued growth that we expect to see in our inventory balance over the course of this year. So we're not just thinking short term. We're making sure we're well positioned to support growth going beyond FY '27.

James Fish

Helpful. And just as a follow-up to the guide side of things, you're talking double-digit software growth. We've been talking about that level for a little while. But is there a way to think about -- it sounds like SaaS, we should think about as that 14% ARR growth continuing at roughly mid-teens in terms of the translation to revenue and we should roughly expect that in terms of term license then? Or how should we think about each of the pieces?

Cooper Werner

Yes. Thanks. Great question. And we wouldn't specifically guide, but that's kind of the right way to think about it. We think we're going to see very healthy growth across both of those areas of our software business. Of course, the ARR growth, you have really good visibility there because of the ratable nature and we expect to continue to see strong momentum with new bookings and expansion across the SaaS business. And then the license term license piece, we think we'll also see really strong growth. That's where that growth will be weighted higher in the back half just because the timing of when some of those opportunities come up for prosecution. But the growth -- we think it will be pretty good growth across the year, just a little bit more growth on the back half.

Operator

The next question is from the line of Meta Marshall of Morgan Stanley.

Meta Marshall

I wanted to ask a couple of questions. First, just on the continued strength out of EMEA and a lot of that having to do with data sovereignty projects. Just are those -- anything that we should be mindful of just in terms of project-based nature of those that create difficult comps? And then maybe just a second question on disaggregating some of the product growth that we saw this year. I know last year, at the end of the fiscal year, you guys kind of gave how much of the product growth you thought was attributable to the product refresh? Just any sort of disaggregation you were doing this year.

François Locoh-Donou

Meta, I will start with the question about EMEA. No, we've seen consistent growth in Europe, Middle East and Africa over the last several quarters that have been at or about 30%. And that is driven by a few things. One, of course, is the continued adoption of hybrid multi-cloud architectures. Specifically in Europe, sovereignty is a driver. Customers repatriating to either local clouds or repatriating applications to private data centers. That we are -- we expect that to continue. And it's not a project-based one-off set of decisions that are made. It's generally a change in philosophy and a change in architecture that then drives an ongoing relationship with a customer an ongoing set of deployment or expansion of deployments into new data centers. So that's how I would think about it. And I wouldn't, as a result, expect a dramatic shift in what we're seeing not just in Europe but internationally in our business.

Meta Marshall

Yes. And then just in terms of kind of disaggregating, I think you're talking about just the systems piece of our business. with the refresh. The mix that we've seen over the last couple of years is about 1/3 of the business has been coming in as new opportunities, whether that's AI -- direct AI use cases or capacity expansion or competitive takeouts and then about 2/3 of the business has been coming through that refresh plus motion. The good news is that the growth has been very high across both of those categories. So the mix has actually stayed stable in spite of the fact that we've been having such a strong refresh. And so, as we go into next year, we think the mix will be more weighted towards the new and expansion opportunities. Some of that business that's come through the refresh motion where it was expansion. It was just timed with the refresh. That business will start to come through in the form of net new opportunities that will be kind of outside of a refresh opportunity. So we would expect in the second half of the year that replacement bookings will be down year-over-year and that we'll continue to see strong growth from new and expansion leading to an overall positive growth profile.

Operator

The next question is from the line of Tal Liani of Bank of America.

Suzanne DuLong

Why don't we move on, and we'll come back to Tal. .

Operator

Yes, we are having problems here in all. So we will move on to Jeffrey Hopson from Needham.

John Jeffrey Hopson

Just wanted your updated thoughts on the pricing environment. I think you guys had gotten into a yearly kind of revisiting the pricing cadence. Just wanted your thoughts on a go-forward basis, considering the memory situation is changing quarterly or even monthly.

Cooper Werner

Yes. In general, we've had a program where we revisit pricing on an annual basis to capture the value of the innovation and adjust to any cost component pressures that we're seeing. That's still our philosophy from time to time, if we see an acceleration in costs, we may make an off-cycle adjustment, but our plans right now would be to stick to an annual cadence with price increases.

Operator

The next question is from the line of Simon Leopold of Raymond James.

Simon Leopold

All right. Okay. I believe we are having trouble hearing Simon. So at this point, there are no more questions. I will now turn the call back to Suzanne DuLong for closing remarks.

Suzanne DuLong

Thank you Connor. Appreciate it. Apologies if there's going to be technical difficulties. We'll definitely look into it with the provider here. But if those of you that weren't able to ask questions, need anything to follow-up, please reach out to me directly. Broadly, thank you all for joining us today. We look forward to seeing you during the quarter. We'll be at several conferences. Thanks so much for your time today.

Operator

This concludes today's call. Thank you for attending. You may now disconnect.

原文链接