FFIVF5, Inc.

“if you look at the first half of the year, -- we had approximately $50 million in sales in the first half of the year on these use cases. That's up more than 200% year-on-year. And we're now approaching about 100 customers that are using F5 for their AI use cases.”

“we have developed an integration with NVIDIA where we have been able to basically refactor our software to work in ARM architectures and specifically work on NVIDIA BlueField technology. We've done a lot of work with NVIDIA over the last 18 months. As of December, we have now been formally put into NVIDIA's reference architecture. Since then, there have been a number of tests, including third-party tests to test the efficiency gains from this integration. Those tests have validated that basically the integration of F5 software on these NVIDIA DPUs helps AI factories generate 30% to 40% more token for a certain amount of GPUs.”

“the number of customers choosing F5 for web application firewalls are up 62% year-on-year. The number of customers choosing F5 for API security is up 54% year-on-year. And for bot defense, it's up 33% year-on-year.”

“we believe that all security is going to be AI-powered. Static security, static signatures are really not going to be able to cope with the power and the speed that these new models have in terms of creating exploits. And so this is a shift that we saw coming. We have been investing in AI-powered security for a while now. Just this quarter, you may have seen this, we released our AI-powered web application firewall. We also released our Agentic Bot Defense solution.”

“Our research shows 78% of enterprises run inference themselves using more than seven models on average. Organizations are standardizing on a new architecture with models distributed across the data center, the cloud and the edge. And the next shift is already underway. AI agents are moving into production and enterprises are adapting their applications for agent interaction. This is driving more compute, more data delivery and more security to protect inference.”

“In an AI data delivery win, a global payments company needed a more resilient way to move rapidly growing AI data between storage and compute as they scale the training and retrieval workloads. F5 improved performance and resiliency while displacing both an in-house solution and a competitor, positioning us at the center of the customers' AI infrastructure strategy.”

“a large healthcare services organization started with a life cycle refresh across hundreds of legacy systems. As the project progressed, they expanded the scope to support an AI-driven consumer engagement platform. F5 became the control point for secure, low-latency traffic and data movement across applications, storage and their GPU server environment.”

“we've also introduced in the last few months, AI Guardrails, which is AI Red Team and AI Guardrails. So technologies that help our customers both detect vulnerabilities in their AI models and mitigate these vulnerabilities. And we have introduced a product called AI Remediate that automates the process of creating mitigation for these vulnerabilities.”

“In an AI factory load balancing win, a major manufacturer an existing F5 customer needed to support operations and established a digital twin of their manufacturing environment for simulation and optimization. They deployed BIG-IP as the production traffic layer across their GPU server environment, improving availability and offloading encryption.”

“We introduced AI-powered capabilities in Distributed Cloud WAF, replacing manual policy tuning with automated outcome-based threat blocking. Our F5 training model helps customers stay ahead of increasingly sophisticated AI-driven attacks that are growing in both speed and complexity.”

“In an AI runtime security win, an industrial automation firm needed a scalable way to assess risk and govern a growing number of AI applications and models. They chose F5 based on the depth of our red teaming insights and strong integration with their existing security stack.”

“We launched Agentic Bot Defense, extending our industry-leading Bot Defense to autonomous AI agents, a new and fast-growing category of traffic. The result is that customers can confidently adopt Agentic AI while ensuring only verified trusted agents reach their applications.”

“We released F5 AI Remediate, which closes the loop between our AI Red Team and AI Guardrails products. It collapses the path from vulnerability discovery to runtime protection from days or weeks into minutes.”

“We are capturing AI runtime security wins, protecting AI applications, APIs and models from emerging threats such as model abuse, data leakage and prompt injection.”

“Our research shows more than 90% of enterprises run hybrid multicloud today across an average of 19 locations.”

“the customers who are today really focused on -- have already started worrying about AI security and protecting AI models and AI applications that have new types of vulnerabilities like prompt injections, model abuse, et cetera. Those customers are a small minority, typically the largest customers in any vertical, the customers perhaps have a lot of sophistication in security, financial services companies, very large technology companies. But today, it's a small minority of the universe of customers we serve. And I think that number of customers is only going to grow over the next couple of years as more and more customers actually implement AI in inference.”

“As AI models become more capable, attackers are using them to launch attacks against production applications at higher volumes and with greater variation than traditional defenses were designed for. Our customers see this and they are responding. They are deploying more application security and prioritizing best-in-class defenses. The era of checkbox security is over. AI applications require best-in-class security to match both the volume and the sophistication of AI-driven attacks.”

“what we are seeing is that a number of customers who are building AI factories are early in terms of sophistication in that their first priority is to get these AI factories, these GPU farms up and running, get them running -- get them working, get these Kubernetes clusters to work. That takes quite a bit of technical sophistication and customers are really focused on that.”

“we are now in an era where the window of time for an enterprise to patch their applications has closed as we have AI models that are very powerful and can now find and exploit vulnerabilities in any application almost in real time.”

“a lot of innovation that is accelerating, is in part -- by the way, because we're also leveraging AI to do that.”