Enterprises are currently fighting a two-front war. On one side, there is an aggressive push toward AI adoption; on the other, an infrastructure landscape so fractured across edge, cloud and on-premises sites that scaling becomes nearly impossible.

This “complexity tax” is stalling innovation. For the modern operations team, the dream of lightning-fast artificial intelligence is being deferred by the manual labor of managing a dozen disconnected tools that provide plenty of alerts but almost no actual signal.

This week at F5 AppWorld in Las Vegas, the conversation shifted from the “what” of AI to the “how.” The message from F5? Organizations cannot secure or scale the AI era using a “Frankenstein” architecture of disconnected point products. To move forward, the industry is eyeing a massive consolidation of the networking and security stacks — a move toward what F5 calls the Application Delivery and Security Platform or ADSP.

This is something F5 has been moving toward for years. The company has been the undisputed leader in application delivery controllers or ADCs for years despite many companies both big and small taking runs at that business. Along the way, F5 has built a strong security portfolio and the coming together of the two products, XOps and F5 Insight, resulting in the ADSP.

The three friction points holding back AI

Before organizations can move forward with autonomous AI agents, they must resolve three fundamental conflicts currently stalling adoption:

1. The signal-to-noise ratio Modern information technology environments are saturated with data but starved for information. “They have a dozen tools, a thousand alerts and not enough signal,” Kunal Anand, chief product officer at F5, noted on a briefing with analysts. Without unified observability, identifying a bottleneck in an AI training pipeline or a security flaw in a large language model becomes a forensic exercise rather than a real-time fix.

2. The agentic security gap As we shift from chatbots to agentic AI — where AI agents autonomously interact with APIs to execute tasks — the attack surface gets exponentially bigger. Traditional web application firewalls or WAFs were built for human-to-app interactions. They are often blind to the Model Context Protocol traffic that defines the AI-to-AI economy.

3. The looming shadow of “Q-Day” While AI is the immediate priority, the “store now, decrypt later” threat of quantum computing is forcing a rethink of encryption. Organizations are hesitant to overhaul their entire delivery stack for AI if it isn’t also “crypto-agile” enough to survive the transition to post-quantum cryptography or PQC.

F5’s strategy: Collapsing the ‘mess’ into a platform

F5’s announcements this week center on the idea that application delivery and security are no longer separate domains. During the briefing, Chief Marketing Officer John Maddison emphasized that the goal is to be a “control point” regardless of where the app lives — whether it’s on-premises, in the cloud or sitting on Nvidia DPUs.

“F5 ADSP collapses that mess into a platform,” he explained. “With F5 Insight, we turn scattered telemetry into a clear story and the next best action. Then we extend that foundation for agentic AI workloads and future-focused cryptography, because the infrastructure is changing, ready or not.”

Key evolutions in the ADSP stack

To address these hurdles, F5 unveiled several major enhancements to its platform:

• F5 Insight for ADSP: This is the “brain” of the operation. It leverages OpenTelemetry to provide unified visibility across hybrid and multicloud settings. Crucially, it uses AI-driven proactive guidance to generate “operational narratives,” allowing teams to prioritize vulnerabilities through natural language rather than digging through logs.
• BIG-IP v21.1: A significant update for F5’s flagship software, introducing NIST-compliant PQC ciphers to protect against future quantum threats. It also adds Dynamic Client Registration to empower agentic AI with secure, automated resource access. The AI-WAF has been deeply integrated to secure the specialized traffic used by LLMs.
• AI Remediate: Bridging the gap between “red teaming” (finding holes) and “guardrails” (blocking them), this new tool automates the creation of security policies to protect AI models in production.
• NGINX Agentic Observability: By inspecting MCP metadata directly in the traffic path, NGINX now provides visibility into “shadow AI” activity — AI agents interacting with services without explicit IT oversight.

The transformative value: Beyond the perimeter

The shift toward a unified platform isn’t just about technical elegance; it’s about business velocity. Research from IDC suggests that by integrating these layers, organizations can “optimize operations, strengthen security and scale across their environments” more effectively than with siloed tools.

For the C-suite, the value proposition is clear: Convergence equals lower risk. By replacing dozens of disparate SKUs with simplified, value-driven bundles — as seen in F5’s new Distributed Cloud Services packaging — companies can reduce tool sprawl and proxy overload.

As Maddison noted during the briefing, the market for cybersecurity is expanding by roughly $3 billion thanks to AI-enabled applications. However, capturing that value requires an infrastructure that is “AI-aware.” Whether it’s ensuring session persistence for AI workloads or providing air-gapped API security for highly regulated industries, the platform approach is becoming the only viable way to manage the “geo-repatriation” of data and the rise of the agentic economy.

The bottom line

The “Age of AI” is quickly becoming the “Age of Complexity.” The winners won’t just be the companies with the best models, but those with the most resilient, observable and converged delivery platforms. As F5 moves to make its entire stack — from BIG-IP to NGINX to Distributed Cloud to a singular, intelligent fabric — it is allowing its customers to simplify an increasingly complex environment and start security for the AI-first future.

Zeus Kerravala is a principal analyst at ZK Research, a division of Kerravala Consulting. He wrote this article for SiliconANGLE.

Photo: Robert Hof/SiliconANGLE