Can Rising Adoption of Agentic AI Accelerate NET's Revenue Growth?

Cloudflare reported that more than 50% of Q2 2026 network traffic was nonhuman, while its Workers platform added nearly 2 million developers to exceed 7.4 million, underscoring accelerating agentic-AI adoption. Q2 revenue rose 36% year over year and dollar-based net retention reached 120%; consensus forecasts revenue growth of 32.5% in 2026 and 28.6% in 2027. New AI-agent transaction products could expand monetization, but competition from Zscaler and Palo Alto Networks remains intense and NET's 34.21x forward sales multiple is far above the industry's 4.20x average.

Analysis

Agentic workloads are incrementally better for edge-network vendors than traditional SaaS because machine-to-machine traffic is persistent, latency-sensitive and security-intensive. The key question for NET is not developer-seat growth but conversion: whether Workers usage, inference routing and transaction tooling lift revenue per customer faster than bandwidth and compute costs. A sustained increase in large-customer expansion and gross-margin stability would justify a premium; absent that, AI traffic can become a low-margin volume story rather than a new software profit pool.

PANW is the cleaner enterprise monetization vehicle over the next 1-3 quarters: agent security can be bundled into an installed platform, reducing procurement friction and supporting platform consolidation. ZS has comparable thematic exposure but faces a more concentrated execution test around converting new agent-security modules into paid workload adoption. NET is more exposed to competitive unbundling: hyperscalers can subsidize developer and edge services, while PANW/ZS can own the security control plane even when traffic runs through Cloudflare.

The market is likely assigning NET substantial optionality to payment and agent-commerce products before there is evidence of transaction take-rate, payment-loss controls, or meaningful GMV. At its current valuation, even a modest deceleration in expansion, gross-margin pressure from AI compute, or a weaker forward revenue guide could drive disproportionate multiple compression over 1-3 months. Conversely, the bearish view is falsified if NET demonstrates two consecutive quarters of accelerating large-enterprise growth, net retention re-acceleration, and stable-to-expanding gross margin while monetized AI products become material to bookings.

Contrarianly, the most durable AI beneficiary may be security vendors rather than traffic intermediaries. Autonomous agents enlarge identity, data-loss and authorization attack surfaces; enterprises will pay to govern them before they pay for agent-to-content micropayments. That favors PANW's platform economics and, selectively, ZS, while NET remains a higher-beta expression best owned only after monetization KPIs—not developer metrics—validate the thesis.

Key Decisions for Investors

• Initiate a 3-6 month pair: long PANW / short NET in roughly beta-neutral dollars. Target 15-20% relative outperformance as PANW captures higher-margin security spend while NET faces elevated execution and valuation sensitivity; cover if NET reports accelerating enterprise expansion plus gross-margin expansion for two consecutive quarters.
• Keep ZS on a catalyst watch rather than add aggressively: buy only after evidence that agent-security products contribute to net-new ARR or billings growth, using the next two earnings reports as the validation window. A reacceleration in billings and large-customer additions would support upside; continued sales-cycle elongation invalidates the setup.
• For existing NET longs, reduce exposure into the next earnings event or hedge with 3-month downside puts financed by out-of-the-money calls. The asymmetric risk is a guidance miss or margin pressure causing multiple compression; remove the hedge only if management quantifies paid AI-product revenue, usage conversion, and unit economics.
• Avoid using QBTS as an agentic-AI read-through. Its earnings sensitivity is primarily to quantum commercialization and financing conditions, not enterprise edge, developer-platform, or AI-security spend.