Is Cloudflare primarily a cybersecurity stock or an AI infrastructure play?

Both, which is part of the bull case. The Zero Trust and SASE business provides durable revenue growth, while Workers AI positions Cloudflare as an inference infrastructure play without requiring the massive GPU capex outlays of hyperscalers.

What makes R2 storage competitively different from AWS S3?

R2 charges no egress fees. When you retrieve data from AWS S3 at scale, egress costs can rival storage costs themselves. For AI training datasets and model weights accessed across multiple environments, R2's zero-egress pricing produces meaningful total cost differences.

How is Cloudflare's 'pool of funds' enterprise model different from traditional licensing?

Rather than negotiating each product separately, large enterprise customers deposit a committed spend amount and draw it down across any Cloudflare service—CDN, Zero Trust, R2, Workers AI. This raises ARPU and retention while giving customers flexibility to allocate budget to whichever service adds the most value.

NET Stock Outlook 2026: Cloudflare at the Intersection of...

Matthew Prince has been building Cloudflare’s moat layer by layer for fifteen years. It started with DDoS mitigation and CDN—a network that sits in front of the internet and makes it faster and safer. Then came Zero Trust: applying the same always-on network to enterprise security. Now the bet is that the same global edge network, with GPU hardware added at strategic points, can become an AI inference platform that is meaningfully cheaper and lower-latency than running inference on centralized cloud regions.

This is a big, multi-chapter story that requires patience. The network advantage is real. The question for 2026 is whether the enterprise monetization of Workers AI and the Zero Trust revenue acceleration can justify the current valuation multiple.

Workers AI: Edge Inference as a Competitive Position

Workers AI lets developers run AI models on Cloudflare’s network of 300+ data centers, physically close to end users. The latency advantage over centralized inference is most meaningful for real-time applications: chatbots, streaming translation, content moderation, personalization.

Prince has spoken publicly about a deliberate choice not to pursue hyperscaler-scale GPU clusters. Instead of buying tens of thousands of GPUs in a few centralized locations, Cloudflare places smaller GPU deployments across its global network. This keeps capital expenditure manageable and aligns with the edge inference thesis, but it also means Cloudflare is not competing for large model training workloads—it is competing for inference, which is where the volume sits once models are deployed.

Pricing is developer-friendly by design. Workers AI is bundled into the Workers platform at accessible entry points, with usage-based scaling for enterprise volumes. The land-with-developers-then-convert-to-enterprise playbook is exactly what Cloudflare executed with Zero Trust, starting with free tiers and graduating to six- and seven-figure annual contracts with large organizations.

Related: AI Stocks Investment Guide 2026 →

Zero Trust: The Three-Way Race

Cloudflare One is the company’s SASE platform, bundling Zero Trust Network Access (ZTNA), Secure Web Gateway (SWG), Cloud Access Security Broker (CASB), and Magic WAN into a unified service. The competitive set is Zscaler, PANW Prisma SASE, and Cloudflare—three companies with meaningfully different architectures and go-to-market approaches.

Zscaler is cloud-native and has been doing SASE longer than anyone. PANW brings the weight of a massive installed base and a full security product suite. Cloudflare competes on price, global network performance, and a developer-first platform that makes integrations fast. In multi-cloud environments where latency to the nearest Cloudflare PoP matters, the network architecture wins deals that Zscaler and PANW find harder to compete for.

Gartner’s analysis of the SSE (Security Service Edge) market positions Cloudflare as a Challenger—technically strong but with less enterprise go-to-market depth than the leaders. That gap is closing as the enterprise sales team scales.

R2 and the Egress Fee Disruption

Egress fees—what AWS, GCP, and Azure charge when you move data out of their clouds—have been a persistent CTO grievance for years. Cloudflare R2 eliminated them entirely. S3-compatible API, zero egress cost on retrieval.

For AI workloads, this matters in concrete ways: training datasets stored in R2 can be read repeatedly by compute resources without accumulating egress charges. Model weights can be served from R2 to inference endpoints globally without per-GB fees. As AI workloads standardize around models that are updated and served at scale, the economics of R2 versus S3 become a real part of infrastructure budget conversations.

The Enterprise Pool of Funds Model

For large enterprise customers, Cloudflare structures deals as committed annual spend that can be allocated flexibly across the entire product portfolio. A customer might enter with a Zero Trust use case, then pull Workers AI and R2 into scope using the same budget pool as AI projects get prioritized internally.

This model drives meaningful ARPU expansion because the customer’s team has already justified and funded the spend—the friction to add new Cloudflare services is organizational rather than budgetary. Cloudflare’s investor presentations highlight that large customers consistently expand their product usage footprint after entering a pool of funds agreement.

Bull Case vs Bear Case

Bull case

Workers AI enterprise contracts start contributing a distinct ARR line in FY2026 earnings calls
R2 adoption accelerates as AI teams optimize infrastructure costs, pulling more storage workloads away from AWS
Zero Trust market share gains continue, with Cloudflare closing the enterprise go-to-market gap to Zscaler

Bear case

AWS, Azure, and GCP bundle competitive edge AI inference and SASE products into their existing enterprise agreements, reducing Cloudflare’s addressable opportunity
GPU capex requirements increase faster than expected, compressing free cash flow margin improvement
Workers AI remains primarily a developer tool with slow enterprise conversion, delaying meaningful revenue contribution

Related: PLTR Palantir Stock Outlook 2026 →

Tax Considerations for US Investors

NET pays no dividend. In a taxable account, the tax event is entirely deferred until you sell, with gains over 12 months eligible for 15-20% LTCG rates. The stock has historically traded at a premium multiple relative to current profitability, reflecting market confidence in long-term FCF generation rather than near-term earnings.

For investors who believe in the multi-year AI infrastructure and Zero Trust thesis, Cloudflare fits naturally in a 3-5% sleeve of a diversified tech portfolio. Pair it with a more profitable compounder for balance—MSFT or ORCL provide the stable cash flow that NET lacks in the near term.

Bottom Line

Cloudflare’s 2026 setup is compelling for investors who can hold through valuation-driven volatility. The network is genuinely differentiated. The Workers AI inference bet is architecturally sound even if monetization timelines are uncertain. Zero Trust ARR is growing on a durable trajectory. The key milestone to watch: when does Workers AI show up as a discrete, growing revenue contribution in quarterly reports rather than a narrative?

This article is for informational purposes only and is not investment advice.