AI will write more code. It won’t rewire the edge overnight.
Everyone’s trying to jam “AI” into every ticker. Cloudflare doesn’t need the costume. This is an edge network + security control plane business. AI can help customers generate more traffic. But it doesn’t magically delete the need for global routing, DDoS absorption, and policy enforcement close to users.
Cloudflare (NET)
Quick verdict: A real network business with real switching costs. Not untouchable, but not getting “GPT’d” either.
ANCHOR Score: 44 / 60
Badge: ABIP ANCHOR Certified (clears gates: Total ≥ 40, H ≥ 6, N ≥ 6)
10-second thesis
Cloudflare sells “one network” security + performance at the edge. Once your traffic, policies, and apps run through it, ripping it out is pain. AI is mostly a tailwind (more traffic, more security need), not a replacement.
Market narrative (what the bull case says)
“Connectivity cloud” consolidates CDN + WAF + Zero Trust + edge compute into one vendor.
Global footprint means low latency and big DDoS capacity advantages.
Developer platform (Workers) and storage (R2) expand wallet share beyond classic CDN/security.
Enterprise expansion engine is real (retention improved recently).
Reality check (what gets hand-waved)
This is still a competitive knife fight (Akamai, Fastly, hyperscalers, security incumbents). “One network” is a pitch, not a law of nature.
Parts of the product set are features customers can multi-source (especially at mid-market).
AI hype can inflate expectations: model traffic ≠ durable pricing power. Customers will negotiate.
Full scoring breakdown (A / N / C / H / O / R)
A — Asset-Embedded: 7/10
Cloudflare runs a large global edge network where traffic is processed close to source, and many services are delivered “on the network,” not as a bolt-on. That’s infrastructure-adjacent, not just an app.
N — Non-Discretionary: 7/10
Security, uptime, and performance aren’t “nice-to-have” once a business is online. DDoS/WAF/Zero Trust tends to be operationally important, especially during incidents.
C — Capital-Intensive: 6/10
Building and operating a global network footprint across hundreds of cities is not a weekend project. But it’s not a regulated utility either—well-funded competitors exist.
H — Hard to Replace: 7/10
AI doesn’t replace edge routing, DDoS absorption, or distributed policy enforcement. The real risk is pricing compression and feature parity—not extinction.
O — Obsolescence-Resistant: 7/10
The core job—move traffic fast, filter bad traffic, enforce access—doesn’t change every year. Interfaces evolve; the function stays.
R — Real-World Demand: 10/10
This business exists because packets have to traverse physical networks and data centers. It’s “digital,” but it’s still atoms: fiber, power, capacity, locations.
What could go wrong
Enterprise buyer consolidation: big customers standardize on a rival stack (or a hyperscaler bundle) and squeeze pricing.
Feature commoditization: CDN/WAF becomes table stakes; differentiation narrows to price and contracts.
Execution debt: expanding product surface area (Workers, R2, Zero Trust) can dilute focus.
Regulatory / geopolitical friction: operating globally invites compliance and localization pressure.
The setup
If I’m right:
Cloudflare keeps expanding inside enterprises (more products per customer), and the network advantage holds. Retention stays healthy.
If I’m wrong:
The edge stack becomes a commodity. Cloudflare is forced to buy growth with discounts, and margins never scale cleanly.
What would change my mind:
Sustained evidence that large customers can swap Cloudflare out with minimal pain (multi-vendor parity + cheap switching).
A clear hyperscaler move that makes “edge security + delivery” a bundled default at lower effective cost.
AI Impact Label: AI Tailwind
AI increases traffic, attack surface, and the need for enforcement at the edge. That’s more demand for what Cloudflare already sells. The risk is pricing pressure, not replacement.
Closing line: AI compresses cognition. It does not pour concrete. Cloudflare is closer to concrete than cognition.
— Connor
Alpha Before It Prints
© 2026 Alpha Before It Prints
Unsubscribe