The business
Cloudflare runs a network in hundreds of cities that sits in front of its customers' websites, apps and offices. Traffic passes through it for security (DDoS protection, web application firewall, bot management), performance (content delivery) and zero-trust networking (replacing corporate VPNs). On top of the same network it sells a developer platform (Workers, storage and AI inference at the edge).
Revenue in FY2025 was $2.2B, up 29.8%.
Moat and switching costs
The architecture is the advantage: every server runs the full software stack, so new products ship to the whole network at once and run cheaply on capacity already paid for. A free tier brings millions of developers and websites, which feeds both the threat intelligence and the sales funnel.
Network services can be swapped for rivals (Akamai, Zscaler, the hyperscalers' own offerings) more easily than an application platform, so stickiness depends on customers adopting several products and building on Workers.
How the technology shows up in the numbers
Gross margin near 75%; free cash flow of $287M in FY2025 (13% of revenue). Stock pay was $451M, larger than the free cash flow, so owner earnings were negative (about -$164M) and the GAAP operating margin was -9.6%. The diluted share count grows about 2% a year. Convertible notes are the only debt; net cash was $836M.
Thesis
If the developer platform and zero-trust products make Cloudflare a default layer of the internet, the network's fixed costs turn high growth into high margins later. Today the evidence for owner economics is missing.
We are watching for margin progress before owning.
Risks
- Dilution continuing at 2% a year with no buybacks.
- Hyperscalers bundling equivalent services.
- Valuation: the price has typically assumed many years of high growth.
What would change our mind
- Free cash flow exceeding stock pay for a full year (a reason to own).
- Growth falling below 20% before margins expand.
- Large-customer growth stalling.
