Cloudflare launches Cloudflare Workers, an edge computing service for developers using its network, charging devs $0.50 for every 1M tasks used by their apps
Context & Ripple Effects
In March 2018 Cloudflare turned its content-delivery network into a programmable platform: Workers lets any developer run code on Cloudflare's edge servers, priced at $0.50 per million tasks rather than by reserved server capacity. The move repurposed infrastructure Cloudflare already operated for security and caching into a billable compute product.
That pricing-and-placement bet aged well inside the company's own roadmap: the same Workers runtime later carried pay-as-you-go GPU access for AI models via Workers AI, and by 2026 it underpins Kitesurf, a cloud-hosted browser for AI agents. The 2018 launch is the foundation layer of that arc.
First-order effects
- Developers gain serverless execution at Cloudflare's edge with usage-based pricing, removing the need to provision servers for latency-sensitive logic near users.
- Cloudflare opens a new revenue line on existing hardware, converting network capacity that previously served only CDN and security traffic into metered compute.
Second-order effects
- Hyperscale serverless offerings face a rival whose unit is the task executed close to the user, pressuring incumbents to compete on per-invocation price and geographic reach.
- Cloudflare's later push to cut networking costs alongside Microsoft, Google, and Backblaze through the Bandwidth Alliance extends the same playbook — using scale to undercut cloud economics across adjacent lines.
Third-order effects
- If edge runtimes keep absorbing new workloads, distribution networks consolidate into general-purpose compute platforms — a reversal in which the network operator, not the centralized cloud, hosts application logic, inference, and now agent tooling.
The trend: Content-delivery networks are evolving into distributed compute platforms, with the edge runtime becoming the product rather than the plumbing.