Nobody chooses CloudFront; teams default to it. Nobody defaults to Fastly; teams choose it. That asymmetry is the entire comparison.
The gravity argument
CloudFront wins by adjacency: same bill, same IAM, free egress from AWS origins, one throat to choke. For AWS-native teams shipping standard workloads, the convenience is real and the integration cost of anything else must be argued for. Defaults are not villains; unpriced defaults are. The integration convenience is worth an actual number, and the number deserves to be compared against what the alternative earns.
The velocity argument
Fastly wins where delivery is part of the product: instant config deploys against CloudFront’s slower propagation, real edge programmability, and dynamic-content performance that developer-led teams measure and keep. Its request-based pricing suits media and API shapes differently, which is exactly why it deserves a quote rather than a table. Propagation speed sounds like a developer nicety until the day a bad rule needs reverting during an incident, at which point seconds versus minutes becomes a business metric.
There is also an organizational-boundary effect worth naming. CloudFront lives inside the AWS bill, which means its costs flow through cloud budget governance, tagged, allocated and reviewed with everything else. A separate CDN contract creates a procurement surface: a negotiation, a renewal calendar, a vendor relationship. Some organizations count that surface as overhead; sophisticated ones count it as leverage, because a line you negotiate is a line that responds to negotiation. The AWS-internal line, by contrast, prices by published tier and moves for almost no one. Convenience and leverage are the same coin, facing differently.
The honest tiebreak
If your team would never touch edge logic, gravity usually wins. If your roadmap lives at the edge, velocity usually wins. And the egress economics that make CloudFront look free deserve an actual calculation, because free egress at a higher effective rate is not free. The egress interaction is the most misunderstood line in the comparison: origin-fetch economics differ structurally between the two, and the difference scales with your miss rate.
In practice
Compute three numbers before deciding: your monthly AWS egress with and without each candidate’s origin-fetch treatment, your team’s realistic edge-logic roadmap for the next year, and each option’s effective per-GB rate at your request profile. The decision usually falls out of the first and third numbers; the second one tells you whether to regret it.
We price both against your profile in the assessment, including the AWS egress math nobody shows.