PIPE FOR S3 · EDGE ACCELERATOR

S3 egress
without the AWS bill.

Keep S3 as your source of truth. Put Pipe in front. Cut AWS egress by 95%+ on the first request, and never pay it again. One DNS change. Three minutes of setup. No migration.

PIPE NETWORK Edge S3 AWS

Three minutes of setup. Zero migration risk.

01 ——

Grant read access

Create a cross-account IAM role with s3:GetObject scoped to your bucket. Pipe assumes it via STS with a unique external ID. Your keys never leave AWS.

aws iam create-role \
  --role-name PipeReadOnly
02 ——

Point a hostname at us

Change your asset URL from bucket.s3.amazonaws.com to yours.pipenetwork.com. That's it. S3 stays your source of truth.

CNAME → edge.pipe.network
03 ——

Watch the bill collapse

Pipe fetches each object from S3 exactly once on first request, then serves it from the nearest of our 1,000+ POPs across 34+ countries. Hit rates above 95% are typical within hours.

→ 94.3% egress saved
S3 Origin Source of truth Pipe Edge 1,000+ POPs · 34 countries READ-THROUGH CACHE End Users Global audience ~5% cold 95%+ edge
Billing & Cost Management

Monthly Bill Summary

April 2026
ServiceAmount
Amazon S3 — Data Transfer Out (100 TB)$9,000.00
Amazon CloudFront — Distribution$8,500.00
Amazon S3 — Request Fees$1,200.00
Pipe Network — Cold Origin Reads (5%)$450.00
Pipe Network — Edge Delivery$620.00
Previous total: $18,700.00
$1,070.00
with Pipe Network
-94.3% $17,630 saved per month

See what happens to a real AWS bill.

A typical customer serving 100 TB/month from S3. Same data, same users, same bucket — just Pipe in front.

No code changes — one DNS update
S3 stays your source of truth
95%+ cache hit rate within hours

Built for scale, priced for sanity.

95%+
Typical cache hit rate
1,000+
Points of presence
34+
Countries covered
3 min
To first byte served

Simple pricing. No egress surprises.

Origin Cache Fill
$10
per TB

Cold misses fetched from your S3 bucket. Only ~5% of requests hit origin after warm-up. Standard AWS egress rates pass through.

Enterprise
Custom
 

Volume discounts, committed-use pricing, SLAs, dedicated support, and custom cache policies. Contact sales

Answered, up front.

Does Pipe own or move my data?
No. S3 stays your source of truth. Pipe is a read-through cache — we fetch objects on demand, serve them, and hold them only as long as cache policy dictates. Delete your bucket and Pipe has nothing.
What about credentials?
Cross-account IAM role assumption with a unique external ID per customer. Pipe never holds long-lived keys to your account. Roles Anywhere with X.509 is available for customers who want zero persistent AWS secrets anywhere in our infrastructure.
How do I invalidate stale content?
Three ways: TTL (per bucket or prefix), purge API (single object or wildcard), and S3 event notifications via SNS/SQS for automatic invalidation within seconds of a PUT. Upload to S3, Pipe refreshes — no manual step.
Range requests, HLS, large assets?
Supported from day one. Video segments, progressive MP4, game patches, ML model weights — anything that speaks HTTP GET with Range headers works against the origin fetch path. Tested at multi-GB object sizes.
What's the catch on the savings number?
The savings depend on cache hit rate. Static assets, video, software distribution, and ML artifacts typically see 95%+ hit rates within hours. Unique, one-shot downloads see less. Our dashboard shows cold-fetch bytes separately from served bytes so you always know where you stand.

Stop paying AWS to serve your own files.

Pipe for S3 is in limited availability. We onboard one customer per week and walk you through the IAM role, the DNS cutover, and the first week of hit-rate tuning personally. Talk to us.