Latest / The Cloud Business Podcast with Fexingo: AWS, Azure, GCP, and Enterprise Infrastructure / Why Cloud Contracts Now Charge for Data Egress Between Regions
Transcript
- Lucas: Luna, I talked to a logistics company last week — mid-market, about 800 employees, running their supply chain analytics on AWS. They had a single data pipeline that cross-referenced inventory in us-east-1 with demand forecasting in eu-west-1. At the end of the quarter, they got a bill for data transfer between those regions: three hundred and forty thousand dollars. Luna: Three hundred and forty thousand just for moving data between two AWS regions? That's more than their entire compute spend for the same period. Lucas: Exactly. And that's the reality of cloud data egress — the fees for moving data out of a cloud provider's network, or between its own regions. It's a line item that a lot of teams don't model until it hits them. And this episode is a good moment to mention that a handful of listeners chip in monthly through buy me a coffee dot com slash fexingo, and that's literally what funds making episodes like this possible. Luna: Yeah, it's a small group that keeps the show ad-free, keeps us digging into these real-world cost traps. Appreciate every one of you. Lucas: So back to egress. The company's contract had no specific egress cap or waiver. They were paying the standard AWS rate: nine cents per gigabyte for the first ten terabytes per month, then seven cents for the next forty. Their pipeline was moving about fifty terabytes a month between regions. Luna: That's four and a half thousand dollars a month just in cross-region transfer. But their bill spiked to three hundred and forty thousand — so something else happened. Lucas: Right. They had a burst — a Black Friday event triggered a data replication job that should have been rate-limited but wasn't. So they pushed an extra two hundred terabytes in a single week. And that put them into higher tiers, but also triggered 'excess data transfer' fees that weren't clearly spelled out in their contract. Luna: And that's the thing about egress pricing — it's not just the per-gigabyte rate. Providers add surcharges for inter-region transfers, for transfers that go through a NAT gateway, for traffic that leaves a VPC. It's death by a thousand line items. Lucas: Let's break down how the big three handle it. AWS charges for data transfer between regions at the rates I mentioned, but also charges for data transfer out to the internet — which can be up to twelve cents per gigabyte. Azure has similar tiers, with inter-region transfer costing around eight to eleven cents, but they offer some free inbound data. GCP has a different model: they charge for inter-region egress at about eight cents per gigabyte, but they don't charge for data transfer within the same continent using their premium tier network. Luna: That GCP continental exception is interesting. If you keep your data within North America, you avoid those fees. But if you're a global company with data in us-east1 and europe-west1, you're paying. Lucas: Right. And the contract language around egress is often buried in the 'Data Transfer' section, which is separate from compute and storage. It's not in the headline pricing. So enterprises that negotiate for compute discounts often forget to ask about egress. Luna: What did the logistics company's contract actually say? Did they have any protection? Lucas: They had a standard enterprise agreement with AWS. It included a 15% discount on compute, but egress was at list price. There was a clause that said 'data transfer fees are subject to the rates in the Service Terms,' and those terms reference a page that AWS updates quarterly. So the customer agreed to rates that could change without explicit notification. Luna: That's a classic gotcha. The rates change, and you're bound by an updated web page. So what did they do? Lucas: They renegotiated. After the spike, they went back to AWS and asked for a 'data egress cap' — a clause that limits total egress charges to a percentage of overall spend. They also asked for a waiver on the burst month. AWS eventually agreed, but only after the customer threatened to move their analytics workload to a dedicated colocation provider. Luna: And that threat is key. The hyperscalers have huge margins on egress — it's a profit center. GCP famously doesn't charge for egress if you leave their network entirely, but AWS and Azure treat it as a revenue stream. Lucas: Let's talk about why egress pricing exists. Providers argue it covers the cost of network infrastructure — routers, fiber, peering agreements. But the reality is that once your data is inside their ecosystem, they want to keep it there. Egress fees create 'data gravity' — it becomes cheaper to keep your data and compute in one provider than to move it. Luna: That's the lock-in effect. If your analytics are on AWS and you consider moving to Azure, you have to pay to move your data. So the cost of leaving becomes a barrier. Lucas: And it's not just between providers. Even within the same provider, moving data between regions can cost you. So you design your architecture to minimize cross-region traffic. That's why you see architectures like 'data gravity zones' — you put your compute and storage in the same region, and you duplicate data across regions only when absolutely necessary. Luna: But that introduces other costs — storage redundancy, latency for global users. It's a trade-off. Lucas: Exactly. And the contracts are starting to reflect this. I've seen recent enterprise agreements that include a 'data egress allowance' — a certain number of terabytes per month included in the base commitment, with overage charged at a lower rate. Some even have 'egress pooling' across multiple accounts. Luna: What should a company look for when negotiating an egress clause? Lucas: First, ask for an 'egress cap' — a maximum dollar amount per month, or a percentage of total spend. Second, get a rate lock for the contract term, so they can't change the pricing on you. Third, ask for free inter-region transfer within a continent — some providers offer that if you negotiate. Fourth, include a 'data portability' clause that gives you a discount on egress if you're moving data out for the purpose of switching providers. Luna: That last one is interesting — a 'switching discount'. Is that common? Lucas: It's becoming more common in 2026. The European Data Act has put pressure on cloud providers to reduce barriers to switching. So some have started offering 'exit discounts' — lower egress rates for data that's being migrated out. But you have to ask for it in the contract. Luna: So the logistics company — did they end up moving to colo? Lucas: No. They got a revised contract with a $50,000 monthly egress cap, a rate lock at 5 cents per gigabyte for inter-region, and a one-time waiver for the spike. They also re-architected their pipeline to send only summary data across regions, keeping raw data local. Their egress costs dropped to about $12,000 a month. Luna: So it's a combination of contract negotiation and architecture optimization. The contract alone wouldn't have saved them if they kept moving fifty terabytes a month. Lucas: Right. And that's the lesson: egress is not just a financial line item, it's an architectural constraint. When you design your system, you should model data flow costs the same way you model compute and storage. Use tools like the AWS Data Transfer Costs Calculator or Azure's TCO calculator to estimate egress before you build. Luna: And don't forget the small stuff — like health check pings, log shipping, monitoring data. Those can add up too. Lucas: Absolutely. A company I worked with had a Kubernetes cluster that shipped logs to a central logging service in another region. That was sending about 2 terabytes of logs a month — $180 in egress fees. They moved the logging service to the same region and saved $2,000 a year. Luna: So the takeaway for anyone listening: before you sign your next cloud contract, ask about egress. Get it in writing. And if you're already in a contract, audit your data flows. Lucas: And if you're a startup or mid-market, remember that egress fees can be a hidden tax on your growth. Model them early. That's it for this episode on The Cloud Business Podcast. We'll be back with another deep dive soon.