Latest / The Cloud Business Podcast with Fexingo: AWS, Azure, GCP, and Enterprise Infrastructure / The Unspoken Cost of Cloud Egress Fees
Transcript
- Lucas: If these conversations are useful for what you're building or running, we hope you stick around. We want to talk about something that quietly eats a hole in a lot of cloud budgets: egress fees. Luna: Egress — that's the cost of moving data out of a cloud provider, right? It's not the compute or storage that gets most of the attention. Lucas: Exactly. And it's one of the most opaque line items on an enterprise cloud bill. AWS, Azure, Google Cloud — they all charge for data leaving their network. The rates vary, but they're non-trivial. For example, AWS charges around nine cents per gigabyte for the first ten terabytes transferred out to the internet each month. Luna: And that adds up fast. I remember reading about a company that moved a petabyte of data out of AWS for a migration — their egress bill alone was over ninety thousand dollars. Lucas: Right. And that's the thing — egress fees don't scale linearly in a way that's easy to predict. They're based on cumulative monthly transfer. So a company running a multi-cloud architecture where workloads talk to each other across providers can see a real shock. Luna: What about the argument that these fees are a form of lock-in? That they discourage customers from leaving or mixing providers? Lucas: That's exactly the criticism. And it's not just from customers. European regulators have been looking into this. In fact, the UK's competition authority and the European Commission have both flagged egress fees as a barrier to cloud switching. Google Cloud actually responded last year by eliminating egress fees for customers who want to leave their platform — though with some conditions. Luna: So Google made a strategic move there. Put pressure on AWS and Azure. But how meaningful is it in practice? Lucas: It's meaningful as a signal, but the actual cost of egress is often just one piece of a larger migration expense. The bigger friction is usually architecture rework, data format changes, retraining teams. Still, egress is the line item that gets people's attention because it's so visible once you look. Luna: Let's talk about a concrete example. You mentioned a mid-size SaaS company earlier — what happened there? Lucas: Yeah. So I talked to the CTO of a company called DataFlow — they're a real company, about 200 employees, running a SaaS platform on AWS. They decided to adopt a multi-cloud strategy, partly for resilience, partly to use some Azure-specific AI services. They started moving workloads gradually. But within three months, their monthly cloud bill jumped forty percent. Almost all of that increase came from data transfer between AWS and Azure. Luna: And they hadn't budgeted for that? Lucas: They had, but they underestimated the volume. Their architecture required databases in one cloud and compute in another — so every query involved egress. They thought they'd move most data over private peering, but the traffic patterns were more complicated. They ended up spending about 150,000 dollars extra over six months before they re-architected. Luna: That's brutal. And it's a reminder that egress isn't just about the cost per gigabyte — it's about architectural decisions that multiply that cost. Lucas: Exactly. And there are workarounds. One is to use Cloudflare or similar CDN providers that participate in the Bandwidth Alliance — they waive egress fees for data going to each other's networks. Another is to use direct connect or private links to keep traffic within the provider's backbone as much as possible. Luna: But those require planning and investment upfront. A lot of teams don't think about egress until the bill arrives. Lucas: Right. And that's the key takeaway for anyone running cloud infrastructure: model your data flows early. Use tools like AWS Cost Explorer or third-party cloud cost management platforms to see what your transfer costs look like. And consider negotiating — especially if you're a large customer. AWS and Azure will sometimes offer egress discounts or credits if you commit to a certain spend level. Luna: Speaking of support — you know, a couple of dollars a month is genuinely what keeps these conversations going. If you've gotten something out of the show, you can help us keep it ad-free at buy me a coffee dot com slash fexingo. Lucas: Yeah, it's a small thing that makes a real difference. And we appreciate everyone who's already contributed. So, back to egress — another angle is the regulatory push. There's been a lot of noise about the European Data Act, which includes provisions on switching cloud providers and making egress fees more transparent. Luna: Right. The Data Act is supposed to remove obstacles to switching, including unfair egress fees. But how much teeth does it have? Lucas: It's still early. The act came into force in early 2024, but implementation is phased. Providers have to make switching easier and fees transparent, but there are exceptions. For example, if the provider can show that egress fees reflect actual network costs, they might not be forced to eliminate them. So the jury's still out on whether it will actually drive down costs for customers. Luna: And what about the argument that egress fees are actually a good thing — that they prevent customers from making frivolous transfers that waste network resources? Lucas: That's the providers' argument. They say egress covers the cost of bandwidth, peering, and infrastructure. But the counter-argument is that they're already making healthy margins on compute and storage, and egress is an additional profit center. Especially when you consider that ingress — data coming into the cloud — is usually free. So the asymmetry is telling. Luna: So what's the bottom line for an enterprise listening today? Lucas: First, audit your data transfer patterns. Use the AWS or Azure billing dashboards to see which services are generating egress. Second, consider architectural patterns that minimize cross-cloud traffic — like putting all compute and data for a given workload in one cloud, or using a CDN for user-facing data. Third, if you're planning a multi-cloud or hybrid setup, model the egress costs before you commit. And finally, negotiate — especially if you're a medium to large customer, you have leverage. Luna: And keep an eye on regulation. If the European Data Act leads to more competitive pricing, that could benefit everyone. Lucas: Absolutely. It's an area where small changes in policy or pricing can have big effects on cloud economics. We'll be watching how the big three respond in the coming quarters.