On January 12, 2027, it becomes illegal for any cloud provider serving EU customers to charge you a fee for leaving. Switching charges and the egress fees bundled into them disappear across IaaS, PaaS and SaaS on that single date, under the EU Data Act’s switching and porting rules. If you run a technology budget, you have probably seen the headline. You may also have quietly assumed the egress line on your monthly bill is about to shrink.
It is not. The ban is real and it matters, but it touches a narrow slice of the egress you actually pay for. The teams that read the rule carefully will use it as leverage at renewal. The teams that read the headline will budget for a saving that never arrives.
What the ban actually covers
The Data Act targets the cost of getting out, not the cost of operating. Article 29 stops providers from imposing switching charges on the customer for the switching process, and that includes the data egress you incur while moving your workload to another provider or back on-premises. Until the January 2027 deadline, providers sit in a transition window where they may still levy reduced switching charges, capped at the direct costs of facilitating the move. After it, the number goes to zero.
The mechanics come with guardrails. A provider’s notice period is capped at two months, the transitional period for the actual switch defaults to 30 days, and IaaS providers have to deliver “functional equivalence” on the far side so your workload still runs. Those obligations are spelled out in the switching rules under Articles 23 through 25. This is a genuine structural change. The one-time exit tax that made a provider migration look expensive on a spreadsheet is being legislated out of existence.
The hyperscalers saw it coming. Microsoft announced free egress for customers leaving Azure back on March 13, 2024, layered on top of the first 100 GB of monthly egress it already gives away. AWS moved the same quarter, offering free data transfer out to customers who are exiting entirely, issued as credits once support approves the request. Google went first. So the exit-fee relief has been available in practice for two years already, on request. What January 2027 does is make it automatic and non-negotiable, EU-wide, for every provider serving European customers regardless of where the provider is headquartered.
The three egress lines it does not touch
Here is where the budgeting mistake lives. Your egress bill is not one number, it is at least three, and the ban only reaches one of them.
The first is operational egress: the per-gigabyte charge for serving content to your users, streaming an API response, pushing a video, answering a request from the public internet. That day-to-day traffic stays fully billable under normal commercial terms. For most consumer-facing and API-heavy businesses, this is the bulk of the egress bill, and the Data Act does not mention it.
The second is cross-region replication: moving data between regions for resilience, latency or data-residency reasons. That is an architectural choice you made, not a switch away from the provider, so it keeps its price tag.
The third is the subtle one: multi-cloud, or what the Act calls in-parallel use. If you run workloads across two providers and data flows continuously between them, Article 34 treats that egress as permitted indefinitely, chargeable at cost with no margin. A provider cannot mark it up, but it does not have to give it away. The distinction between “I am leaving” and “I run here and there at the same time” is doing enormous work in this regulation, and it is exactly the distinction a vendor’s finance team will lean on.
Put plainly: if your egress spend is dominated by serving customers and replicating across regions, January 2027 changes almost nothing on the invoice. Model a zero-dollar egress saving into next year’s budget on the strength of this law and you will be explaining the variance in Q2.
Why the real win is at the negotiating table
The value of this change is not on the cost line. It is in your leverage.
For fifteen years the egress fee did quiet work as a lock-in mechanism. The threat was never stated; it did not need to be. Everyone knew that pulling a few hundred terabytes out of a provider carried a five- or six-figure exit charge, and that number sat in the back of every renewal conversation, shrinking the customer’s willingness to walk. Remove the exit charge and you remove the club the incumbent was holding under the table.
Running network operations at a large telecom, and later doing fractional COO work through Ops Harmony, I have sat on both sides of contracts built on exactly this asymmetry. The vendor’s pricing power was never really about the sticker; it was about how painful and expensive it looked to leave. The moment a credible exit costs nothing, the renewal conversation changes character. You can put a competing offer on the table without first burning budget to prove you could act on it. That is the same dynamic that makes outcome-based and pay-per-use pricing shifts so disruptive to incumbents: they strip out the friction that used to keep customers in place.
Do not expect the providers to absorb the loss quietly. Legal analysts already warn that vendors will recover foregone switching revenue through higher baseline pricing rather than eliminate the cost outright. The exit fee does not vanish from your total cost of ownership so much as move upstream into the rate card, where it is harder to see and harder to attribute. FinOps and vendor teams that only track the egress line will miss it. The ones tracking effective unit cost across the whole contract will catch the shuffle.
What to do before January
Three moves, none of which require waiting for the deadline.
Audit your exit cost now. Most teams have never modeled what a full migration off their primary provider would actually cost, because the egress charge alone made the exercise feel academic. It is not academic anymore. Knowing the number, and knowing it is about to be legislated toward zero, is the precondition for using it. This is the same exit-readiness discipline that underpins any serious cloud repatriation or provider-switch decision.
Read your renewal for baseline creep. If a provider quotes a new agreement between now and January 2027, scrutinize the base rates against your current deal. A quiet few percent added to compute or storage is how the switching revenue comes back. Name it in the negotiation.
Separate your egress by type in reporting. If your cost tooling reports egress as a single line, split it into operational, cross-region and switching-related before you brief finance. The version of the story where “egress fees are being abolished” and the version where “our egress bill is flat” are both true, and the only way to tell that story without looking wrong is to have the breakdown ready. If you have not tagged spend to that level yet, it is the same cloud waste and allocation hygiene that makes every other cost conversation legible.
The Data Act is a win for cloud customers. It is a lock-in win, not a bill-reduction win, and treating it as the latter is how a good regulatory change turns into a bad forecast. For a refresher on how these disciplines fit together, our plain-English guide to FinOps and our walkthrough of Azure cost management both cover the tagging and allocation groundwork you will need to read this change correctly.