The pitch for cloud computing was, at its heart, a promise of liberation. Stop buying servers, stop running data centres, stop sinking capital into hardware that is obsolete the moment it is installed. Rent computing as a utility, scale up and down at will, and pay only for what you use. The economics were compelling and the migration was vast. But a less advertised consequence has settled in alongside the benefits: many organisations have discovered that the cloud they moved to is far harder to move out of than they expected.
This is the phenomenon of cloud lock-in, and it is one of the most important and least understood dynamics in enterprise technology. It is rarely the product of a coercive contract. It is the cumulative result of sensible decisions that, taken together, make leaving expensive, slow and risky — a dependence built by gravity rather than by force.
How Dependence Accumulates
Lock-in is best understood not as a single trap but as several overlapping ones, each reasonable on its own terms.
The first is proprietary services. The major cloud platforms compete by offering ever more sophisticated managed tools — databases, analytics engines, machine-learning platforms, serverless functions — that handle complexity so customers do not have to. These services are genuinely valuable and genuinely convenient. But many are unique to their provider, with no direct equivalent elsewhere. An application woven through a provider’s bespoke services cannot simply be picked up and dropped onto a competitor; it has to be partly rebuilt.
The second is data gravity. As an organisation accumulates data in a provider’s storage, that data exerts a pull: the applications that use it want to be nearby, and the more data there is, the more costly and disruptive it becomes to move. Petabytes do not relocate cheaply or quickly.
The third, and most human, is skills and process. Teams learn a provider’s tools deeply. Operations, monitoring, security and deployment all become tailored to one platform. Switching means retraining people and rebuilding institutional muscle memory — a cost that never appears on an invoice but is often the largest of all. None of these forces is sinister; together they explain why businesses so often find themselves more committed than they intended to be.
The Egress Fee Problem
If lock-in mostly operates through soft friction, one mechanism is sharp and explicit enough to have drawn the attention of competition regulators: egress fees, the charges providers levy to move data out of their platform.
The asymmetry is striking. Moving data into a cloud is typically free, and storing it is comparatively cheap. But pulling large volumes back out — to migrate to a competitor, or even to run a service across more than one provider — can carry meaningful per-gigabyte charges. Critics argue this is not merely a cost-recovery measure but a deliberate barrier to exit: the more data a customer wishes to take elsewhere, the more they must pay for the privilege of leaving.
This has become a live issue for authorities. Competition bodies including the UK Competition and Markets Authority and the European Commission have examined the structure of the cloud market and the role such charges play in dampening switching. The concern is straightforward: a market where it is easy to join but costly to leave is one where competition on price and quality weakens over time, because providers face less pressure from the threat of customers walking away. These are precisely the questions that animate debates over market power in the digital economy.
The Limits of the Obvious Fixes
The instinctive response to lock-in is to avoid depending on any single provider, and there are recognised strategies for doing so. But each carries a cost that complicates the picture, and pretending otherwise would be dishonest.
A multi-cloud approach spreads workloads across several providers to preserve leverage and resilience. It works, but it sacrifices much of the convenience that made the cloud attractive: teams must master multiple platforms, and software written to the lowest common denominator forgoes the powerful proprietary services that justified the move in the first place. Open standards and portable technologies — using widely supported tools rather than provider-specific ones — reduce dependence by keeping systems movable. Yet they too can mean leaving performance, features or efficiency on the table.
The honest conclusion is that lock-in is a trade-off to be managed, not a problem to be eliminated. Deeper integration with one provider buys efficiency and capability at the price of flexibility; insisting on portability buys freedom at the price of some of that efficiency. There is no setting that wins on every axis, and the right balance depends on how much an organisation values the option to leave.
What’s at Stake
The cloud market’s structure matters well beyond the IT department. A handful of providers now sit at the foundation of a large share of the digital economy — running the services of governments, banks, hospitals and countless businesses. When switching is hard, that concentration becomes self-reinforcing, with implications for pricing, for innovation and for systemic resilience if a dominant platform suffers an outage.
For decision-makers, the practical lesson is to treat exit costs as a first-class consideration from the outset rather than a problem discovered too late. That means weighing the convenience of proprietary services against the dependence they create, negotiating with the cost of leaving in mind, and designing systems with an eye to portability where it genuinely matters. For regulators, the question is whether the market’s frictions are fair competition or quiet entrenchment. Either way, the era of treating the cloud as simply rented freedom is over; what it really offers is a sophisticated dependence, and the organisations that thrive are the ones that enter it with their eyes open.
Frequently Asked Questions
What causes cloud lock-in?
Lock-in accumulates from several overlapping factors: reliance on a provider's proprietary managed services that have no direct equivalent elsewhere, data gravity as ever more data makes migration costly, and the specialised skills and processes teams build around one platform. None is coercive on its own, but together they make leaving expensive, slow, and risky.
What are egress fees and why do they matter?
Egress fees are the charges cloud providers levy to move data out of their platform. Because getting data in is typically cheap while getting it out is not, these fees create a direct financial barrier to leaving and have drawn the attention of competition regulators.
What is data gravity in cloud computing?
Data gravity describes how accumulated data exerts a pull: applications that use the data want to be near it, and the more data an organisation stores with a provider, the more costly and disruptive it becomes to move. Petabytes do not relocate cheaply or quickly.
Why is it hard to switch cloud providers?
Switching is hard because applications woven through a provider's bespoke services must be partly rebuilt, large volumes of data are expensive and slow to move, and teams must be retrained on a new platform. Egress fees add a further explicit cost on top of this soft friction.