Software Licensing Is Becoming Cloud Computing’s Next Antitrust Battleground

· · Views: 2,433 · 6 min time to read

Cloud competition is usually discussed in terms of processor performance, storage prices, network latency or the breadth of services offered by AWS, Microsoft Azure and Google Cloud. Increasingly, regulators are looking somewhere less visible: the software licenses sitting above that infrastructure.

The European Commission is now examining Oracle’s licensing practices, with Reuters reporting that regulators are seeking information from third parties to determine whether there is evidence worth pursuing. No formal investigation has been opened, and Oracle had not responded to Reuters’ request for comment.

The scrutiny comes shortly after SAP agreed to change practices that regulators believed restricted customer choice. Although that case concerned maintenance and support for on-premise ERP software rather than cloud infrastructure itself, the European Commission said the outcome should serve as a warning against similar conduct in increasingly important cloud markets.

That connection matters because cloud lock-in can begin long before a company decides where to run its next virtual machine.

A software licence can quietly choose the cloud

Enterprise workloads rarely consist of infrastructure alone. Companies run databases, operating systems, ERP platforms, collaboration software and other commercial applications that carry their own licensing terms.

Those terms can alter the economics of deploying the same workload on different clouds.

The OECD’s 2025 study on cloud competition explains that software vendors with market power may use licensing policies that raise competitors’ costs or make rival cloud providers less attractive. It points to restrictions involving bring-your-own-license arrangements, additional charges and terms that can make software more expensive when deployed outside the vendor’s preferred ecosystem.

That creates an architectural constraint disguised as a commercial one.

An engineering team may determine that another provider offers better infrastructure pricing, geographic availability or specialized services. But if moving an important database or server workload means purchasing additional licenses, losing existing discounts or changing support agreements, the technically superior platform can become economically irrational.

Cloud choice is therefore partly encoded in contracts.

Lock-in is a stack of switching costs

Software licensing is only one layer of a wider lock-in problem.

Ofcom’s cloud market study identified egress fees, committed-spend discounts and technical restrictions as features that can make switching cloud providers or adopting multiple clouds more difficult. It also specifically examined whether software licensing practices discourage customers from using rival infrastructure providers.

Those barriers compound each other.

Moving a production system can require transferring large datasets, rewriting applications around different APIs, retraining engineers, rebuilding security and identity controls, renegotiating contracts—and then discovering that a critical software licence becomes more expensive on the destination platform.

Research submitted to an OECD competition discussion offers a useful indication of how companies experience those costs. Among Danish cloud customers surveyed, 14% identified licensing as a barrier to switching, while the proportion rose among Azure customers. Technical barriers, other switching costs and limited price transparency were also frequently reported.

The significance is not that licensing is always the largest barrier. It is that several moderate barriers can combine to make migration prohibitively expensive.

Regulators increasingly see a competition mechanism

Economically, switching costs matter because they weaken the pressure customers can place on suppliers.

If moving becomes sufficiently painful, an incumbent provider may have greater freedom to increase prices, alter contract terms or reduce service quality without immediately losing customers. New cloud providers face the inverse problem: even a better product may not be enough to persuade customers to absorb the cost of leaving an existing ecosystem.

The OECD describes this as a potential form of market-power leveraging, where control over widely used software can influence competition in the adjacent cloud infrastructure market. Its analysis says restrictive licences may create financial disincentives for customers using rival providers and increase barriers faced by competitors.

The UK Competition and Markets Authority reached a similar conclusion after its multiyear cloud investigation. The CMA found that technical and commercial barriers can lock customers into their initial cloud choices and said Microsoft’s position in software made it harder for AWS and Google to compete for customers that needed Microsoft products.

By March 2026, the CMA said its final cloud investigation had identified software licensing, egress fees and interoperability barriers as limits on customer choice. It subsequently opened a strategic market status investigation into Microsoft’s wider business software ecosystem.

SAP shows why contract design matters

The SAP case provides another example of regulators examining commercial mechanisms that make leaving harder.

Reuters reported that SAP agreed to make it easier for customers to switch to rival service providers or terminate contracts. Its commitments include an alternative method for calculating licence fees, removing reinstatement fees and reducing back-maintenance charges for returning customers. The commitments apply globally for 10 years.

EU competition chief Teresa Ribera said the changes give SAP customers more freedom to choose maintenance and support providers without restrictions that increased costs and suppressed competition. The European Commission has made those commitments legally binding.

The lesson for cloud markets is straightforward: switching does not have to be technically prohibited to be commercially discouraged.

Cloud architecture is becoming a competition-policy issue

For engineering leaders, licensing can easily appear to belong to procurement or legal teams. That separation is becoming harder to defend.

A licence determining where database software can run, how many processor cores must be paid for, whether an existing entitlement can move to another provider or what support fees apply after migration can materially reshape infrastructure architecture.

For cloud vendors, those rules can determine whether superior performance translates into actual customer wins.

And for regulators, the concern is increasingly about whether companies controlling one important layer of the technology stack can use that position to influence another.

The European Commission’s current Oracle scrutiny may ultimately produce no enforcement action. Reuters makes clear that regulators are still gathering information rather than conducting a formal investigation.

But the broader regulatory direction is difficult to miss. The Commission’s cloud discussions now explicitly cover interoperability, financial conditions and contractual practices that may affect switching, while authorities in Britain and elsewhere are examining similar constraints.

The next phase of the cloud wars may therefore be decided by more than chips, regions and benchmark performance.

Sometimes the most powerful piece of infrastructure is the licence agreement that determines where everything else is allowed to run.

Share
f 𝕏 in
Copied