Buying the Code Is Not Buying Independence
Business · Analysis · 16 November 2026
By iGaming Review
Allwyn’s agreement with Kambi separates control of a sportsbook platform from the services that keep it running. For operators considering a similar move, the important question is which decisions they can make for themselves after the deal.
A source-code licence can give an operator more room to develop its product. Whether it gives the business more freedom depends on what happens when it needs to change a feed, fix a failure or replace a supplier.
Allwyn’s agreement with Kambi, announced on 10 November, makes that distinction concrete. Kambi will license its sportsbook source code globally and in perpetuity, help implement it and build and train a team for later transfer to Allwyn. Implementation is planned for 2027–2029. A separate long-term agreement covers Kambi’s trading and risk-management services.1
This is a route to greater internal capability with external services still in the arrangement. It does not establish that Allwyn has bought Kambi’s intellectual property outright, or that the operator will become self-sufficient. Nor does it suggest that self-sufficiency is necessarily the best objective. An operator may reasonably want control over its product while purchasing specialist services.
Allwyn is a lottery-led group whose UK subsidiary operates the National Lottery.1, 2 The Kambi agreement concerns its sportsbook operations. Allwyn points to product-roadmap control, local-market adaptation and scale from a shared platform. The announcements do not identify mandatory local hosting or national technology audits as drivers.3
In our analysis, the hybrid model offers a way to gain development capability while retaining specialist services. The strategic question is whether that division of responsibility serves each business and market—not whether every component becomes internal.
Define the decision that needs to move
The useful starting point for a board is a business constraint. Perhaps a local feature repeatedly loses priority on a supplier’s roadmap. Perhaps a new data provider cannot be connected without a long development queue. Perhaps the operator cannot explain how a critical decision was made. Each problem calls for a different kind of control.
Owning a development backlog addresses the first problem only if the operator also has the right and the ability to change the relevant software. Connecting a new feed requires usable interfaces and the necessary commercial permissions. Explaining a trading decision may depend on records and models outside the platform code.
“Build, buy or license” is therefore too broad a choice to make once for the whole sportsbook. The more useful exercise is to identify a decision, locate the capability behind it and establish who can change it. The framework below is our analysis for buyers, not a description of Allwyn’s unpublished contract.
A perpetual licence still needs an operating model
Duration answers how long a right lasts. It does not, by itself, settle who owns later improvements, which changes are permitted or who supplies future security fixes. The announcements reviewed for this article do not disclose those provisions.1, 3
For a prospective licensee, the practical test is whether its team can maintain a working product, not merely possess a repository. That means being able to build a release, test it, deploy it, diagnose failures and recover service. Documentation and access to the necessary development tools matter alongside the code itself.
Before approving the deal, the board should ask:
- Rights and ownership: Which changes can our team make, and who owns the improvements it develops?
- Updates and fixes: Who supplies security patches and bug fixes, and can we obtain and apply the supplier’s later updates?
- Compatibility: Who tests changes against our integrations and accepts responsibility when something breaks?
- Funding and support: Who pays for this work, and who supports the platform after the handover?
These obligations need an operating budget and named owners. A perpetual licence does not remove the continuing cost of maintaining a usable product.
Kambi’s planned team-building, training and transfer address the capability gap.1 For a buyer considering this model, a transfer plan is only a starting point: knowledge transfer should be demonstrated through a release completed, an incident rehearsed and a critical integration changed by the receiving team. Building the team addresses the threshold for taking control; funding and retaining it remain ongoing responsibilities.
Follow the recurring cost
Allwyn reports €50 million of consideration over the three years to 2029, expected to be capitalised, with approximately half expected in 2027. Separately, it expects added central capabilities in trading, pricing and risk management to increase annual central costs by a low-tens-of-millions-of-euros amount from 2027. Savings and wider financial benefits are expectations, not delivered results.3
Bringing code in-house changes the cost structure; it does not guarantee a saving. A supplier fee can give way to a substantial continuing commitment to developers, DevOps, quality assurance, security and support. Much of that cost must be funded regardless of short-term demand. Data, trading and other external services can remain additional expenses.
The public figures do not establish whether Allwyn’s model will cost more or less overall. The board’s test should be total cost of ownership: whether the value of faster, better-controlled product changes and any costs actually eliminated justify the internal organisation and external contracts that remain.
For another operator evaluating a similar proposal, a useful cost model would separate initial rights and implementation, the steady-state internal team, continuing external services and the cost of an eventual exit. It should also identify which existing expenses will actually stop, when they stop and who is accountable for removing them.
The business case becomes more credible when a claimed saving is attached to a contract that can end or a duplicated activity that can cease. Moving a cost between a business unit and headquarters does not itself remove it. Equally, a higher central technology budget can be justified if it delivers changes the business values. The test is the resulting capability and economics, not where the expense is recorded.
Map responsibility before comparing suppliers
The following questions can turn an abstract promise of control into a procurement discussion. They apply to prospective arrangements generally; the public announcements do not supply Allwyn’s detailed answers.
| Layer | Responsibility to define | Evidence to request |
|---|---|---|
| Code and updates | Rights to modify and maintain the software; ownership of new work; responsibility for patches. | A permitted change built, tested and released by the receiving team. |
| People and support | Staffing, knowledge transfer, incident cover and support after handover. | A documented handover and a recovery exercise without reliance on one specialist. |
| Data and feeds | Rights to use, retain and export data; responsibility when an input fails. | A sample export and a tested replacement feed with the necessary permissions. |
| Trading and risk | Who sets policy, supplies decisions, can override them and keeps the audit trail. | A traced decision and an agreed escalation route, including service interruption. |
| Exit | Notice, assistance, data return, transition charges and rights that survive termination. | A replacement plan with named owners, dependencies and a cost estimate. |
An operator should be able to assign each obligation to an internal owner or a supplier. A gap between those assignments is more revealing than a broad statement that the platform is now “in-house”.
More choice can still sit behind one connection
Sportradar’s announcement on 12 November illustrates a different purchasing model. DATA.BET, Esportsbull and GRID are being added to its Unified Odds Feed, allowing operators to choose an esports odds provider by title through one integration, subject to availability and commercial terms.4
That offers a way to change a content choice within an external service. It is not equivalent to bringing the underlying capability inside the operator. The relevant buying questions concern access, pricing, continuity and whether changing provider is practical under the contract, not simply whether several names appear on a supplier menu.
Neither model is inherently superior. A single integration may be appropriate when the operator values convenience and breadth. Internal development may be appropriate where product changes are strategically important and the business can support them. The point is to compare the freedom obtained at the layer that matters.
Test control through an ordinary change
Before approving a source-code deal, a board could ask management to work through one ordinary future request: add a permitted data feed, alter a risk rule or release a market-specific feature. Who authorises it? Who performs the work? Which supplier must cooperate? What does it cost, and what happens if that supplier declines?
The same exercise should be repeated for a service failure and an exit. This is how a procurement team can distinguish a valuable dependency, deliberately retained, from a dependency the business thought it had removed.
Allwyn’s eventual results will depend on execution. For other operators, the immediate lesson is to buy a clearly defined ability to act—and budget for the people and agreements needed to keep that ability usable.
Sources and research notes
- Kambi, “Kambi to power Allwyn’s sportsbook technology consolidation in source code licensing agreement”, 10 November 2026. Licensing, team transfer, implementation period and separate trading-services agreement.
- Gambling Commission, “The current National Lottery licence”, last updated 20 July 2026; accessed 16 November 2026, “Allwyn” section. Identifies the UK National Lottery licensee; not evidence about the Kambi agreement.
- Allwyn, “Allwyn advances strategy to own and manage key technology and content capabilities”, 10 November 2026, p. 1, “A unified sportsbook platform” and “Additional transaction details”. Stated strategic rationale and company cost estimates.
- Sportradar, “Sportradar Expands Unified Odds Feed with Leading Esports Providers”, 12 November 2026. Announced provider choice through one integration; availability and commercial conditions apply.
Research checked on 16 November 2026. This analysis uses public company announcements and the Gambling Commission’s licence information; the full contracts have not been examined. The responsibility table and procurement tests are iGaming Review’s analysis. No independent performance assessment or total-cost estimate is implied.