Costa Rica
Costa Rica iGaming market in numbers
| Metric | 2025 | 2026 |
|---|---|---|
| Total GGR | $330m | $360m |
| Regulated GGR | $50m | - |
| Offshore GGR | $280m | - |
| Channelization | 15% | - |
| Mobile share | 70% | - |
| YoY growth | - | +9.0% |
| CAGR 2021–2026 | +8% | - |
Regulated and offshore split
Legal status by vertical
Operator's read on Costa Rica
Costa Rica is the licensing hub that is not actually a licensing regime, and an operator has to understand that distinction before relying on it. There is no gambling-specific regulator and no iGaming licence in Costa Rica. Operators incorporate a local company and obtain a generic business or data-processing permit, and the resulting "Costa Rica licence" is a marketing label rather than a regulatory authorisation. It permits running operations from Costa Rica provided the operator does not target Costa Rican residents or use local banks. For how the offshore options compare, the Curaçao versus Costa Rica and Anjouan versus Costa Rica comparisons are the relevant reads.
There is no real licence, and that is the point. Because Costa Rica has no gambling regulator, there is no fitness vetting, no gambling-specific compliance regime and no authorisation that a payment partner or another regulator would recognise as a gambling licence. Operators choose it precisely because it is cheap, fast and flexible, with none of the scrutiny of a real regulator. That absence of oversight is the entire proposition, and it is also the entire risk.
The reform that would change this has stalled. A bill to modernise the framework and actually license digital operators, drafted with input from the state lottery body, was approved by a legislative commission in late 2025 but then rejected by that same commission in an internal vote in January 2026. So as of mid-2026 the genuine-regulation path is stalled, and Costa Rica remains the unregulated hub it has long been. An operator should not plan around an imminent real licensing regime, because the most recent attempt failed.
The reputational and banking risk is rising. The flip side of zero oversight is that banking and payment partners increasingly discount a Costa Rica setup, and the AML and reputational risk attached to it is growing as other jurisdictions tighten. For an operator, that means a Costa Rica base may be cheap to establish but increasingly costly in terms of the payment relationships and market acceptance it fails to unlock. It is a low-cost shell, not a trust signal.
What winning looks like. Winning with Costa Rica, to the extent it makes sense at all, looks like using it only as a low-cost operating shell for markets that accept it, with clear eyes about its lack of regulatory credibility and the banking friction that comes with it. An operator that needs reputation, banking access or a recognised licence should look to Curaçao under its new regime, Panama, or a tier-one European hub instead.
The regional play. Costa Rica sits at the bottom of the offshore hub credibility ladder, below Curaçao and Antigua, and well below Panama or the European bases. Which hub fits depends on the operator's reputation and banking needs, as discussed in the multi-market sequencing piece.
The biggest mistake. The biggest mistake is treating the Costa Rica setup as a genuine gambling licence that confers credibility, when it is a generic business permit with no gambling regulator behind it. The related mistake is choosing it for low cost without pricing in the banking and acceptance friction that absence of regulation now creates. Use Costa Rica only as a cheap shell where it is accepted, and look elsewhere when reputation or banking matters.
What's changing
Itself a licensing hub; liberalisation trend.
Where these figures come from
- Slotegrator 2025
- Statista
GGR figures are 2025 estimates or actuals where regulator data is available; 2026 projections drawn from the most recent published forecasts. Offshore figures are inherently more uncertain than regulated figures and should be treated as directional. Where reputable sources disagree materially the dataset uses the midpoint of the range.