The cheapest way to build a regulator is to buy one. Burundi did.

On June 30 the Loterie Nationale du Burundi stops operating as a commercial lottery and becomes the country's sole gambling regulator. Lotteries, sports betting, horse racing, online. Casino oversight in a later phase. Eight existing licensed providers move from competing with LONA to being monitored by it. The same staff. The same building. The new mandate.

What does the new regulator run on. NSoft, the Sarajevo-based gaming technology vendor, presented the results of its supervision framework to LONA on February 27. The Gambling Management System monitors player registration, operator turnover, jackpot payouts, declared revenues, tax liabilities. The whole supervisory stack of a country. Centralised. Single supplier.

This is the part most observers will not notice. Burundi is doing the thing that on paper is correct. Independent regulator. Centralised monitoring. Compliance enforcement. Tax transparency. The architecture reads like a textbook. The architecture is also operationally identical to "NSoft is up today, regulator is up today." Outage in Sarajevo, outage in Bujumbura. Pricing renegotiation in Sarajevo, regulatory pause in Bujumbura. NSoft sells to a competitor, the eight operators in Burundi inherit a counterparty they did not choose.

This is not a Burundi problem. It is the structural condition of every brand-new African regulator that buys its supervisory stack from a single vendor instead of building it. Kenya GRA is procuring monitoring. Liberia LNLA just signed Seven Blue. Gabon GDJ will need a platform by July 8. The vendor list is short. The contracts are long.

The operators who care about platform selection have started asking the harder version of the question. Not which platform do we run. Which platform does our regulator run, and what happens to us when it fails.

The market that solves this in five years is the market that builds in two-supplier mandates the way regulators built two-bank mandates a decade ago. The market that does not solve it discovers that "regulator independence" was always conditional on a service-level agreement nobody at the regulator could read.