Brazil’s REDATA Turns Tax Relief Into a Compute-Resource Bargain

Brazil has stopped treating a data center as just another building full of imported equipment. Under REDATA, the country’s new special tax regime, a qualifying project is expected to deliver something back: computing capacity for the domestic market, electricity with a low-emissions profile, tightly controlled water use and investment in Brazilian research.

That bargain is the real policy. The tax suspension is the instrument.

Law No. 15,504, sanctioned on 15 September, allows approved projects to suspend federal taxes on equipment and inputs used to install, modernize or expand data centers. The benefit can last for five years. But it becomes a zero-rate benefit only when the operator demonstrates that it has met the program’s conditions. Failure can trigger repayment with interest and penalties.

This structure matters because it changes the sequence. The state is not simply awarding relief in the hope that wider benefits will appear later. It is defining performance tests and retaining a route to recover the concession if they do not.

The first test is domestic usefulness. Participants must generally make at least 10 per cent of their data-processing, storage and handling capacity available to Brazil’s domestic market. An operator can replace that obligation with an additional research-and-development commitment, but the trade is explicit: capacity not reserved for the home market must be compensated by a larger contribution to the national technology base.

The second test is electricity. A participating project must cover all of its contracted demand through renewable or otherwise low-emissions electricity, either through supply contracts or self-generation. That does not remove pressure from the power system. It does, however, make the operator’s procurement strategy part of its eligibility rather than a voluntary sustainability claim.

The third test is water. REDATA sets an annual water-use-efficiency ceiling of 0.05 litres per kilowatt-hour. This is unusually concrete for an industrial incentive. A national number cannot by itself account for every catchment, cooling design or local scarcity condition, but it gives regulators and operators a measurable floor from which project-level scrutiny can begin.

The fourth test is capability. Operators must invest 2 per cent of the value of supported goods and services in research, development and innovation in Brazil. Projects in the North, Northeast and Center-West receive lower thresholds for some commitments, reflecting a second ambition: to spread investment beyond the country’s established digital centers.

Together, these conditions make REDATA a compact among five systems. Fiscal policy lowers the cost of investment. Energy policy shapes how new demand enters the grid. Water policy sets an operating constraint. Digital policy claims a share of capacity for domestic users. Innovation policy tries to keep more knowledge and technical work inside Brazil.

The arrangement is promising precisely because it is testable. It will also be difficult to administer. Ministries will need consistent definitions for clean supply, annual water performance, usable domestic capacity and eligible research. Operators will need to report in forms that allow compliance to be checked across years, not merely at entry. Local authorities will still have to judge whether a nominally compliant project is appropriate for its power network and water basin.

The strongest feature of REDATA is therefore also its main implementation risk. By linking several public objectives to one incentive, Brazil has created leverage. It has also created a monitoring problem. If the evidence remains fragmented across tax, energy, water and technology agencies, the conditions could become a checklist rather than a real operating discipline.

The next documents matter as much as the law: participation rules, evidence standards, audit procedures and public reporting. They will show whether REDATA is a durable exchange of public benefit for public support or simply a cheaper route to equipment imports.

Brazil’s wager is clear. Data center growth is welcome, but only on terms that strengthen the country’s own digital economy and respect the physical systems on which computation depends. The measure of success will not be announced investment. It will be verified domestic capacity, energy provenance, water performance and research that would not otherwise have happened.