Summary

  • Anatel’s proposed 6 GHz auction would direct part of its economic value toward neutral regional hubs and smaller edge nodes in Brazil’s North, Northeast and Centre-West.
  • The draft would require a market test and a proven viability gap before funding, then condition the final 30% on twelve consecutive months of measured operation. This is a proposal, not an approved auction or a built network.

The most consequential part of Brazil’s proposed 6 GHz auction may be the money that does not get spent. Anatel’s draft would attach a regional digital infrastructure commitment to the auction, but it does not promise a fixed number of data centres or require the entire available ceiling to be used. Before a project receives support, the design asks whether the market would build an equivalent facility without it. After construction, it asks whether the facility actually delivers the service it promised.

That distinction matters because a subsidised building is not the same thing as a new network capability. A public contribution can displace investment that a private operator had already planned. A completed shell can still lack power, diverse routes, customers or useful interconnection. The draft tries to separate these outcomes with several gates, although none has yet been tested in operation.

The consultation covers authorization to use 6,425–7,125 MHz, the upper portion of the 6 GHz band. Anatel proposes 700 MHz in three 200 MHz blocks and one 100 MHz block, with national and regional lots; its published planning horizon contemplates an auction by 2028. The draft also contains separate mobile-coverage and transport obligations. The regional infrastructure commitment is a different obligation, funded from a portion of the auction’s economic value. It would support neutral hubs and smaller edge nodes in service areas covering the North, Northeast and Centre-West. The agency says these sites are intended to bring interconnection, hosting and cache capacity closer to places now farther from those services. That is the policy rationale, not evidence that latency or service quality has already improved. (Anatel’s proposal summary; full Consultation 37 text)

The first gate is additionality. The proposed administrator would map existing, under-construction and announced neutral colocation and interconnection capacity. A locality could be excluded if an equivalent neutral offer already operates nearby, if the power supply or physically separate fibre routes are inadequate, or if environmental and site conditions make the project unworkable. For places that remain eligible, a market test lasting at least 60 days would let a private provider declare that it can deliver equivalent capacity within 24 months.

A declaration would need evidence of readiness, such as land, a formal power request, approved budget or equipment orders; a general statement of interest would not suffice. Others could challenge the declaration, and a credible project would pause local eligibility while its delivery is tracked.

If no private project displaces the proposed site, the next question is how much support it needs. The draft caps the contribution at the project’s calculated viability gap: eligible costs minus projected revenue, assessed over a reference horizon. It says the contribution should not cover the full project where commercial revenue can repay part of the cost. A competitive call would select the operator, normally using the lowest requested contribution among technically and financially qualified bids.

If the eligible portfolio needs less than the fund’s ceiling, the remainder would not have to be spent on data centres; the draft redirects unused money to separate transport and coverage commitments.

Payment is then staged. Up to 40% could be released for verified construction spending, subject to a retention. Another 30% follows energisation and physical interconnection. The final 30%, together with retained amounts, depends on twelve consecutive months of operation. The proposed evidence includes availability, round-trip latency at the 95th percentile, connected autonomous systems, occupancy, energy and water efficiency, and compliance with a public reference offer. Independent measurement and access to primary data are part of the draft. Failure on essential availability, interconnection or open access would block release rather than be averaged away by construction progress. (Consultation 37, Annex IV-C)

That is a more demanding test than counting buildings or announcing investment. Yet a carefully written test is not proof of a workable institution. The administrator is meant to manage funds rather than own or operate the assets; private operators would be selected competitively and required to offer open, equal and non-discriminatory access. The proposal also includes occupancy review, repayment of excess returns and public reporting. Those controls could limit capture by a subsidised operator, but they depend on a reliable baseline, comparable measurement, competent auditors and a regulator able to enforce the contract over time.

The consultation was still open on 14 October, with contributions due by 23 October 2026. Public hearings in Brasília and São Paulo were scheduled for 16 and 21 October. The final auction terms and the infrastructure annex may change before approval. Until then, there are no awarded sites, selected operators, committed amounts or performance results to report. (Current consultation timetable; Anatel hearing notice)

The proposal’s central wager is that funding can be made conditional on two forms of proof: that private capital would not have supplied the same capacity on time, and that the funded asset later operates as an open network resource. The auction will be judged not by the size of its ceiling or the number of projects announced, but by whether those tests remain credible when money, deadlines and regional priorities collide.