Summary

  • Juan Carlos Duarte Dure's public record during Paraguay's 2024-2025 5G tender connects consultation-stage design, two 200 MHz allocations, about USD 4 million in licence payments, about USD 114 million in stated infrastructure commitments, and formal engineering and service-start deadlines.
  • The award record establishes who received spectrum and what was stated or required; it does not establish completed spending, coverage, speed, service quality, legality, wrongdoing, or nationwide operational continuity.

The allocation ledger is the beginning of the test

Radio spectrum is an input to a mobile network, not the network itself. A regulator can define a band, accept applications, determine eligibility, allocate frequencies, record payments, impose obligations, and set milestones. Each step creates evidence about a legal or administrative state. None of those steps, standing alone, shows that antennas have been installed, transport capacity is available, software is stable, coverage reaches intended areas, or users receive a reliable service. The distinction is easy to state and frequently lost when a tender is announced as if it were a completed infrastructure program.

The accepted public record identifies Juan Carlos Duarte Dure as president of Paraguay's national telecommunications regulator, CONATEL, during the relevant 2024-2025 tender chronology. That description is deliberately source-dated. It supports an account of his public statements and regulatory role during the process; it is not a claim about what office he holds at the time of publication. It also does not convert CONATEL's institutional work into Duarte's personal technical authorship. Operators, engineers, lawyers, evaluators, and other public officials remain part of the process even when a named official explains a decision.

Within those boundaries, Duarte's record is useful because it carries a decision through several stages. Trade reporting from December 2024 attributed to him a plan to use consultation input to balance infrastructure investment, social commitments, and direct revenue. Later reporting documented the allocation result, monetary categories, and implementation deadlines. Still later reporting recorded questions about bidder eligibility information and a public investigation. Read together, the articles show a regulatory choice, the constraints applied to it, the recorded award, and the limits on what the award can prove.

That chain is more informative than either praise or accusation. A favorable narrative might treat the stated investment figure as delivered infrastructure. A hostile narrative might treat low direct revenue, incomplete knowledge, or an investigation as proof of misconduct. The evidence supports neither shortcut. It supports a harder question: what did the regulator choose to record and require, and what later observations would be needed to determine whether those requirements became working service?

A consultation-stage choice among unlike values

Convergencia Latina's December 2024 account described a draft tender and attributed to Duarte an intention to incorporate consultation input. The stated design problem involved at least three categories: money paid directly for spectrum rights, investment expected in network infrastructure, and social-connectivity commitments. Those categories can all carry public value, but they are not interchangeable. A dollar received by the state is an observed payment. A dollar committed to future infrastructure is an obligation whose completion must be verified.

A social-connectivity promise needs a defined beneficiary, delivery condition, time frame, and measurement method.

Treating the categories separately does not decide which one should dominate. Direct revenue can be counted immediately and may support public finances, but a high acquisition cost can reduce capital available for deployment or discourage participation. Infrastructure obligations can direct resources toward towers, radio equipment, transport, power, and operations, but their value depends on clear definitions and enforceable milestones. Social obligations may address areas or populations that a purely commercial rollout would not prioritize, but they can become symbolic if eligibility, service levels, and continuity are not measurable.

The regulator's task is therefore not simply to maximize one number. It is to make the trade-off legible. If direct payment is lower because investment obligations are higher, the documents should distinguish the two. If an obligation substitutes for part of a licence price, the substitution should be testable over time. If a bidder receives credit for a social commitment, the record should state what performance closes the obligation. Otherwise, an appealing total can combine cash already paid with expenditure not yet made and service not yet delivered.

The five-year licence term reported at the design stage adds another constraint. A time-limited right changes the investment horizon for an operator and the monitoring horizon for the regulator. Deployment has to occur early enough for the network to operate meaningfully within the term. Renewal expectations, depreciation, equipment cycles, and uncertainty can affect investment choices, even though the accepted evidence does not disclose the operators' internal financial models. For oversight, the term means that milestones cannot be allowed to drift indefinitely without reducing the practical value of the licence period.

Concentration risk was also part of the reported design discussion. Spectrum in a desirable band is finite, and the way blocks are structured can shape participation. Yet avoiding concentration cannot be measured only by counting awardees. A nominally separate recipient still needs financing, technical capability, lawful eligibility, access to supporting infrastructure, and a credible route to service. Conversely, an incumbent's existing assets may allow faster deployment but can reinforce market power. The accepted articles identify the concern; they do not provide enough information to settle the competitive outcome.

Staged base-station obligations offer one way to connect an award to observable progress. A staged requirement can break a large promise into dated checkpoints and make delay visible earlier. Its usefulness depends on what counts as a compliant station, where obligations apply, whether shared infrastructure qualifies, and how operation is verified. The frozen record confirms that staged obligations were part of the design discussion, but it does not provide a complete compliance schedule. Any later assessment should use the final enforceable documents rather than fill gaps from the draft-stage report.

Infrastructure sharing created another reported constraint. Sharing can lower duplication and speed entry when a new operator needs access to towers, sites, ducts, power, or other facilities. It can also produce disputes over price, access conditions, timing, and control. The December 2024 account described unresolved pricing disagreement around sharing. That fact matters because an award can be formally complete while a practical dependency remains unsettled. The evidence does not establish which party was right or what final terms emerged.

Two awards, two monetary categories, several clocks

By October 2025, accepted reporting described the tender as concluded with 200 MHz allocated to Claro and 200 MHz allocated to Nubicom. The combined allocation was therefore 400 MHz. This is the clearest administrative result in the record: two recipients and equal quantities of spectrum. It says nothing by itself about how much of the spectrum was in use, where service was available, or how traffic performed.

ABC Color reported a combined commitment of approximately USD 118 million and attributed the infrastructure-over-revenue rationale to Duarte. The frozen evidence separates that total into about USD 4 million in licence payments and about USD 114 million in stated infrastructure commitments. The distinction must remain visible whenever the figure is repeated. Calling the full USD 118 million auction revenue would be inaccurate. Calling the infrastructure portion completed spending would also be inaccurate.

A licence payment is an event that can normally be evidenced through a receipt, accounting entry, or formal confirmation. An infrastructure commitment is a future-facing obligation. Verification may require procurement records, installation evidence, site acceptance, operational tests, coverage measurements, service availability, and continued performance. The public articles summarized here do not supply that complete chain. They establish the stated categories and amounts, not the final execution of every commitment.

The award also introduced clocks. Accepted reporting described a 90-day period for detailed engineering and a six-month deadline to begin service. The two milestones answer different questions. Detailed engineering should make a network plan concrete enough to evaluate against technical and regulatory requirements. Beginning service should produce an operational state. Neither milestone should be treated as a substitute for the other, and neither necessarily proves broad or mature service beyond the defined acceptance criteria.

A 90-day engineering deadline can be meaningful only if the expected deliverable is clear. A plan could include radio design, backhaul assumptions, site sequencing, interference analysis, dependencies, security considerations, testing, and rollout phases. The accepted reporting does not disclose the contents or acceptance criteria, so an outside observer should not claim that a particular design was filed or approved. The defensible point is narrower: the process reportedly imposed a near-term engineering obligation after allocation.

The six-month service-start milestone moves the inquiry from plans toward running infrastructure. Even here, "begin service" needs definition. It might refer to commercial availability, a minimum number of active sites, a defined geographic area, an accepted technical test, or another contractual condition. Without the governing document and later verification, the phrase cannot establish nationwide coverage or a particular user experience. It does establish that the award contemplated a transition from administrative rights to observable operation on a stated schedule.

These clocks make delay and ambiguity important. If a dependency such as infrastructure sharing remains unresolved, the regulator has to decide whether it affects compliance, whether an extension is justified, and how equal treatment is maintained. If engineering is filed but service is not demonstrable, the two milestones should remain separately recorded. If a limited launch satisfies a formal threshold, public communication should still distinguish that threshold from broad coverage. Good recordkeeping protects both the regulator and the public from category drift.

Eligibility is a gate, not a prediction of performance

Ultima Hora's October 2025 report added a specific precondition around Nubicom's legal formation and an external subscriber verification. The accepted capsule records that CONATEL used a formal-declaration acceptance rule and that verification remained a constraint. This is important because a tender needs an administrable way to decide whether an applicant has supplied required information. A formal declaration can support a decision process, but it does not make the underlying fact immune from later checking.

DPL News later reported that Duarte said he did not know Nubicom's exact subscriber count, asserted that it had more than 100,000 users and met the requirement, and said he would review documents. Those statements should not be converted into an independent finding about the subscriber total. Nor do they establish that the bidder was ineligible or that accepting its declaration was unlawful. They document a gap between the exact detail publicly known by the named official during an interview and the conclusion he attributed to the process.

Eligibility and performance are separate again. Even a fully verified subscriber count would answer a tender-entry question, not whether a 5G network could be financed, built, secured, and operated. Conversely, uncertainty about an exact count does not prove that later engineering or deployment must fail. The regulator needs both gates: a defensible admission record and continuing technical oversight. Combining them into a single judgment would hide where evidence is strong and where it remains incomplete.

The formation of a new company can also be an administrative milestone without being proof of operational capacity. Corporate existence, licences, spectrum rights, network assets, interconnection, staffing, and working service are distinct states. A reliable public record shows when each state is reached. That sequence matters particularly for a new entrant because external dependencies may be greater and historical operating evidence may be thinner. It also matters for an incumbent, whose existing network should not be assumed to prove performance in a newly allocated band.

The practical lesson is not that formal declarations are inherently weak or that every fact must be re-litigated publicly. It is that the decision record should preserve what was declared, what was independently checked, what remained pending, and who owned the next verification. This gives later reviewers a way to distinguish a procedural acceptance from a confirmed operational fact. It also reduces the temptation to reinterpret an early administrative step after controversy emerges.

Investigation is an oversight fact, not a verdict

ABC Color reported in December 2025 that the Nubicom award was under investigation by Paraguay's Comptroller General. The article also recorded Duarte saying that he did not know a company leader when questioned and that documents would be provided later. The frozen evidence contains no finding or outcome from that investigation. It therefore supports reporting that scrutiny existed, not an allegation that corruption, illegality, favoritism, concealment, or personal benefit occurred.

This boundary is not a courtesy that weakens accountability. It is what makes accountability auditable. An investigation creates questions, document requests, and a process. A finding would create a different evidentiary state. A charge, judgment, or sanction would be different again. Collapsing those stages harms both due process and public understanding. It can also distract from concrete, answerable questions about eligibility records, allocation conditions, milestones, and delivery evidence.

The same discipline applies to Duarte's reported knowledge. A statement that an official did not know an exact subscriber count or a company leader at a particular interview is not proof of what the official knew at another time. It does not establish motive, concealment, friendship, influence, or negligence. The record does support asking how decision-critical information was held institutionally, what was verified before award, what documents supported the decision, and how later review was handled.

Institutional recordkeeping is especially important when individual recollection is incomplete. A regulator should not depend on one person being able to recite every fact. It should be able to retrieve the declaration, verification result, evaluation record, licence condition, deadline, and compliance evidence. In that sense, the strongest response to a knowledge gap is not a stronger personality claim. It is a better evidence chain.

A spectrum right is a controlled record

Spectrum governance works when rights are unique, bounded, transferable only under defined rules, and connected to accurate operational conditions. The allocation record serves as a ledger: it identifies the holder, frequency range, term, obligations, and status. Its authority lies in the ability of other actors to rely on those limited facts and to see changes recorded. It should not be asked to certify facts outside its scope.

This is why an allocation cannot establish network legitimacy by ceremony alone. A public announcement can accurately report that rights were assigned. The running network remains the reality layer. Engineers and users encounter coverage, capacity, interference, transport, power, routing, security, maintenance, and support. Regulators encounter milestone evidence and exceptions. A complete accountability model connects the ledger to those observable states without pretending they are identical.

The principle also limits biographical storytelling. Duarte's relevance here is not that a regulator personally embodies the network. It is that his attributed statements expose the reasoning and constraints attached to a national resource-allocation decision. His record can be evaluated because the tender produced dates, quantities, categories, conditions, and later questions. The analysis should remain attached to those records rather than expand into an unsupported account of his career or intentions.

The allocation described in the accepted articles is significant as a resource record, but those articles do not provide a complete engineering picture. They do not establish site locations, interference coordination, radio configuration, transport paths, security controls, or coverage results. These omissions are not defects to be filled by assumption. They define the evidence needed for a later operational assessment.

What observable delivery would look like

The first observable layer after allocation is documentary. A compliant record should show the final licence terms, payment status, engineering submission, accepted milestones, and any approved changes. For infrastructure commitments, it should separate obligated value from contracted value, paid value, installed assets, accepted assets, and assets in service. These categories can move at different times. Reporting one as another creates false precision.

The second layer is physical and operational. Evidence may include activated sites, tested links, available spectrum use, service areas, and operational acceptance under the licence. The exact metrics must come from the governing conditions, not from an outside author's preferred definition. Public claims about coverage or service should be traceable to a dated measurement method. Marketing availability and regulatory acceptance may answer different questions.

The third layer is continuity. A network that begins service still needs maintenance, security, capacity management, fault response, and sustainable access to shared infrastructure. Five-year rights make continuity an ongoing obligation rather than a launch-day event. A single successful test does not prove resilient operation. Conversely, an early incident would not by itself prove the whole allocation failed. The record should accumulate measurements over time.

The fourth layer is social delivery. If obligations are intended to support particular communities or public goals, the beneficiary and service condition need to be defined. A tower count may be relevant but not sufficient. Availability, affordability, usable performance, and persistence may matter, depending on the obligation. The accepted source set does not specify the final measurement design, so this article cannot declare whether social commitments were met. It can identify why that definition is essential.

The fifth layer is financial reconciliation. The approximately USD 114 million figure is a stated infrastructure commitment, not a verified asset register. Later assessment should show which expenditures qualify, when they occurred, and how double counting is prevented. Shared infrastructure complicates this further: one asset may support more than one service or operator, while a payment for access may not equal ownership of an asset. Clear accounting rules prevent a large headline number from becoming impossible to test.

Transferable lessons for spectrum governance

The first lesson is to keep monetary categories separate from the beginning. Auction proceeds, annual fees, performance bonds, qualifying capital expenditure, shared-infrastructure payments, and social obligations should not be merged into one success figure. Each has a different verification method and timing. A total may be useful for orientation, but the components carry the accountability.

The second lesson is to define the conversion from commitment to result. A requirement should identify the evidence that closes it, the date by which that evidence is due, and the consequence of non-compliance. Staged obligations are valuable because they expose problems early, but only if the stage has an observable acceptance condition.

The third lesson is to separate admission from ongoing performance. Eligibility checks decide who may participate. Engineering review tests a plan. Service-start acceptance tests an initial operational state. Continued monitoring tests continuity. None should be assumed from another.

The fourth lesson is to preserve institutional evidence so that accountability does not depend on memory. A regulator should be able to show the documents and checks behind a decision even when a public official cannot recall an exact number during an interview. This protects fair review and makes later correction possible.

The fifth lesson is to treat scrutiny as a process with states. Questions, investigations, findings, and remedies are different events. Reporting them accurately keeps attention on evidence and prevents an unresolved inquiry from becoming either proof of guilt or a reason to ignore legitimate concerns.

The sixth lesson is that operational service is the ultimate test of an infrastructure-centered tender. Allocation, promises, and deadlines matter because they structure that test. Their public value is realized only when the resulting network can be observed, measured, maintained, and held to its conditions over time.

Sources