Colombia is about to promise ten private mega-prisons. It has adjudicated exactly one, and only one company bid on it.

Colombia is about to promise ten private mega-prisons. It has awarded exactly one prison concession in its history.

One company submitted a bid. It scored 970 out of 1,000 because there was nobody to score it against. That company was a Mexican construction firm called Prodemex, operating through a local vehicle named Carmet Medellín S.A.S. The contract was signed in October 2023. Abelardo de la Espriella takes office August 7 with a plan to do this ten more times.

The entire evidentiary basis for believing that is possible is this one contract, in one city, for a facility one tenth the size of what he is describing.

1 bidder on Colombia’s only prison public-private partnership (PPP, known locally as APP, Asociación Público-Privada) · COP 675bn contract value, single offer · 1,339 beds, pretrial detainees only · 0 entities named to structure the ten

The Medellin facility is a municipal remand jail for 1,339 sindicados, people awaiting trial, five pabellones in the corregimiento of San Cristobal. It was procured by the city’s own APP agency, not by any national body. Construction is actually happening, roughly 9% complete as of February, delivery penciled for 2027.

The financing closed in early February 2026. Grupo Aval announced it on the 9th: COP 215,000 million, about USD 58 million, provided by Banco de Occidente and Banco de Bogota, structured by Aval Banca de Inversion. The concessionaire carries the rest of the COP 675 billion project cost and draws roughly COP 56 billion a year across a twelve year operating phase. It is an ordinary, competently structured piece of project finance. It is also the only one of its kind in the country.

What makes it interesting is the shape of the competition. The interventoria contract, the oversight mandate to supervise construction, drew three bidders when it was awarded in March 2024: Consorcio Intercarceles, Consorcio Metropol, and Consorcio MZCJ 2024.

Three consortia competed to supervise the prison. One bid to build it.

Colombian officials describe Intercarceles as two national firms plus a Chilean one with prison concession experience. Which Chilean firm is not disclosed in any government record or press account, and the consortium’s full membership sits in a SECOP filing that is not publicly retrievable. That is a small detail, but it captures something about this entire program: the supervision layer is legible and contested, and the layer where the money actually goes is neither.

Prodemex is not a household name outside Mexico. It should be for anyone trying to price this thesis.

The company is led by Olegario Vázquez Aldir, whose father Olegario Vázquez Raña, the founder of Grupo Empresarial Angeles, died in March 2025. Prodemex is family linked to that conglomerate but, by Aldir’s own account, not formally part of it.

Through its Proinfra arm, the company holds the concessions on two Mexican federal penitentiaries: Cefereso 14 in Gomez Palacio, Durango, and Cefereso 17 in Buenavista Tomatlan, Michoacan. Each holds 2,520 inmates. Each carries a 22-year term. The structure is the one now arriving in Colombia: private capital builds, the state pays per available bed for decades.

The part that matters: Zero competitive tenders were held for the Mexican prisons Prodemex operates. Both Ceferesos were awarded by adjudicacion directa, direct award, no bidding, a fact confirmed in the Cefereso 17 contract text itself.

Mexico’s federal government later put the total projected cost of the eight privatised prisons at roughly 266.3 billion Mexican pesos across their twenty year terms. The Lopez Obrador administration attacked the arrangements as contratos leoninos, one sided contracts written against the state. That criticism was aimed at the scheme and at all eight contracts, not at Prodemex by name. The distinction matters. No Mexican court has found against the company.

But the pattern still travels. A firm whose domestic prison portfolio was handed to it without a tender has now won Colombia’s first prison concession in a tender nobody else entered. Neither event is an allegation of wrongdoing. Together they are a description of how this market clears.

The most useful thing in the record is not the Medellin contract. It is the program that came before it and never happened.

Between 2015 and 2018, under a national push involving the DNP, the justice ministry and USPEC, Colombia floated a set of prison APP projects. The named candidates were Popayan in Cauca and Barrancabermeja in Santander, with Uramita in Antioquia cited as a third.

Not one was structured. Not one was adjudicated. Not one reached financial close. None disclosed a private originator, because none attracted one. Barrancabermeja was eventually built the old way, as a conventional USPEC public works prison at around COP 214 billion, explicitly outside the APP framework. Popayan left no adjudication trace at all.

Colombia has announced a national prison concession program before. The projects were named. None were built.

The lesson is not that private prisons cannot be financed in Colombia. Medellin proves they can. The lesson is that the binding constraint has never been political will or legal framework, both of which existed in 2016. The binding constraint is that when Colombia opens a prison tender, almost nobody bids.

De la Espriella takes office in under two weeks. Against that deadline, the ten prison program currently consists of the following procurement artefacts: none.

No structurer: No entity has been designated to structure the program. Not the DNP, not the ANI, not USPEC, not the Agencia Virgilio Barco, not the FDN, not an investment bank, not a new agency. The question of who would hold the contracts is genuinely open.

No paperwork: No CONPES document, no enabling law or decree, no vigencias futuras authorisation, no SECOP process, no RUAPP registration, and no identified sites.

No bidders: No firm, Mexican, Spanish, Chilean, Brazilian or American, has publicly expressed interest. No unsolicited private initiative for prison infrastructure has been filed.

No costing: No affordability analysis from the Comite Autonomo de la Regla Fiscal, ANIF or Fedesarrollo. The outgoing justice ministry has called budget financing non viable, which is precisely why the concession route is being proposed.

One name does exist. Ivan Cancino is both the justice sector transition coordinator and the minister of justice designate. He has confirmed the concession route, promising ninguna carga fiscal, with a three to four year build. He has not said who would run the process.

And here the program argues with itself. De la Espriella campaigned on eliminating INPEC and folding the justice ministry into Interior. Those are the institutions that would legally hold and administer the concessions. The plan proposes to demolish the counterparty before signing the deal. Cancino has already begun retreating from the ministry fusion, telling W Radio in July that he would not recommend it.

The honest read: None of this means the program is theatre. Colombia’s prison crisis is real and non partisan: roughly 102,949 people in facilities built for 81,025, about 27.1% over capacity, under a Constitutional Court estado de cosas inconstitucional that remains in force. A state that cannot house the people it arrests has no security policy. The question is not whether Colombia needs this. It is who shows up to build it.

Everything above this line is documented. What follows is not, and we are labelling it plainly, because the most interesting question about this program cannot be answered from the record: nothing is concrete yet, and the reveal almost certainly comes after August 7 rather than before it. A president elect with no majority does not announce his contractors while he still needs the votes to abolish the agency that would sign them.

There are two ways this goes, and they lead to completely different balance sheets.

The continuity case is that Colombia does what Colombia does. The Medellin template gets scaled because it is the only template that exists. Prodemex bids, Grupo Aval finances, Colombian civil works firms take the local content, and the program moves at the speed of Colombian infrastructure, which is to say slowly, through courts and cost overruns and a delivery date somewhere past the end of the term. This is the base case for the simple reason that it requires nobody to do anything new.

The break case is the one worth thinking about. De la Espriella campaigned on Bukele’s results, holds American citizenship, lived in Miami for over a decade, was congratulated by Trump within hours of the count, and wants Colombia inside the hemispheric security architecture Washington is assembling. It is entirely reasonable to expect that a man in that position looks at the regional contractors who have delivered slowly and expensively for thirty years, and decides to go somewhere else entirely.

What the Bukele route actually was: If Colombia copies El Salvador, it does not get American contractors and it does not get a concession. CECOT was direct-awarded by the public works ministry to two Salvadoran firms and one Mexican one, paid for out of the state budget at roughly USD 115 million, with the construction records sealed for two years. No US, Turkish or Chinese company appears in the file. The core build took about seven months. That is the model: skip the concession, spend budget money, award without tender, classify the paperwork.

That detail cuts both ways, and it is why the speculation is worth stating rather than assuming. It argues against American builders, since the playbook being admired never used any. It argues for the broader instinct, since the Salvadoran route bypasses the concession model entirely, which means it bypasses Prodemex, Grupo Aval, and every listed company an investor could actually buy. Ten prisons at Salvadoran cost is on the order of USD 1.2 billion of budget money. Ten prisons at Medellin’s cost per bed, financed, is a multiple of that and arrives years later.

American participation, if it comes, is likelier to arrive as security technology and training through cooperation channels than as concrete. That is how Washington has historically shown up in Colombia, and it is the part of the procurement that moves fastest. You cannot build a prison in a hundred days. You can sign a perimeter surveillance contract in a hundred days.

Continuity: A new APP tender that looks like Medellin’s. Vigencias futuras requests to Congress. Prodemex or a Mexican peer named. Grupo Aval structuring. Slow, legible, and buyable.

Break: Prison construction moved into the national budget rather than a concession. Direct awards. An emergency or conmocion interior decree used to bypass ordinary procurement. Records reserved. Fast, opaque, and unbuyable.

The tell: Watch whether the first prison money appears as a vigencia futura or as a budget line. That single distinction tells you which government you are dealing with, and it will be visible well before any company is named.

We hold no view on which is likelier, and we would be inventing one if we claimed otherwise. The point of setting both out is that the program is currently a promise with no procurement attached, and a reader who assumes it will arrive through the existing, investable channel is making a bet they have not been told they are making.

There is no clean way to own this theme, and the honest version of the trade is mostly a warning about the unclean ones.

Grupo Aval (BVC: GRUPOAVAL) is the only listed company with a demonstrated position. It financed the one deal that exists, through two of its banks, with its own investment banking arm structuring it. If the program scales, it holds the template and the relationship. It is also a large Colombian bank, and prison lending is a rounding error on its book. You would be buying Colombian credit and financial sector recovery, with prison PPP as a small optional kicker. That is a reasonable thing to buy. It is not a prison trade.

Conconcreto (BVC: CONCONCRET) and El Condor (BVC: ELCONDOR) are the listed Colombian civil works names. Neither has built a prison. Both build the kind of heavy infrastructure prisons sit on, and both benefit from any construction cycle, with de la Espriella’s broader infrastructure agenda offering catalysts well beyond corrections. A third name, Conciviles, appears on screeners at around 171 pesos and shows almost no trading activity. Treat it as unbuyable rather than cheap.

CoreCivic and GEO Group are not Colombian plays. Neither has any presence in Colombian or Latin American prison construction. They operate facilities under government contract. They do not develop infrastructure under concession. Their 2026 performance is a function of United States detention policy and nothing else. If Colombia ever reaches them, it will be as minority operating partners inside somebody else’s consortium, years from now. Buying them on this thesis is buying an American immigration trade and calling it Colombian.

The most direct exposure is the one nobody can buy. Prodemex is private, and so is the concession vehicle.

The structure works. That is the finding, and it is genuinely bullish for Colombia. A city government wrote a fifteen year availability payment concession, a foreign operator took it, a domestic bank syndicated it without state guarantees, and there is a building going up in San Cristobal today. Colombia’s PPP framework is not theoretical.

What has not been demonstrated is competition. The one prison concession Colombia has ever awarded went to the only company that wanted it, and that company’s domestic prison portfolio was assembled without tenders at all. Ten times that scale, run through institutions the incoming government intends to abolish, with no structurer named and no second bidder identified, is not a pipeline. It is an intention.

Watch three things after August 7. Whether a structuring entity is named. Whether the first vigencias futuras request reaches Congress. And whether a second firm, any second firm, publicly signals interest.

The trade is not the announcement. Colombia has had the announcement before. The trade is the moment somebody bids against Prodemex.

• ColombiaOne: “De la Espriella’s New Prison Model for Colombia” (July 23, 2026)
• Expansion.mx: “Prodemex prepara el debut de una Fibra E para cárceles” (Nov 27, 2017)
• Caracol Radio: “Abelardo de la Espriella oficializó a Cancino como MinJusticia” (July 7, 2026)
• Cambio Colombia: “El Ministerio de Justicia podría desaparecer” (June 24, 2026)
• Infobae: “De la Espriella estaría estudiando fusionar los ministerios” (June 24, 2026)
• KienyKe: “Iván Cancino descarta eliminar la JEP” (July 2026)
• Reuters: “Colombia presidential candidate de la Espriella proposes mega-prisons” (2026)
• Insight Crime: “Report Slams Colombia Failure to Control Prison Overcrowding” (2026)
• Wikipedia: “Terrorism Confinement Center” (CECOT)
• Grupo Aval: financing announcement (Feb 9, 2026)
• PRODEMEX: Cefereso 14 and Cefereso Michoacán project pages
• World Prison Brief: Colombia country data