High-altitude energy: maintaining the solar and the batteries already installed in the Puna
One of the four lithium plants already producing in Salta declared that its ramp-up, in the first quarter of 2026, was limited by a shortage of gas supply verif verbatim in the company half-year report. That is exactly what sells here: firm energy, which is what holds production back today. The market has three distinct clocks: maintaining the megawatts already spinning —that is paid today and never switches off—, building the generation under construction through 2029, and converting diesel engines to gas while the power line has not arrived. And what pinches has to be said: six out of every ten pesos in this market, and up to seven and a half, never reach tender in Salta —the owner of most of the capacity solved it with its own chain, and whoever built the hybrid plant also kept its maintenance—. One warning about the figure above: it counts the maintenance of generation (0.13 dollars per tonne milled), not the cost of generating (8.07). The feasibility study publishes the two lines together and they look alike; that is 62 times the difference, and confusing them would make a single project not yet built look larger than the market of the whole province.
What the market is made of
The number is contractable service, not energy spend: it does not include fuel or imported equipment, which are the two big lines and the two not bought here. It is built in three blocks with three different clocks — one already running that does not stop, one that switches off on a date, and one that closes by itself when the transmission line arrives.
The rule that moves it
No rule creates this niche's demand — the megawatts already spinning create it — but two define how much is left after tax and who can bill it, and in this line of business that weighs more than in any other. Each one opens its own page, with the rule, since when it applies and its primary source.
enablesSalta cuts the rate 20% for retail and hospitality, and exempts newly registered taxpayers for 12 monthsIt is the one that finances year one and it has a date: a new taxpayer registering voluntarily pays a zero rate for up to twelve months, and the benefit expires at the end of 2026 unless extended. But here it matters above all for what happens after those twelve months, and it is the niche's most counter-intuitive fact: the activity can be classified under different lines with twice the difference between repairing and building, and the expensive line is precisely the one for the recurring business. Whoever plans cash flow looking only at the grace year gets a surprise in month thirteen.see the reform →touchesSalta: 70/60 local mining procurementIt defines who counts as a local supplier and leaves the door open to the outsider with the joint venture from 30%. But in this line of business it has a worse hole than in the others and it is worth knowing before investing in the paperwork: the register the province measures compliance with has no energy label, so the category is invisible to the mechanism that ought to push it. Added to the fact that the law says preferentially and sets no penalties, local content here is an auditable sales argument, not guaranteed demand.see the reform →touchesSalta ratified first, and its Gazette publishes the annex that Catamarca’s does not: the 50/50 split is there in writingIt creates no demand for this service, but it orders the shared area where the brine field of one of the projects in the number sits, and its ninth clause says the agreement ceases to have effect once Congress settles the boundary between the two provinces. For energy the effect is indirect: it defines which side the installation falls on and therefore which suppliers' roll applies. In its favor, the protocols are declared a reference base for future projects in the border area, so the regime survives the project that launched it.see the reform →What forces someone to pay for this
The underlying policy that opens up this demand. It is the same inference declared by the rules and the program pillars pushing in this direction.
the RIGI promise is keptWhich projects already buy this
This market does not float on its own: concrete megaprojects drive it. These are the ones moving demand for this niche — each with its investment and status.
The largest RIGI commitment in Argentine lithium: Rio Tinto is building a 53,000 tpa plant at the Salar de Rincón using DLE (nanofiltration)…
see the project →Gold and silver mining in the puna, on the Salta/Catamarca border area: feasibility development, a 3.15 Mt/yr processing plant and full infrastructure…
see the project →Second stage of POSCO’s Sal de Oro lithium complex in the Salar del Hombre Muerto, on the disputed Salta/Catamarca border strip: it adds a 23,000…
see the project →The niche in depth
Who splits the market, where you get in, what pays and what could break it.
already in
split
An Argentine company founded in 1936 prob trade press. It built Lindero's hybrid system, the country's first hybrid mine: 10,908 bifacial panels for 6 MWp prob trade press, 12 MWh of batteries, 30 conversion units and 2 substations, at 3,800 m above sea level prob ibid., with a plant controller designed by its own Argentine engineering team prob Panorama Minero. Agreement with Lindero's operator in 2024, plant operational in September 2025 prob ibid.. And here is the key to the competitive map, AND IT IS NOT AN INFERENCE: the same article declares it responsible for the design, execution and operation of Lindero's energy infrastructure prob ibid.. Its declared business lines are electricity generation, gas compression and treatment, industrial production and materials handling equipment, with sale, rental and fleet outsourcing as commercial arrangements prob trade press. Its model is not to sell the works and leave: it is to stay operating what it installs. Lindero's maintenance is already awarded and does not come back to the market.
Mariana operates 100% off-grid with 120 MW of solar and 288 MWp of batteries, and the solar infrastructure declares USD 190 M within the project's USD 980 M prob company and financial press. It is more capacity than all the rest of the Salta mining fleet put together. Note the unit: the source says 288 MWp of batteries, and that is not a valid unit for storage — either it is solar capacity mislabelled, or it is energy mislabelled, and the available material does not allow a decision. It is cited as it stands, with the inconsistency in plain view, and the reading that enlarges the market is not chosen. Who operates and maintains those 120 MW today does not appear in any public source: it is the most expensive missing piece of data in this whole analysis.
It is migrating from diesel to natural gas plus photovoltaic, with no capacity declared in any source prob provincial press. The mining suppliers' chamber complained in February 2026 that the operator brings in Korean suppliers and turns them into service companies once construction ends prob trade press: it is not future competition, it is a competitor already inside and with Argentine residence.
It is one of the province's four lithium plants in production and it was not possible to determine what powers it — whether diesel, gas or grid unconf no public source. What is documented is that its first-quarter 2026 ramp-up was limited by a gas supply restriction and by equipment maintenance, according to the company's own report verif verbatim in the half-year report. That sentence is the best proof of demand for this niche that exists in the province: firm energy is what holds production back.
208 MWp, 554,580 panels over 350 hectares, 650,000 MWh a year, with a 20-year supply contract prob company and province. It is not mining self-generation: it is a commercial park that already has an operator and a signed contract, and counting it in a market for mine plant maintenance would be inflating. It matters anyway for two reasons: it is some 60 kilometers from the salar corridor prob company and province and competes for the same technical crew, and it is the only reference for the scale of photovoltaic operation running in the Salta Puna.
It built Rincón's camp for Rio Tinto: 26,000 square meters in 18 two-storey buildings, with a kitchen sized for more than 2,000 people, at 3,600 m above sea level and in the 2025-2026 window, with own generation of 4.4 verif the contractor's own project sheet, opened with our own eyes on 2026-08-05. Note the unit: the source writes 4.4 MV, and MV is not a unit of power — it will be MVA or MW. It is an error of the source that is not corrected here, only flagged. What matters about the case: it is the counter-example to the idea that generation always comes imported, because here the package was delivered by a Salta firm.
Three transmission line projects competing for the same corridor: Genneia with EDESA for USD 400 M, up to 300 kilometers and 350 MW prob financial press; Central Puerto with YPF Luz and the World Bank's private arm for 140 kilometers extendable to 350, USD 250-400 M and up to 400 MW prob ibid.; and the national State through a May 2025 resolution under a public works concession prob opinion of the provincial economic and social council. None is under construction as of August 2026. When any of them arrives, mine-site generation goes from being the system to being the backup.
The provincial registry of local suppliers to mining companies classifies its roll with 24 labels and NONE OF THEM IS ENERGY verif official roll dashboard. There are labels for catering (41 bidders), drilling (23), fuel (7) and even vehicle sales (4), but not for energy. The mining suppliers' chamber, which is a different roll, does have the category: 4 members out of 316 verif chamber directory. Under pumping and generation appear ING. RAMÓN RUSSO, GÓMEZ ROCO Y CÍA S.R.L. and ARIZARO INDUSTRIA Y servicio by DUO S.R.L.; under electrical works and supplies, af ingeniería s.R.L. verif ibid.. Two consequences, and both matter: an entrant cannot register under energy, it registers under Services (300 bidders) or Engineering (95) verif official roll dashboard, the two most saturated boxes, and its speciality becomes invisible to the buyer; and the province cannot measure local content in this category, so the mining promotion act does not protect it in practice.
The Diablillos feasibility study budgets ancillary services personnel at 0.24 dollars per tonne milled, USD 10.1 M over the mine life prob operating cost table of the feasibility study, as in-house staff. Not the whole energy system of a mine is outsourced, and a competitive map that does not say so is selling a bigger market than there is.
The large works of new generation are mostly placed already, and competing head-on against whoever built them is the worst fight in this line of business. That does not leave the assembly contractor out: it puts him in step 4, won underneath whoever won the works and not by disputing it. If assembly is your capability, that is your segment; the rest is what repeats every month over what is already installed. It is worth doing so in this order:
Panel cleaning and thermography, as a subcontractor to whoever holds the contract today — low capital, a fast first invoice and no approval process of your own. It is the service the international cost model explicitly names within fixed maintenance, and the one nobody wants to do at 4,000 meters. It is the door that finances year one and the one that shows you the site from the inside. It is worth not confusing market with workload: the TAM counts Mariana's 120 MW and Lindero's 6 MWp, and leaves out — rightly — the 208 MWp of the commercial park at Olacapato, which already has an operator and a twenty-year contract prob company and province. But that park is some 60 kilometers away and calls for exactly the same crew: it does not enter the addressable market and it does enter the calculation of who you will compete with for the technician — it is the only utility-scale photovoltaic operation running in the Salta Puna.
Settling the tax classification before signing — zero capital and it is the niche's margin play. The provincial schedule has a line for network construction, one for network repair and one for mining support services, and none of the three names the maintenance of a generation plant. You request it in writing from the Directorate General of Revenue, with the contract's subject matter drafted exactly as you are going to invoice it — not as the bid document describes it — and with the detail of whether spare parts supply is included, which is what usually moves the classification. Do it before quoting and not after: once the price is signed, the extra point cannot be passed on to the client.
Battery health and medium voltage — the technical step and the one that separates you from the generic competitor. The international model allocates 2.5% of capital per year to storage and includes module replacement there: it is a market that grows on its own as the banks age, and today it has no declared local supply. It requires instruments and trained people, not scale.
Electromechanical assembly as a subcontractor to the main contractor — the high ticket, and it is won beneath whoever won the works, not by arguing over it. It is the route through which the local content law admits an outsider: a joint venture with a Salta partner from 30%. And you do not have to go out and invent the partner: Salta's steel fabrication industry already exists and is already on the roll — IMEC Servicios Mineros, INDASA, METALNOR, MAHR Hnos. Construcciones Metálicas, Galvanizados del Norte and INMMECSA. You form the joint venture with them or you buy the structure from them. It also works out on the tax side, which in this niche runs backwards: for this product you invoice at 2.50% turnover tax as network construction, which is the cheapest of the lines this niche can fall into verif provincial schedule RG 16/2022.
Gas connection engineering — the bet with a clock, and the clock is not controlled by whoever invests: the window closes when the transmission line reaches the salars. There are 321 kilometers of gas pipeline already laid to the puna and at least one operator declaring the migration, but there is no published tariff for the service and no observed contract.
~USD 8.5-10.5 M/year (60-75% of the total market): six out of every ten pesos in this market, and as many as seven and a half, never reach a tender in Salta. Four documented routes: (a) the operator's own chain - Ganfeng resolved Mariana with its own chain, USD 190 M that were not tendered in Salta; the Korean operator of Sal de Oro brings in suppliers and turns them into service companies once construction ends; Eramet operates with its own structure. They are the three owners of most of the installed capacity. (b) the builder that keeps the maintenance - the article documenting Lindero's hybrid puts its builder in charge of the design, execution and operation of the energy infrastructure: that maintenance is already awarded and does not come back to the market. (c) what the operator internalizes - the ancillary services personnel in the Diablillos study is in-house staff, 0.24 dollars per tonne milled. (d) the contractor that brings generation inside its fixed price: Rincón's camp was delivered with its 4.4 of generation inside the builder's package. The 60-75% band is wide on purpose: it depends on a fact that is not public today, which is who operates Mariana's 120 MW.
~USD 3.5-6 M/year, midpoint ~USD 4.5 M/year (~32% of the total market), addressable by a local or national entrant: panel cleaning and thermography, battery state-of-health testing and management, medium-voltage and substation electrical maintenance, maintenance of gensets and dual-fuel power stations, electromechanical assembly as a subcontractor on the large works, civil works for foundations and piling, and gas connection engineering. BEWARE THE 70%, THE 60% AND THE 21.02%: the local content of the provincial mining promotion act (70% of the contracted amount and 60% of the payroll) and the supplier plan committed by Sal de Oro II (21.02% of the amount for suppliers, goods and works, against the 20% floor of the national regime) enlarge the attackable market and are a real, auditable sales argument, because they are in the text of the resolution. But they are commitments on the investment amount, not on this line of business: they are not a quota, they are not guaranteed demand, and the operator meets them by buying transport and earthworks, which is where the volume is. In this niche the problem is worse than in the others: the register the province uses to measure local content does NOT even have THE ENERGY LABEL, so the category is literally invisible to the compliance mechanism.
USD 0.4-1.2 M/year for ONE entrant within 2-3 years. Realistic composition: (year 1) a panel cleaning and thermography contract for one park, or the medium-voltage electrical maintenance of one site, entering as a subcontractor to whoever holds the contract today; (year 1-2) the electromechanical assembly of part of the new generation, beneath the main contractor; (year 2-3) an own annual maintenance contract with guaranteed availability over a 20-40 MW asset. It is a 15 to 35 person SME, it is replicable to Jujuy and Catamarca, where there are equivalent parks and salars, and it is not a unicorn nor does it pretend to be.
When you get paid, and what blocks it
effect
It is a niche of few people and high qualification - the opposite of the camp, and it has to be said that way rather than selling mass employment that is not there. La lectura completa para el que busca trabajo, en la hoja de este nicho para la gente →
calculate it
Concentration Three levels with opposite structures, and the entrant picks the wrong level if it does not separate them.Level 1 - assembly and construction of new generation (49% of the market): concentrated. A single name with completed, documented mine generation work in the Salta Puna, plus the Asian operators' own chains, which own most of the installed capacity. There is no public register of awards, so any percentage would be invented; what can be stated is that of the generation projects with a known builder, there is only one with a public project sheet.Level 2 - recurring maintenance (30% of the market): unknown structure on the large asset, and that is the gap. It does not appear in any open source who operates Mariana's 120 MW today: it may be the operator itself, it may be whoever installed it, it may be staff brought in from outside. At Lindero, by contrast, it DOES appear, and maintenance was awarded within the same package as the construction: those 6 MWp already have an owner and do not come back to the market. The remaining unknown — the province's largest asset — is exactly what makes the directly attackable market move between 25% and 40% of the total, and it is not resolved with more searching: it is resolved with two phone calls.Level 3 - grid and transmission: a regulated monopoly plus three private developers competing for the same route. The provincial distributor, the regional transmission company, the national dispatcher and the regulator that authorizes any transmission work with a certificate whose processing can take up to two years. It is not a market for an SME entrant, but it is what sets the expiry date of part of the niche.And gas conversion has no identified local supply: no Salta company was found on any roll declaring industrial gas connection engineering or regulation and metering stations. The system's players are the provincial company that owns the pipeline, the gas distributors and the national sector regulator.Beware the easy reading: of the 91 categories the industry surveys, 73 have local supply in Salta and 18 do not — and the 18 are heavy machinery, mills and crushers, not this. This niche's gap is NOT one of electrical companies existing: it is one of specialization, of altitude accreditation and of the category having no box to declare itself in.
Who really pays?
In this niche the client is almost never the mining company, and that is the fact that most orders the sale: there are four doors and only the third is the holder operator.
It is the first invoice and the fastest, because the client is not the mine but whoever already won the service and needs hands at altitude. The flip side is the category's visibility barrier: the provincial mining suppliers' register has no energy category — its schedule does not include the activity — so whoever registers with this speciality disappears into the services box and the local content mechanism cannot see them. The correct statement is not that there are no companies: it is that the regime does not register them as such.
Here the buyer is the construction company and not the operator, and that saves you the mine's supplier register. The provincial mining promotion act is what opens the door to the outsider: it expressly allows a joint venture with a Salta partner from 30%, and asks for 70% of the contracted amount towards registered local suppliers and 60% of the payroll. It is worth reading the section before building on top of it: it says shall preferentially contract and sets no penalties — it is a preference with a reward, not a captive market. It is the door that switches off in 2029.
The door with the most margin and the slowest, and both for the same reason: the annual contract with guaranteed availability is signed by the holder and not by a contractor, so the decision goes up to the operations department and sometimes to head office. It is measured in quarters, not weeks, and it almost always comes in as a renewal of someone already inside — which is why the two doors above are the route, not the shortcut. In exchange it is the only relationship that does not switch off when construction ends. The catch is not in the price but in the tax authority, and it is dealt with in «What we watch»: on this line the margin is decided in the classification as much as in the quote.
It is the door with an expiry date and the only one with no anchor of its own: there is no published tariff for the service in the province and no observed contract. What is laid is the infrastructure — 321 kilometers of gas pipeline reaching the puna — and there is at least one operator declaring the migration. It is declared as a bet and not as a market, and that is why the number's midpoint does not rest on this block.
When the window opens
This is not «what breaks it»: it is the dashboard for not signing a contract with the wrong classification. In this niche the expensive mistake is not coming in late or quoting low — it is invoicing on the line that costs twice as much, every month and for the whole life of the contract.
It is the most expensive line in the province's mining services chain — more than mining support services — and that is exactly where part of this niche invoices. The consequence reads the opposite way to what you would expect: the tax authority charges twice as much to repair a network as to build it, meaning the block that switches off on a date is the cheap one and the block that never switches off is the expensive one. And there is a nuance that decides the invoice: the expensive entry is the one for the distribution network; the maintenance of electrical and electromechanical installations has an entry of its own and pays 3.60%. For a newcomer that inverts the order of priorities, because the contract most worth fighting for is precisely the one that is taxed worst. A caveat the resolution itself puts in writing, worth knowing before quoting: the nomenclator's rates are «de carácter orientativo» and «podrán adecuarse a la realidad económica de cada hecho imponible» (art. 2), and the legally applicable rate comes from Ley 6.611 and its amendments. What to watch: the general resolutions that amend the schedule in the Official Gazette — a change of entry here is worth more than any price adjustment, and it does not change the size of the market but who captures it.
Salta's Directorate General of Revenue — schedule of activities and its rate by code, plus the general resolutions amending it in the Official Gazette ↗Three companion signals, and all three move the market without anyone in this line of business being able to do a thing. The three branch lines to the salars: none is under construction today, and when one arrives it switches off at once the rationale for self-generation and the rationale for gas conversion — it is the niche's killer, not its market, and that is why the perpetual block is measured over what is already installed and not over what is announced. The expansion of the largest park, from 120 to 150 megawatts: it is planned and has no announced date, so it enters only in the band's ceiling. And the question that defines half the number: who operates those megawatts today, a fact that is not public and that decides whether the captive share is 60% or 75%.
There are three transmission line projects competing for the same corridor and none is under construction as of August 2026 prob financial press and opinion of the provincial economic and social council: Genneia with the provincial distributor for USD 400 M, up to 300 km and 350 MW prob financial press; Central Puerto with YPF Luz and the World Bank's private arm for 140 km extendable to 350, USD 250-400 M and up to 400 MW prob ibid.; and the national State through a May 2025 resolution prob opinion of the provincial economic and social council. When any of them arrives, mine-site generation goes from being the system to being the backup, and the gas conversion block dies entirely. The date is written into a technical study: Diablillos connects to the grid IN YEAR 3 prob feasibility study. And the flip side pushes in favor of connection, not against it: the region's transmission capacity is already over-supplied by some 500 MW of newly authorized solar verif First Quantum technical report.
The province already has the case: the Rincón Lithium plant (Argosy / Puna Mining), 2,000 tonnes a year, was listed as suspended as of the Aug-2026 cutoff prob industry press and our own survey of the province's plants; there is no operator statement or administrative act backing it, against the backdrop of the roughly 80% fall in the lithium price. An idled asset still needs minimum maintenance, but the contract is renegotiated downward immediately and there is no order book to cushion it, because the unit of sale is the megawatt made available.
Its 100 MW in grinding alone -two 28 MW semi-autogenous mills and two 22 MW ball mills at 40 million tonnes per year verif technical report with effective date 31-Dec-2025- are the largest power figure in the province and are not a market for this niche: the project is fed by a new 122.5 km line tied into the 345 kV grid verif idem. An entrant sizing capacity against Taca Taca gets it wrong twice: it is waiting on a project for which there is no public evidence that it has filed its application to the large-investment incentive regime —the title holder stated in Feb-2026 that it was preparing to file, and the official portal does not publish the detail of the projects under review, so non-filing cannot be verified there unconf status of the filing, and it is waiting on a self-generation market that this project is not going to have. The total demand of 200-250 MW reported by the press is consistent but does NOT appear in the technical report unconf.
Ganfeng resolved the province's largest solar park with its own chain: USD 190 M of solar infrastructure that was not tendered in Salta prob company and financial press. The Korean operator of Sal de Oro brings in suppliers and turns them into service companies once construction ends, as the suppliers' chamber complained in February 2026 prob trade press. And the Argentine player in this line of business appears as responsible for the design, execution and operation of Lindero's hybrid prob Panorama Minero, meaning its model is to stay operating what it installs.
49% of this market is assembly and construction of new generation estim our own calculation, and it switches off with the legal investment deadlines, which expire on 30 June 2029 for Rincón and 31 July 2029 for Sal de Oro II verif resolutions of the large investment regime. What stays alive is the perpetual core of maintenance: on the order of USD 4.3 M/year across the whole province and among all bidders put together estim our own calculation. Whoever builds a construction structure without a maintenance line closes in 2030; whoever builds maintenance alone needs to combine it with another service to reach scale. It is the least favorable peak-to-perpetual ratio of this province's niches.
The repair of infrastructure works or networks pays 5.00% turnover tax on gross revenue, TWICE as much as building them (2.50%) and more than the 3.60% of mining support services verif Annex I of the schedule, codes 422200 note 2 —repair of electricity, gas, water and telecommunications distribution networks— and 429090 note 2 —repair of civil engineering works—. A business in this line straddles two entries rather than one, and it pays to know which is which before invoicing: repairing the distribution NETWORK falls under that expensive entry, but the maintenance of electrical, electromechanical and electronic INSTALLATIONS has an entry of its own and pays 3.60%, or 3.00% for a simplified-regime taxpayer verif code 432190 note 2 of the same Annex I. On a net margin of 8-12%, typical of a technical service, 5.00% of gross revenue eats on the order of 40-60% of the bottom line estim own calculation, and there is no materials cost to dilute it because the tax is on revenue, not on value added. The contrast with the other end of the chain is brutal: extraction pays 0.75%, and 0% with an exemption certificate verif ídem. And the entry is still unassigned: depending on which of the three ends up applying, the same invoice is taxed at 5.00%, 3.60% or 2.50% — up to 2.5 points of gross revenue of difference, which on this margin is half the bottom line. Step 2 of «how to get in» explains how that question is settled.
The province's largest generation asset is published in a unit that does not apply to what it measures: 288 MWp of batteries prob company. MWp is panel peak power and is not a storage unit -either the figure is solar power mislabeled, or it is energy mislabeled, and the available material does not allow deciding which. An entrant sizing crew, spares and stock against a figure whose unit cannot be settled sizes it wrong. You call before investing, not after.
The two conflicts local press recorded at Tolar Grande in the twelve months before the Aug-2026 cutoff were over service contracts and over local hiring, not environmental prob local press; the count of two episodes is our own reading of that coverage. It cuts both ways: it is the way in for whoever brings local payroll, and it is the risk for whoever already won. A blockade on the access road leaves a crew stranded 300-430 kilometers from site estim road distances from Salta city, on stretches of gravel.
The Supreme Court took original jurisdiction on 17 October 2025 in the amparo brought by Kolla and Atacama communities of the basin, with an injunction suspending permits that was still pending resolution as of the Aug-2026 cutoff prob national press and the organizations that are parties to the case. It does not touch the salt flats of the firm ring today, but it is the only front able to halt permits in the province.
How the number is built · and how fresh each data point is
The calculation is installed capacity times an annual maintenance ratio, done separately for three markets with different timeframes and added at the end. It was not derived from any percentage of capex, and there is a concrete reason: the line that looks like the market — the cost of generating — is 62 times larger than the real market, and using it would have made a project not yet built worth more than the whole province.
A robustness check along three paths. The target is the same in all three: the annual maintenance ratio per megawatt, which the page declares to be the input that decides everything. Installed capacity does not need it —it is counted plant by plant and deliberately understated—; the ratio does. (a) The third data point, and it comes from this very dataset. The calculation already converges along two routes that do not touch —the Diablillos technical study gives ~USD 21,000 per MW per year and the US national laboratory model 24,000—. What is missing is a third opinion, and this project had one sitting unused: Neuquén's distributed generation niche publishes O&M of USD 20 per kW per year, that is 20,000 per MW. It is urban photovoltaics —commercial rooftop, no altitude, no isolation, no batteries— and that is precisely why it works as a hard floor: it lands just 5% below the floor adopted here. The three points sit within 20% of each other, coming from a puna mine, a cost model from another continent and an urban market in another province. The floor is anchored three times over. The 35,000 ceiling still has no single point of support, and the page says so: it is the only one of the four figures without a source. (b) The same ratio read as a percentage of capex, using two different capex figures — and the result brackets the band. If annual maintenance of an energy asset runs between 2.0 and 2.5% of what it cost to install, then the ratio per MW depends on one thing only: what an installed MW costs. And there are two such prices in this repository. The urban one: USD 900 per kW turnkey. The puna one: the Mariana plant, USD 190 M for 120 MW, that is USD 1,583 per kW — 76% more expensive. Applying the same percentage to both, the ratio should land between 18,000-22,500 (urban) and 31,700-39,600 (puna). The published band, 21,000-35,000, sits bracketed almost exactly between the two. And there is something finer than the bracketing: the width of the ratio band is ×1.67 and the width of capex per kW between urban and puna is ×1.76. Both widths measure the same thing —the altitude cost premium— by different routes, and they almost coincide. The band is not rhetoric about uncertainty: it is measuring something. Honest caveat: the high end of this sandwich is not independent, because it reuses the percentage and the capex the calculation itself already applies in its route 2; what is new is the low end, which brings in a capex from another province and another market. (c) The scope check, which here is worth more than a third number. This page was defined against an error, and it is worth measuring how close it came: the Diablillos study publishes the cost of generating and the cost of maintaining the generation one line apart, and they differ by 62 times. A third party bills the second. San Juan's power niche legitimately measures the first —capacity times hours times price per MWh—, and if that method were applied here with a realistic solar capacity factor it would give USD 19-26 M/year. And there lies the danger: against the USD 9-29 M this TAM publishes, the two ranges overlap. Confusing the value of the energy with the market for the service would not have produced an absurd figure that leaps off the page — it would have produced a similar one, measuring something else. That the page chose the small line item with the large one within reach is the decision holding up everything else. What still stands: the ratio's floor is anchored by three independent sources and the width of the band is explained by a second route; the ceiling remains an assumption and is declared as such. What none of the three paths touches is the acknowledged understatement: three of the five operating plants do not publish their generating capacity and enter at zero. The bias of this figure, then, points upward — and the gap has the shape of a datum that is not public today: the installed capacity of Sal de Oro, Centenario-Ratones and Rincón.
The number rests on a few variables. The formula shows how it moves when each one changes; and each variable carries its freshness seal — how often it is worth revisiting. estim
How we validate this figure
The ratio that decides the whole calculation — how much it costs to maintain an installed megawatt per year — does not exist published in Salta or in the country, so it was built by two routes that do not touch and we looked at whether they converged. The first comes from the definitive feasibility study of a project in the Salta puna itself: its generation maintenance line, applied to the declared milling rate of 9,000 tonnes per day, gives on the order of twenty-one thousand dollars per megawatt per year. The second is the cost model of the United States national renewable energy laboratory, which sets fixed maintenance of an industrial photovoltaic plant at twenty-four dollars per kilowatt per year and explicitly includes asset management, insurance, site security, *cleaning*, vegetation control and component failure. The two routes converge within 14%, and one is a technical study of a salar and the other a model from another continent. Three assumptions are declared, and all three narrow downwards and not upwards — each is written at the foot of its variable in the table above: plants with no published capacity enter at zero, the construction package enters only by its service fraction, and the gas conversion block has no anchor of its own. None was padded with the comfortable assumption, and the market's midpoint does not rest on the weakest of the three. And a unit trap left in plain view rather than resolved in our favor: the source for the largest park declares its batteries in a unit that is not valid for storage, so either the number is mislabelled or it measures something else — it is cited as it stands, because choosing the interpretation that enlarges the market would be exactly what this method does not do. Finally, the area's commercial solar park is left out: it has a twenty-year supply contract and is not mine self-generation.
How to cite this figure: Despegue (2026). High-altitude energy: maintaining the solar and the batteries already installed in the Puna · Salta. despegueargentina.com/en/salta/operacion-mantenimiento-generacion-boca-mina · terms of use
Where the capital is best placed · the neighboring markets of High-altitude infrastructure, compared
There are 4 RIGI projects in Salta that will buy from this trade, and each one opens its window in a different phase. You already have 9 named competitors on this page. We cross what your company makes or does against the projects that buy this sector and tell you which ones you fit into, when each one buys and through which door. Two pages, with the evidence behind them.
- 10+ years in data science for clients across Europe and the Americas
- Certified in AI governance (ISO/IEC 42001)
- Machine Learning (Google Cloud)
- Registered expert with the European Commission
Get on board the takeoff
Tell us what you are looking for and we will reply. This is what we work on: the map of high-altitude energy: maintaining the solar and the batteries already installed in the Puna and the niches opening up.