Fifteen Times Per Kilometre, a Third Per User: the Two Units That Decide Argentina's Mining Push-to-Talk

There are two ways to describe the cost of putting a radio network over an Argentine mine, and they disagree by a factor of five and a half.

The first is per square kilometre. A lithium brine operation on a salar covers something in the order of 180 square kilometres of salt flat. Covering it takes about USD 1.3 million of radio and backhaul, which works out to USD 7,222 per square kilometre. An underground copper mine is a different animal: perhaps twelve kilometres of drift and decline, USD 1.32 million of infrastructure, and USD 110,000 per kilometre covered — fifteen times more.

The second is per user. That same brine field carries around 500 people on shift, so its coverage capital is USD 2,600 per user. The mine carries 1,400, so its coverage capital is USD 942.86 per user. Per user, the mine is cheaper — by 2.76 to one.

Both sets of numbers are correct. Both come out of the same two capital figures. And no document that carries only one of them is describing a market. This is the finding at the centre of Pheonix Research’s Argentina mining push-to-talk report, and it is the reason the country’s communications demand is going to behave strangely for the rest of the decade.

A salt flat’s problem is area. A mine’s problem is the first kilometre.

A network designed for a salar is a small number of high sites aimed across a large empty plain. The engineering problem is distance and line of sight, and the answer is a tower and a directional antenna. One site can serve six hundred square kilometres if the ground is flat, which on a salt flat it is. The constraint is not the network at all — it is that the people are spread across the middle of nowhere doing a slow chemical extraction, and there are only five hundred of them.

A network designed for a decline is the opposite. There is no line of sight. There is rock. The signal has to be delivered along a corridor, which means leaky feeder or a chain of distributed antennas, each radiating element cabled back to the next, running the length of the drift and down every branch that has people in it. The cost structure is linear in metres, not in area. And because the metres are dense with people — 1,400 of them in a working mine — the per-user figure lands in a completely different place from the per-kilometre figure.

The spread across site archetypes is worth stating plainly, because it is smaller than most operators expect. Capital per user runs USD 3,711 at a copper-scale site, USD 4,100 at a lithium or gold site and USD 4,840 at an exploration camp — a range of only 1.30 times end to end. The spread in coverage cost per square kilometre is 15.2 times. The reconciliation between those two statements is density, and density is the one variable in this market that a mine cannot choose. It is a geological fact before it is a commercial one.

The plan book has a sequence, and the sequence is the story

Argentina’s large-investment regime, RIGI, is usually described in terms of what it does to a tax bill. Its more consequential property is temporal. The Ministry of Economy’s public dashboard recognises a project as approved only once its resolution is published in the Boletín Oficial, and it publishes planned investment by sector, by commodity and by year. That means the schedule of future mine construction — and therefore the schedule of future communications demand — is on a government website, in advance, for anyone who wants to read it.

Read that way, the mining series runs from USD 2,579 million of planned investment in 2026 to a peak of USD 4,128 million in 2028, then declines to USD 105 million by 2033. USD 18,334 million of plan across the forecast window. Total push-to-talk spend inside the same window is USD 198.71 million, which is 1.08 per cent of it — and 0.061 per cent in the base year alone. One thousand six hundred and forty to one.

Now put the two findings together and the market’s shape appears. Lithium outruns copper 1.8 to 1 across the 2024–2026 plan years. Copper outruns lithium 4.37 to 1 across 2027–2031. Argentina is building its worst coverage economics first: the salars, the high plateaus, the sites where 500 people are spread across 180 square kilometres and the network has to reach all of them. Then, roughly from 2027, it gets density — four to five times the users per covered kilometre, arriving in the form of underground copper.

A supplier who read the 2025 market as a proxy for 2030 would misprice both halves. He would overbuild for a brine field, having learned that coverage is expensive, and underbuild for a mine, having learned that it is cheap per head. The two halves of the same country price the same network in opposite directions.

The horizon decides it, not the datasheet

There is a second version of the same mistake, and it lives in the amortisation table rather than the site survey. Take the lithium-scale site: USD 1.3 million of radio and backhaul. Amortise it over five years and it costs 260,000 dollars a year, which with service of 186,000 and support of 24,000 comes to 470,000 — USD 940 per user. Amortise the identical asset over ten years and it falls to 130,000, taking the all-in figure to 340,000, or USD 680 per user.

A 27.7 per cent reduction in cost per user from a change in accounting horizon and nothing else. No new hardware, no negotiation, no competitive tension. Just a longer view of an asset that was always going to last twenty years.

This is where the architecture argument is actually won, and it is not won on capability. A leaky feeder is not a better product than a public cellular subscription. It is a capital asset being compared against an operating line, by a procurement team working to a five-year business case, for an owner who has just signed a stability guarantee specifically so that a twenty-year view becomes rational. The supplier arguing the short view is arguing against the instrument his customer has just committed to.

What 1,640 to one does to the specification

If push-to-talk is six-hundredths of one per cent of the capital programme it serves, nobody senior is in the room when it is specified. That is not a criticism of the buyers. It is arithmetic: a project director with USD 3,000 million of construction to deliver is not going to spend a week on a decision worth USD 400,000. The specification gets delegated, and when it is delegated it gets inherited — from the last site, from the EPC contractor’s template, from whoever answered the phone during the feasibility study.

Which produces the number that gives the market its structure. Devices are 15.6 per cent of a site’s capital. Radio and backhaul are 63.4 per cent. The buyer spends his attention on the fifteen per cent, because a handset is a thing you can compare against a rival handset on a datasheet, and the sixty-three per cent — the network, the coverage, the thing that actually determines whether the mine can talk — gets decided by the inherited template.

So the moat is not the handset. It is the file: a radio survey, a spectrum authorisation, an intrinsically-safe equipment schedule for the classified areas, a safety case that ties coverage to emergency response times, and increasingly the integration layer that connects voice to fleet management, autonomous haulage and the mine’s identity provider. Nine to eighteen months of work, specific to one site, completed before a purchase order exists. A competitor with a cheaper box does not win next year’s mine, because next year’s mine was specified before this year’s box existed. On Pheonix’s modelling no supplier clears fifteen per cent of the combined pool across all four layers, which is why the report counts zero tier-one players — the concentration that exists sits inside the radio layer, where the top two hold a modelled 64 per cent, and evaporates as soon as you integrate across layers.

The loss that is not a loss

Lithium’s share of push-to-talk value falls from 42.9 per cent in 2026 to 28.1 per cent in 2033. That sentence reads like a decline, and it is the single most misreadable line in the report. Lithium’s absolute spend rises over the same period from USD 4.80 million to USD 11.23 million — a multiple of 2.34. The share falls because copper grows faster, not because lithium shrinks.

The commercial consequence runs the wrong way from the headline. A supplier who reads share loss as demand loss will under-resource the lithium installed base in exactly the years it generates its most reliable revenue — brownfield expansion, additional shifts, contractor mobilisation, the boring repeat business that arrives without a tender — while chasing copper projects that will not commission until close to the end of the decade.

The same trap sits in the coverage economics. Contractors reach 20.7 per cent of users by 2033, and contractors buy service rather than infrastructure. That is a mix shift, not a market shift: the same people, doing the same work, with the capital line sitting on somebody else’s balance sheet. A forecast built on owner capital budgets will miss a fifth of the demand and then be surprised by the revenue.

What would make this wrong

The forecast is a count, not a trend. There are roughly 45 push-to-talk sites in the mining sector in 2026, and the ten largest buying entities account for 61 per cent of value on Pheonix’s modelling. Four copper mines have to reach operation inside the window for the base case to hold. Each one that slips past 2033 removes a dense site and the 1,000 to 1,400 users attached to it, and a single deferral moves the national number by four to seven per cent.

So the band is wide and lopsided: USD 30.60 million downside, USD 39.95 million base, USD 46.90 million upside — 23.4 per cent below against 17.4 per cent above. The asymmetry is not pessimism about mining. Slippage subtracts, because a site that is not built contributes nothing, while upside has to be earned by several things going right on the same site at the same time: copper on schedule, video and data joining voice on one platform, contractor fleets fully served.

And the honest limitation is that a 19.95 per cent compound rate is not a description of smooth growth. It is what a series of discrete commissioning decisions averages out to across seven years. Between any two of those decisions the line is flat. In a market this size, that is not a rounding error — it is the whole experience of being in it.

None of which makes the market unimportant to the people inside it. A mine that cannot talk stops. The interesting question is not how big the push-to-talk market becomes. It is which unit the next business case is written in — per user or per kilometre — because on the answer to that question, two identical networks in the same country come out looking like a bargain and a disaster.

Figures in this article are drawn from Pheonix Research’s Argentina Mining Push-to-Talk Market: RIGI-Driven Demand, Size & Forecast 2026–2033, which models site capital stacks, coverage economics and scenarios from the RIGI plan book and Pheonix’s own critical-communications coverage. RIGI parameters are stated as the operative terms of the establishing law and were not re-read from the consolidated text for that edition; site capital figures are models, not quotations.