The Cabildo de Gran Canaria, the island council that sets transport policy on Gran Canaria in Spain’s Canary Islands, had fitted RFID tags to roughly 800 taxis out of a target of about 1,000 when it discovered the tags did not work. The fault surfaced on Monday 14 September 2026 at a fitting session in an annex to the Gran Canaria Stadium, and fitting stopped. A replacement batch of the same tags was brought in, with refitting due to resume on Friday 18 September 2026. That sequence is the story, because the RFID trade press almost never carries it.
What the tags are for
Gran Canaria Airport, at Gando, sits in the municipality of Telde and borders Ingenio. Taxi licences from those two municipalities operate under a special regime, keep unrestricted access to the airport and already carry an identifier. The Cabildo has opened the airport to taxis licensed elsewhere on the island, capped at 20 transfer services per licence per month, a transfer being a run carrying arriving passengers away from the terminal. The Cabildo presents it as the first Canary Islands airport to admit licences from outside its own municipality.
To police the cap, taxis on non-local licences must carry an RFID tag. Fitting is free to drivers, funded by the Cabildo and carried out by SGMAS Software, the technical company contracted for the rollout. The scheme is run by the ConsejerÃa de Transportes, whose director general de Transportes is Manuel López. Fitting sessions ran in mid-September 2026 at the Maspalomas taxi cooperative and at the stadium annex.
A second, separate layer sits alongside the tags. Aena, the state-controlled company that operates Spain’s main airports, has fitted barriers with automatic number plate recognition at the airport entry and exit, in trial from around 15 September 2026, checking in real time whether a vehicle is authorised. The Cabildo’s telematics platform holds the permissions and shares data with Aena’s system. The plate tells Aena which vehicle is at the barrier; the tag is the Cabildo’s own identifier on it. All of it exists to close off the picaresca, the fiddle whereby a driver cancelled a booked transfer to reset the monthly counter.
The fault was found at tag 800, not at tag 10
Canarias7, the Las Palmas daily that reported the failure, says the tags go on the luneta trasera, the rear window, and that they were not working correctly. Telde Actualidad described it as “la última partida recibida presentaba problemas de funcionamiento”, the latest batch received presented functioning problems. Neither phrasing is a diagnosis, and the vaguer of the two is doing a great deal of work.
The order of events is clear enough. Around 800 tags went onto vehicles before anyone established that the batch was bad. That is not a fitting problem, it is a goods-in problem. An incoming batch of passive tags should be sample-tested before a single vehicle is booked in, against what it will have to do: a sample from each delivered reel, on the real glass, read by the real reader and antenna at the real portal, at the range and angle the site demands. A bench read with a handheld proves only that the chip is alive. Even without a formal sampling regime, a first-article check would have caught this: fit two or three vehicles, run them through the lane, confirm they read, then carry on.
Vehicle glass is a hostile substrate
The obvious hypothesis, and it is a hypothesis rather than a finding, is that the problem was never in the tags but in what they were stuck to. A passive RFID tag is an antenna tuned for a particular environment, with the chip impedance matched to it. Put it against a different dielectric and the resonance shifts, the match degrades and read range collapses. That is detuning, and it is the commonest reason a tag that passed on the bench fails on the vehicle.
Automotive glazing is worse than most substrates. Solar control glass carries a thin metallic coating. Heated rear screens carry printed resistive elements across the width of the luneta, plus busbars down each side, and many carry a radio antenna too. That is conductive material directly behind the tag, reflecting and absorbing. Manufacturers know it, which is why they usually leave a small uncoated aperture in the windscreen, behind the rear-view mirror, so toll transponders work. There is no such convention for the rear window.
On a mixed fleet of privately owned taxis the substrate changes with every vehicle, and two cars of the same model with different glazing can behave differently. An earlier report on the Aena plate readers put the existing Telde and Ingenio identifier on the parabrisas, the windscreen, so the mounting position may not be uniform across the scheme either. The RFID glossary covers the terms used here.
The case for a genuinely bad batch
Set against that, the buyer’s own behaviour is evidence. The Cabildo ordered more of the same tags. Had anyone read the fault as a substrate or mounting problem, the fix would have been a different product: a tag built to tolerate a metallised surface, a foam or spacer standoff to lift the antenna off the glass, or a different position. Reordering identical stock says whoever made the call concluded the design was right and the delivery was wrong.
Inlay defects that take out a whole batch are ordinary enough. The chip-to-antenna connection, usually a flip-chip or strap bond, can fail across a production lot if the attach process drifts. Antenna etch or print faults propagate the same way, as do lamination problems or moisture ingress in storage. Such faults cluster by reel and by lot, which is how hundreds of units end up failing alike rather than as a scattering of duds. Note too that tags reading short, tags reading intermittently and tags not responding at all point in three different directions, and no source says which was seen.
What the reporting does not contain
No source names the tag supplier. SGMAS Software is contracted for the fitting and the technical side of the rollout, and no report connects it to the manufacture or supply of the tags. That distinction is easy to blur. No source gives a technical cause, and no official has been quoted on the failure, which is notable given how willingly the Cabildo has put its name to the scheme. Nor does any source state the tags’ operating frequency, the detail that would let a reader judge the glass explanation, since different bands behave very differently against metallised coatings.
That gap is typical rather than exceptional, and it is the wider problem. Vendors publish launches. Buyers publish launches. Neither publishes the batch that came back, so the sector’s institutional memory is built almost entirely out of successes. A public body buying about 1,000 tags, fitting 800 and only then finding the batch faulty is more instructive than any number of go-live press releases, and it surfaced only because a local newspaper was covering the session.
Eight hundred tags to come off
The practical consequence falls on the Cabildo and on the drivers. All of the roughly 800 vehicles already done have to come back in by appointment, have the failed tag removed and a new one fitted, at the Cabildo’s cost since fitting is free to drivers. Taking an adhesive-backed label off a heated rear screen is a careful job if the printed elements are not to be damaged, and residue affects how the replacement sits. Each visit is time a working taxi is not earning.
What is still unknown
As of Monday 21 September 2026 no follow-up reporting has appeared confirming that the Friday 18 September session went ahead, or that the replacement tags work. The scheme may well be running normally, which would be the ordinary outcome. Nobody has yet said so.
The question worth putting to the Cabildo and to SGMAS Software is not who is at fault. It is whether the second batch was sample-tested on real vehicle glass against the real reader before fitting resumed, and if so, what was different the first time round. That answer would be useful to every transport authority planning to put a tag on vehicle glass, which is most of them.
