McNulty's Worlds Gold: The 32km Solo Bet
Stage 11 gave the sprinters’ teams their second clean chance since Puçol, and by every account it was supposed to. Stage 12 of the 2026 Vuelta a España doesn’t offer that kind of mercy again for a while. It’s 166.6km from Vera to the Calar Alto Observatory on September 3, climbing through the Sierra de los Filabres over five categorized tests before finishing at roughly 2,151m — the highest point of this year’s race, by every route guide that publishes stage-by-stage summit altitudes.
Five climbs of increasing difficulty is a shape we’ve seen before this Vuelta. What’s different here is the ceiling. Every prior mountain stage this race has covered — Aitana’s back-loaded 22km, Andorra’s four-climb circuit, Valdelinares’ uniform grind — asked riders to solve a gradient problem. Calar Alto asks that question too, but it adds a second one on top: at 2,150 meters, the same power output simply costs more than it did at sea level. Get the pacing math right on paper and still misjudge the thinner air, and the legs run out before the numbers say they should.
Quick Verdict: Pacing a Stage That Gets Harder Twice
Section Distance/Gradient Role Target Effort Alto del Cabecico (cat. 3) 8.3km at 3.3%, km 16 Leg-opener, doesn’t decide anything 68-75% FTP Alto de CĂłbdar (cat. 3) 4.9km at 4.5%, km 57.5 Second minor test, breakaway sorts itself here 72-80% FTP Collado GarcĂa (cat. 2) 18.5km at 2.4%, km 86.4 Longest climb of the day, deceptively shallow 74-80% FTP — patience, not power Alto de Velefique (cat. 1) 13.1km at 7.2%, cresting ~31km from the line The real selection point 85-92% FTP Calar Alto Observatory (cat. 1, finish) 16.9-17km at 5.6%, summit ~2,151m Decisive climb at the race’s highest altitude 80-88% FTP — thinner air lowers the ceiling Race date: September 3, 2026 Stage distance: 166.6km, Vera to Calar Alto Observatory — roughly 4,460-4,568m of elevation gain (route trackers differ by a few hundred meters, as they usually do) across five categorized climbs The lesson: Velefique’s 7.2% will get all the pre-race attention. The number that actually decides the stage is 2,151 — the altitude where a rider’s engine stops producing what the power meter says it should.
We’ve covered stages with more climbs than this one. Stage 4’s Andorra loop had four, each separated by a descent that handed back some recovery. Stage 12 stacks five, and the first three — Cabecico, CĂłbdar, and Collado GarcĂa — are genuinely minor. An 8.3km climb at 3.3%, a 4.9km climb at 4.5%, and an 18.5km climb that never breaks 2.4% add up to a long day in the legs, not a hard one. Riders who’ve done their homework will treat all three as transport, not terrain, and they’d be right to.
The stage lives in its last 44km, where Velefique and Calar Alto sit back to back with nothing but a descent between them. That’s a structure we’ve seen work as a selection tool before — a hard climb, a short reset, then the finish — but Velefique isn’t a warm-up act here the way Puerto de El Miserat was on Stage 9. It’s 13.1km at 7.2%, opening with double-digit ramps in its first few kilometers before settling into something merely hard, and it crests with over 30km of racing still to come. A rider who spends real matches on Velefique because it’s the steeper of the two climbs is making a specific, avoidable mistake: Velefique is where the race gets sorted, but Calar Alto, not Velefique, is where it gets won or lost.
Here’s the part a gradient chart won’t show you. Peak oxygen uptake declines by somewhere around 5-7% for every 1,000 meters of altitude gained, with the effect measurable from as low as a few hundred meters, according to a review of altitude’s effects on cycling performance published in Sports Medicine. Calar Alto’s finish sits high enough above that onset point that a rider arriving at speed, already deep into a five-hour stage, is asking their body to sustain a wattage number their legs could hit easily at sea level — except the oxygen to support it just isn’t as available.
That doesn’t mean Calar Alto rides like altitude training camp material. It means the last 17km punish anyone who paces off a number they tested at home, at 200 meters, months ago. The same 85% of FTP that felt controlled on Velefique’s lower slopes can feel like 90% by the time the road tips up toward the observatory, and it’s not because the rider got weaker in the intervening 30km. The air did.
Alto del Cabecico (8.3km at 3.3%, cat. 3, km 16): The day’s first test, arriving early enough that nobody with GC ambitions should be spending anything real on it. A breakaway looking to survive 166km needs to use this climb to build a gap, not just survive it.
Alto de Cóbdar (4.9km at 4.5%, cat. 3, km 57.5): A second, slightly steeper minor climb. Same instruction as Cabecico — this is accumulation, not selection, and treating it otherwise costs matches that matter far more four hours later.
Collado GarcĂa (18.5km at 2.4%, cat. 2, km 86.4): The longest single climb of the stage on paper, and also the gentlest average gradient of the five. An 18.5km climb that never exceeds 2.4% is a category 2 by length, not by difficulty — it’s a long, low-grade drag that rewards a steady tempo and punishes anyone who treats “category 2” as a signal to dig in.
Alto de Velefique (13.1km at 7.2%, cat. 1, cresting roughly 31km from the finish): The stage’s first genuinely hard climb, and per Cyclingstage.com’s route guide, it opens with double-digit gradients before easing after the first five kilometers. That front-loaded shape is worth noting on its own — it’s close to the mirror image of Alto de Aitana’s back-loaded profile three stages ago, and it demands the opposite instinct. Control the opening ramps here. The climb gets more manageable as it goes, not less.
Calar Alto Observatory (16.9-17km at 5.6%, cat. 1, the finish, ~2,151m): The decisive climb, and one more front-loaded surprise: several steep ramps in the first three kilometers, into double digits by some route guides, before the gradient settles toward its 5.6% average for the bulk of the ascent. A 5.6% average sounds survivable next to Velefique’s 7.2%. It is survivable, on fresh legs, at sea level. It’s neither of those things by the time the peloton reaches it — tired from 150km and five prior climbs, and finishing at the highest point this Vuelta will visit, at an observatory that sits above the Tabernas desert and normally requires a booked guided tour to visit, not a rider who can just ride up.
Racing or riding to a summit above roughly 2,000 meters changes the pacing question from “how hard is this gradient” to “how hard is this gradient, given that my legs and lungs are working with less oxygen than they were an hour ago.” Whether the finish is Calar Alto or a local mountain pass that tops out somewhere similar:
You can’t replicate 2,150 meters of altitude on a home trainer, but you can train the pacing discipline this stage demands: a front-loaded hard climb, a controlled recovery, and a final effort you deliberately under-pace early to protect the finish.
Warm-up: 15 minutes progressive, ending with two 30-second openers at 100% FTP.
Block 1 — “the minor climbs”: 35 minutes at 70-80% FTP (RPE 4-5/10), with three short surges of 3-5 minutes at 82-88% FTP scattered through it to simulate Cabecico, CĂłbdar, and Collado GarcĂa. Nothing here should feel taxing by the end.
Block 2 — “Velefique”: 20-22 minutes total — the first 5 minutes at 90-95% FTP (RPE 7-8/10) to simulate the steep opening ramps, then 15-17 minutes at 85-90% FTP (RPE 7/10) as the gradient eases.
Block 3 — “the descent”: 8-10 minutes at 55-62% FTP (RPE 2-3/10). Actually recover here. The temptation to keep pushing is the same mistake a rider chasing time on the real descent would make.
Block 4 — “Calar Alto”: 22-25 minutes total — the first 3-4 minutes at 88-93% FTP (RPE 8/10) for the steep opening ramps, then 18-20 minutes at 78-84% FTP (RPE 6-7/10), deliberately below what Velefique felt like, to build the habit of discounting effort on the climb that actually decides the stage.
Cool-down: 8-10 minutes easy.
Total time: 100-115 minutes.
The test: Compare your average power across Block 4’s steady section to Block 2’s steady section. It should be lower — that’s the point, not a failure. This session trains restraint on the second climb specifically because a stage like this punishes anyone who rides the harder-looking gradient (Velefique) and the higher-altitude gradient (Calar Alto) as if they cost the same. If your FTP number is more than six or eight weeks old, retest it before setting targets off it — a day this dependent on precise discounting is a bad place to be pacing off a stale number.
Stage 12 isn’t the hardest climbing day this Vuelta will produce — Stage 9’s Aitana packed more total elevation into a shorter route, and later mountain stages will ask harder gradient questions than either Velefique or Calar Alto. What Stage 12 has that no other stage in this race can claim is altitude: a finish above 2,100 meters that turns a well-paced power number into an educated guess once the oxygen thins out. Velefique will get the attention because 7.2% looks harder than 5.6% on paper. Calar Alto is the climb that actually decides the day, because it’s the one where the numbers on your head unit stop telling the whole story.
The transfer to your own riding doesn’t require a Spanish observatory. Any climb that tops 1,800-2,000 meters — plenty of mountain passes in Colorado, the Alps, or the Rockies qualify — deserves a discount on your normal power targets, applied earlier than feels necessary. Pace the steep opening ramps with control. Save real restraint for the altitude, not just the gradient. That’s the whole lesson, and it’s one this Vuelta hasn’t had to teach until today.
Stage distance, climb data, and finish altitude from La Vuelta’s official stage 12 page and Cyclingstage.com’s stage 12 route guide; total elevation-gain figures vary by a few hundred meters between route trackers. Altitude physiology figures from a peer-reviewed review of altitude’s effects on cycling performance. GC standings entering the stage from Stage 10 reporting, referenced in our Stage 10 preview. Training targets are estimates based on power-based training principles. Adjust for your own tested FTP and conditions.