23.67 seconds – the number that made the Safety Car the race's strategy

When the field arrived in Shanghai a single Grand Prix had been run with the 2026 cars. China was also a sprint weekend: one hour of free practice before Sprint Qualifying, and Pirelli noted after Friday that the teams had run mediums and softs with no significant long runs – the hard had been saved, per F1.com. Going into the sprint there was no real race-distance data at all.

Pirelli had gone a step harder than in Melbourne: C2 as hard, C3 as medium, C4 as soft. The worry was the front-left. Shanghai is a strange mix – the long Turns 1–3, the fast 7–8, the kilometre-long back straight – and the smooth 2024 asphalt had produced heavy front graining the year before, especially in the sprint. A year of ageing should have helped, but nobody knew how much. On top of that Pirelli pointed out that the combination of corners and straights would stress energy recovery in the new power units, per F1.com. New car, new tyres, effectively a new track, graining, a new energy system, a sprint format, almost no practice – all at once.

Most started the sprint on mediums. Racing Bulls put Liam Lawson on hards, reasoning that the C2 would resist graining better. It did: when the Safety Car came on lap 13 Lawson and the others on hards could stay out and gain places, P7 at the flag, and the team said afterwards that the unusual choice had worked as hoped, per F1.com. The whole paddock got a data point: the hard works, and appears very durable.

But the hard was also a limited resource. On a sprint weekend each driver has two sets, and those who used one in the sprint had only one left for Sunday – Pirelli pointed out that this could limit their options. Strategy was not just about which tyre was fastest but which ones you had left.

By Saturday evening Pirelli were clear: one stop clearly fastest, medium → hard with the stop on laps 17–23 or soft → hard on laps 15–21, very close in race time. The soft was still a question mark – nobody except Hadjar in the sprint had run the C4 for more than nine laps. Red Bull took the risk: softs on Verstappen for the start. Mercedes and Ferrari, at the front, chose control: mediums.

The starting order was Antonelli, Russell, Hamilton, Leclerc – and McLaren P5 and P6. Neither of them started. Norris never reached the grid: an electrical fault on the power-unit side was found during preparations, and the mechanics could not fix it before the pit lane closed. Piastri made it to the grid, but shortly before the formation lap a different electrical fault, also on the power-unit side, was found and the car was wheeled back. Andrea Stella stressed that they were two separate faults, per McLaren. Two DNS – Piastri's second in a row after the reconnaissance-lap crash in Melbourne. For Mercedes and Ferrari, two cars that would otherwise have shaped both pit windows and traffic were gone.

Before the race F1 put the pit-stop time loss at 23.67 seconds, calculated with 2.5 seconds stationary, per F1.com. The actual stops bear it out: Hamilton 22.4, Lawson 22.7, Antonelli 23.0, Russell 23.4, Verstappen 23.7 in total pit-lane time. Under the Safety Car we measured 13.4 – ten seconds cheaper. That is the whole race's strategy in two numbers.

Hamilton took the start and led the first lap, Antonelli went back past on lap 2 and Russell, briefly behind both Ferraris, was second from lap 5. Hadjar on softs stopped as early as lap 1. Lawson stopped on lap 9 – the lap before. On lap 10 the Safety Car came, and the calculation was simple: without an SC, mediums to laps 17–23 then hards; with one – take the discount now and try to make the hard last 46 laps. Antonelli, Russell, Hamilton, Leclerc and almost every medium starter came in. Russell rejoined in traffic behind Colapinto and Ocon, who had started on hards and stayed out, and had to work his way back, per F1.com. A cheap stop is not the same as good track position.

Click a scenario to load it in Simulate below.

The question on lap 10 was whether the C2 could last 46 laps. The answer was more than yes: degradation was 0.000 s/lap – the slopes of Antonelli's, Russell's, Hamilton's and Leclerc's stints are zero – while the C3 medium degraded at 0.093 s/lap from graining. Ten drivers ran more than 45 laps on the same set, and Antonelli set the fastest lap, 1:35.3, on lap 52, per Pirelli. An old hard was not a slow hard. The simulator says the risk was smaller than it felt: even had the tyre not held, a second stop would have cost Antonelli 35 seconds and three places – and Mercedes had 0.6 s/lap over Ferrari to draw on (97.16 against 97.76 fuel-corrected).

The hard starters – Colapinto, Lindblad, Bottas – did the opposite and stayed out when everyone else came in. Colapinto ran second on lap 11. But behind them came cars on fresh hards against their old ones, and track position without pace is temporary: by lap 25 Colapinto was 24 seconds back, P10 at the flag. Racing Bulls said the Safety Car compromised the hard starters' races in particular, per F1.com. Lawson, who had stopped the lap before the SC, got "completely the wrong moment" from Alan Permane – but the simulator prices the bad luck at 2.7 seconds: the Safety Car bunched the field straight afterwards and ate the difference. P7 regardless.

Pirelli were right before and after: one stop was fastest, the hard was the race tyre. The Safety Car moved the window, not the strategy. But what the teams took away was bigger than the forecast: the C2 was in practice unwearable, a stint of 45+ laps was possible, and the medium – not the soft – was the tyre graining hit. The clever thing in Shanghai was not an ingenious alternative strategy. It was updating the one-stop plan the second the Safety Car changed the price of the stop, without knowing for certain the tyre would last. Two races into the season the field had learned its first rule for 2026: the neutralisation is the cheapest pit stop there is, and the tyres last longer than the model says.

How we did it: lap data from FastF1 (F1 Live Timing). Degradation = within-stint slope on green laps, fuel-corrected at 0.035 s/lap. Pit loss under green per F1's official 23.67 s (our own measurement on three green stops was too high and is not used). The simulator runs everyone else on their actual lap times; under the Safety Car the order is frozen. Sources: F1.com (pre-race facts, sprint analysis, race report, team comments), press.pirelli.com (post-race note), McLaren (race report), FIA timing.