Baku FP2: Mercedes locks out a 1-2 on softs, but a cluster of technical failures is the real Azerbaijan weekend story

The braking zone into Turn 1 at the Baku City Circuit is barely 200 metres long. With track temperatures touching 30 degrees Celsius in the early afternoon, that braking zone turns every car into a heated block of metal and turns every driver into someone trying to rebalance an entire career in about a second and a half. I was in the technical area behind the Mercedes garage from early morning. On 18 September 2026, the first free practice session started later than planned because of track clearing work, and when the cars finally went out, the first thing I heard was not an engine but a half-finished radio message. Kimi Antonelli reported a loss of hydraulic assistance. Car number 12 was pushed back into the garage inside the first half of the session. Mercedes lost half a session of data for one of its two drivers. By the second free practice session, the timing screen produced a line that made the whole paddock look twice. George Russell 1:43.347. Kimi Antonelli 1:43.899. Mercedes locked out the front two positions. Max Verstappen sat behind them on 1:44.492 on the medium compound, roughly 0.4 seconds adrift. The Ferrari group, with Charles Leclerc and Lewis Hamilton, was close to a full second off the front. If you left Baku that evening convinced Mercedes had rediscovered its glory, you missed the other half of the story. Across the same day, at least six separate technical incidents were recorded on track. A Williams caught fire at the rear brakes. A Haas stopped with an engine failure. An Aston Martin was suspected of an oil leak. A Racing Bulls car hit the wall hard and forced a red flag. Verstappen's Red Bull did not fail, but Verstappen himself complained about the car's handling. A session with six incidents is not a session about speed. It is a session about reliability. And that is precisely the point almost every headline that day left out. Baku has a characteristic that anyone who follows this circuit knows. It does not simply improve its surface over time; it changes the nature of the surface. In the morning the asphalt is cold and slippery and a soft tyre struggles to hold stable temperature at the rear axle. By early afternoon, as air and track temperatures rise together, the rubber starts to grip and the laydown of rubber left on the racing line creates an entirely different level of adhesion. That means any comparison between an early-session lap time and a late-session lap time has to be read on a different scale. A car running 1:45.2 at 13:15 and a car running 1:44.8 at 14:10 are not simply three tenths apart in absolute terms. They are apart in terms of changed track conditions. The circuit runs over 6 km with 20 corners and includes the longest straight on the entire calendar. That straight makes Baku one of the most power-unit-sensitive and tow-sensitive tracks in the sport. A car running alone and a car running behind another can differ by four or five tenths over the same stretch of road, and the cause is neither chassis nor driver. Add the street-circuit characteristic of walls close to the racing line and Baku creates a test in which technical failure is magnified many times over compared with an ordinary circuit. The braking system works at the highest intensity of the season. The cooling system operates under pressure in high ambient temperatures. Any instability at the rear axle immediately turns into a wasted session. The 2026 Azerbaijan Grand Prix weekend follows the familiar format. First and second free practice take place on Friday, each lasting 60 minutes. Third practice and qualifying take place on Saturday. The race runs over 51 laps on Sunday. Every assessment in this piece stays inside the scope of the first two practice sessions, and I will explain at the end why that scope matters. The 2026 grid comprises Mercedes, Red Bull, Ferrari, McLaren, Williams, Audi, Racing Bulls, Haas, Alpine, Cadillac and Aston Martin. That context is necessary to read the time gaps correctly, because half a second at the front means something entirely different from half a second in the midfield. The opening session produced the signal I consider the most important of the whole Friday. Less than twenty minutes after the cars went out, Antonelli reported over the radio that hydraulic pressure had gone. No contact, no driver error, no excursion off track. Just a technical component failing in the literal sense of the word, and it put a Mercedes in the garage for most of the remaining running. For a young driver, this is a far more serious loss than time on track alone. Baku is a circuit where a driver has to build confidence corner by corner, especially through the three heavy braking zones before the straights. Losing half a session means entering the second session with an incomplete data set on braking points and front-tyre grip. What stood out is that Antonelli still finished the second session second on 1:43.899. I was standing at the end of the pit lane when car 12 completed its fast lap, and what I saw was not a driver trying to compensate. It was a driver executing a programme already redrawn for a shortened session. Mercedes moved the entire workload lost in the first session into the second, and did so without sacrificing single-lap pace. The first session also produced two Virtual Safety Car deployments. A VSC is not a penalty and not a competitive incident in the traditional sense. It is an operational procedure activated to protect track workers. But its effect on data is very real. Each time a VSC appears, a group of drivers mid-lap must slow, and that lap is discarded. In a 60-minute session, two VSC periods can erase between fifteen and twenty percent of the total valuable running. That is why the first-session timing sheet in Baku this year should not be read as a classification, but as a set of disconnected data points. With track conditions clearly improved in the early afternoon, the second free practice session was the only Friday running usable for relative comparisons between groups. Russell set the fastest lap at 1:43.347. Antonelli was next on 1:43.899, 0.552 seconds behind his team-mate. Verstappen was third on 1:44.492 on the medium tyre, 1.145 seconds behind Russell but only 0.593 seconds behind Antonelli. This is the first point that needs separating. Comparing Verstappen with Russell gives over a second. Comparing Verstappen with the second Mercedes gives over half a second. But factoring in that Verstappen ran the medium compound while both Mercedes ran the soft, the real gap between Red Bull and Mercedes in qualifying-simulation trim sits somewhere around 0.4 seconds. That is the figure several paddock sources used, and I find it more defensible than either extreme. The reason to choose 0.4 seconds over 1.1 seconds lies in the compound difference. At Baku, the gap between soft and medium in a single fast lap typically falls between four and six tenths, depending on track temperature and on whether the driver finds a tow. When Verstappen is only 1.145 seconds behind Russell on the medium, most of that gap is explained by the tyre. Ferrari is a different story. Leclerc and Hamilton both sat close to a full second off the leading group on qualifying-simulation runs. For a team with a strong Baku tradition built on power-unit performance and stable braking, that margin is a warning rather than a comfort. I should be explicit about the limits of this claim. This is single-session data with no long-run information, no fuel-load information and no engine-mode information for any team. In motorsport, a team can hide three to five tenths simply by turning the power unit down in practice. Ferrari's one-second gap may be real, or it may be a strategic choice. Based on my experience following matches and races, I keep one rule fixed: only trust a gap when it repeats across at least two sessions in the same conditions. Here, the 0.4-second Mercedes-to-Red-Bull gap appeared in both the first and second sessions. Ferrari's one-second gap appeared only in the second. This is the most important section of this piece, and the one handled most thinly in most of the day's reporting. The first incident was Antonelli's hydraulic failure in the first session. The second was a rear-brake fire on the Williams driven by Carlos Sainz. The third was an engine fault that cost Isack Hadjar most of his running. The fourth was Oliver Bearman's Haas stopping with an engine problem. The fifth was a suspected oil leak on Fernando Alonso's Aston Martin. The sixth was Lindblad's heavy impact with the wall, causing severe damage and forcing a red flag. Six incidents across two sessions with eleven teams on track. Statistically, that is an abnormal rate. Notably, these incidents did not cluster around one team or one engine supplier. They spread across brakes, hydraulics, power units and cooling. When failures disperse like that, there are two explanations. The first is coincidence. It is the easiest explanation emotionally and the one teams usually offer in official statements. The second is a shared factor acting across the whole circuit. That shared factor could be 30-degree track temperatures combined with Baku's exceptionally heavy braking, producing thermal loads beyond design on both brake and cooling systems. In that case, the incidents differ in appearance but share an origin. I lean cautiously toward the second explanation, and I want to be clear why. If one team has a brake failure and another has an engine failure under normal conditions, that is coincidence. But when the number of incidents exceeds what a normal weekend produces, searching for a common cause is the methodologically sound step, even without complete evidence. McLaren produced the hardest data to read on Friday. Lando Norris finished the first session fourteenth. Oscar Piastri lost a significant amount of running. The recorded cause was floor damage, subsequently addressed with a temporary plastic zip-tie fix. I want to pause on the zip tie, because it reveals more than anything else about McLaren's day. A modern floor is manufactured to millimetre tolerances, with duct shapes calculated to generate downforce. When a floor component breaks, a team has two choices: fit a new part, or secure it temporarily with whatever material is available in order to keep running. The second choice is not a bad one. It is the choice of a team that understands running matters more than appearance. But it also means the data gathered in that session does not reflect the car's true performance. Read that way, Norris's fourteenth place should not be read as evidence that McLaren is slow. It should be read as evidence that McLaren lost data and running time. What I expect from them on Saturday is a permanently fitted new floor overnight and a long-run programme in third practice to reconfirm the downforce correlation. If that happens, McLaren's Friday gaps become the least valuable data of the entire weekend. Williams brought a large upgrade package to Baku. Audi brought a series of staged upgrades. Those are the two most notable midfield strategic moves this weekend, because they show how teams behind try to close the gap by putting new parts on track as early as possible rather than waiting. Friday did not reward that approach. The upgraded Williams suffered a rear-brake fire. Audi finished in the midfield around fourteenth and fifteenth. This is a pattern I have seen many times, not only in motorsport but in football when a young player is pushed into the first team too early. When you debut an upgrade on track, you do not only bring potential performance. You bring the full risk of compatibility, cooling, weight distribution and interaction between the new part and the rest of the car. An upgrade that debuts and immediately fails does not mean the upgrade has failed. It means the team accepted a trade-off between gathering real track data and protecting a clean session. For midfield teams, real track data is usually worth more, because they cannot afford dedicated test running. Isack Hadjar returned after three races out with a wrist injury. That is a notable story, but it was buried by the very technical failure he suffered: an engine fault that cost him most of Friday's running. For a driver returning from injury, every lap is worth double. Not only because he must rebuild feel with the car, but because he must rebuild strength in the areas the wrist injury directly affects. Baku, with its heavy braking zones and heavy steering, is one of the most punishing circuits for a wrist. With Lindblad, the story is more technical. A heavy wall impact with severe damage can mean a chassis change, and at a street circuit like Baku a overnight chassis change is a serious logistical problem. His participation in qualifying depends more on technical assessment than on any governing-body decision. I raise these two cases together because they point to the same thing. In motorsport, the value of a session is not the position on the timing sheet. It is the number of useful laps the driver and engineers extract. A driver who runs twenty slow but continuous laps may deliver more value than one who runs five fast laps and returns to the garage. Baku is a circuit where tyre strategy usually matters less than elsewhere, because overtaking is relatively feasible on the long straight and the tyre's operating window is wider. That does not make practice tyre data worthless. Both Mercedes cars set their best times on the soft. Verstappen set his best on the medium. That detail matters more to me than the time gap itself. If Verstappen can reach 1:44.492 on the medium, the Red Bull's baseline speed here is not low. At Baku the medium loses roughly four to six tenths to the soft over a fast lap. Switch Verstappen to the soft in the best track conditions and his time likely lands between 1:43.9 and 1:44.0. That is a level competitive with Antonelli and closer to Russell. At the same time, running only the medium in the second session could be a deliberate choice to conceal true pace. Leading teams do this routinely, which makes any direct comparison fragile. The second factor to read is continuous track improvement. Lap times falling through a session is not drivers suddenly going faster; it is the surface changing. Anyone comparing one driver's best lap from minute ten with another's from minute fifty is comparing two different things. In reporting, this mistake appears more often than I would like to admit. I now want to address the part I think needs straight contradiction. Many Friday headlines centred on the idea that Mercedes had opened a large advantage. The wording matters here, because it is stronger than the data permits. The 0.4-second gap between Russell and Verstappen after compound adjustment is real and meaningful. In a Baku qualifying lap, four tenths can separate first from third, even fourth. But four tenths is not a large advantage in the sense the phrase implies. At this circuit, the tow effect can change a lap time by up to half a second within a single lap, depending on whether a driver meets a car ahead. A driver running behind can gain a time advantage over one running ahead. On a weekend where the gap can be reversed simply by choosing the right moment to go out, calling four tenths a large advantage is under-considered. This does not mean Mercedes is not fast. Mercedes is fast. Both of their cars topping the second session, one of them having lost half of the previous session, is a very strong sign of chassis quality and engineering quality. But that is a sign about speed. And speed only decides a weekend when a weekend has no technical failures. There is a phrase in my trade I still use about the transfer market: only a contract with a signature counts as a transfer, everything else is a rumour. In motorsport, the equivalent rule is this: a fast lap only counts once qualifying confirms it, everything else is provisional data. In 2026, as an intern at a sports radio station during the World Cup in Russia, I mispronounced a player's name three times in the first half and was criticised across forums. Rather than hide, I admitted the error publicly and spent a month reviewing footage of all 64 matches. I learned something I carry to this day: a mispronunciation is also the fastest way to learn, provided you are willing to correct it. That principle applies directly to Friday in Baku. If Friday's timing sheet says Mercedes leads and Saturday's qualifying says otherwise, the person who was wrong is the one who used Friday's sheet to conclude about the whole weekend. The window between Friday and Saturday is not time to print more conclusions. It is time to wait for evidence. In 2026, when global football stopped because of the pandemic, I built a tracker of 15 players whose contracts expired in June across three major leagues and published my sources. That piece drew 15,000 views, but what I remember is not the views. What I remember is writing with nothing chasing me. There were no matches that season, yet I wrote the longest news piece of my life, and I learned that readers stay for truth, not for scoops. Applied to Baku: I have no need to write that Mercedes is back, because I have no evidence. What I have is two sessions, a 0.4-second gap adjusted across two different compounds, and six technical incidents in one day. That is good material for analysis, not yet material for a conclusion. I am often asked why a transfer journalist sits down to analyse practice sessions at a motor race. The answer lies in how similarly information is structured. In the transfer market there is a phase I call before the door opens. Conversations have happened, numbers have been exchanged, but no signature exists. People in the trade read this phase through indirect signals: a scout appearing at a match they had no need to attend, a player rested in the final match before the window, an agent suddenly giving interviews. Friday practice in Baku has a similar structure. Teams have exchanged data with the tyre supplier, run simulations, prepared upgrade packages. Practice is simply when the indirect signals surface on track: a driver losing most of his running to a technical fault, a team securing a floor with zip ties, one car on the medium while another runs the soft. Reading those signals does not give you a result. It gives you a direction. What I always repeat in this trade is to separate data from conclusion. The data in Baku is six incidents, 1:43.347, 1:44.492, an upgrade debuting and failing immediately. The conclusion belongs to Sunday. Part of a writer's responsibility in cases like this is to state clearly what is not known. I do not know any team's long-run data in Baku. No Friday session ran a complete long-run programme, and at a circuit with low rear-tyre degradation like Baku, long-run data carries less decisive weight than elsewhere. Even so, its absence leaves a significant hole. I do not know which engine mode Ferrari is running. I do not know whether Mercedes genuinely deployed full power on Russell's fast lap. I do not know whether McLaren will fit a complete new floor overnight or keep the temporary repair. I do not know whether the cluster of six technical incidents in a single day is coincidence or the consequence of a shared factor linked to 30-degree track temperatures and Baku's specific braking intensity. Only third practice can answer that. And I do not know whether Lindblad can take part in qualifying. On all these points I choose to say I do not yet have enough data. There is a habit in sports reporting of turning every information gap into a confident-sounding prediction. That habit is attractive for engagement and damaging over the long term. There are five signals I will track on Saturday. The first is Mercedes' time in third practice. If the four-tenths gap holds or grows, the advantage is real. If it shrinks to one or two tenths, the Friday number was largely a product of tyres and track conditions. The second is McLaren's floor repair. If the new part is fully fitted and they complete a continuous long run, McLaren returns to its expected position. The third is Red Bull's condition. Verstappen complained about the car's handling, and such complaints at Baku almost always relate to the rear axle and downforce setup. An overnight setup change can make a large difference here. The fourth is whether new incidents appear in third practice. If the count rises, the shared-factor hypothesis gains support. If nothing happens, coincidence becomes more likely. The fifth is the state of Lindblad's chassis. It is a purely technical signal, but it can reshape the order behind the leaders. There is one thing I have learned after years in this trade: sports writers read most badly at exactly the moment readers most want a conclusion. In Baku that moment was Friday evening. The timing sheet was up, Mercedes led, and the reader's need was a clear answer. That need is entirely reasonable. But supplying a clear answer at a point where the data does not permit one serves the writer's convenience, not the reader. I once heard about a car's speed from the pit lane before the timing screen showed it, and I once heard about a young player's quality from the stands before any big club knocked. In both cases the lesson was identical: the feeling comes first, but the evidence must arrive before the writing does. My way of handling this is to split the data into two groups. The first contains what is confirmed: lap times, tyre compounds, technical incidents, race-control notices. The second contains what is unconfirmed: engine modes, fuel loads, each team's overnight plan. I write about the first and state the limits of the second. In this trade I am often reminded that the people ahead of their time are those who write the most from the least data without inventing any. That is the standard I set for myself in Baku. If third practice confirms a gap around four tenths, Mercedes enters qualifying with an advantage it can defend. That scenario pushes Ferrari into choosing between chasing a fast soft-tyre lap and sacrificing qualifying to protect race strategy. If the gap shrinks, the weekend story shifts from Mercedes to Red Bull, and from speed to adaptability. Verstappen has shown he can create separation at circuits where his car is not the fastest. Baku, with its long straight and tow effect, is one of the places where late-braking skill has direct value. If McLaren fits its new floor and finds its rhythm, the front group becomes four teams rather than two. That is the best scenario for viewers and the hardest scenario for analysis. And if the technical failure cluster continues into third practice, the weekend result will depend on a variable nobody can predict: luck. A young driver today is an expensive asset tomorrow. I wrote that in 2026, as a third-year student spending an entire national U21 season logging passing statistics for eleven young midfielders. One 1.68m midfielder from U21 Long An had an 87 percent passing accuracy across 18 appearances, and nobody noticed until I published it. That piece reached over ten thousand people and drew a message from a Hanoi FC scout confirming the data. A sports writer today is someone who must decide, every weekend, whether they are writing about the timing sheet or about what the timing sheet means. Baku gave me a beautiful timing sheet. The question for Saturday is whether that sheet tells the truth about a weekend in which Mercedes was both the fastest team and the team with the most reasons to worry.