Cadillac F1: When the American Dream Splits Into Three Time Zones

In the first three races of the 2026 season, both Cadillac F1 cars finished. Not ten races — just three. But in exactly that window, both Red Bull and McLaren — two teams considered title contenders — each had at least one occasion where a driver was left behind on the track. A debut team, with no prior operating experience at this level, kept both cars running to the line while two front-running teams could not. That was the signal I noted in late March, in the notebook I keep for non-verbal details. Then came the Canadian Grand Prix. From that point on, across the following ten races, there were only two occasions when both Valtteri Bottas and Sergio Pérez reached the finish — the British and Italian Grands Prix. In the other eight, at least one car failed to finish. The team sat bottom of the Constructors' Championship. A project once spoken of in the same breath as championships was now clinging to occasional finishes. The question I carried through the season was not whether the Cadillac was fast. It was obviously slow. The question was why a structure designed with ample resources could not maintain the minimum stability it had itself demonstrated over the first three races. And when Otmar Szafnauer — the former Alpine team principal — appeared on a podcast with a critique aimed squarely at the operating structure, I realised the answer he proposed was not on the track. It was on the office floor plan. Cadillac F1 is the racing team of TWG Motorsport, entering the Formula One World Championship with the ambition of restoring an American automotive brand to the sport's highest level. Symbolically, this is a national project: America returning to a stage where it once left a mark in a different era of F1 history. But that project is not based in Detroit, or Charlotte, or Austin. It spans three headquarters. Design and core engineering functions are based at Silverstone — the heart of Britain's motorsport industrial cluster. The aerodynamics department sits in Indianapolis. The simulation department is in South Carolina. And according to the information I gathered, the Indianapolis facility is still under construction — meaning the team is operating a model whose finished version does not yet exist. The structure is not technically impossible. In the modern world, working across time zones has become normal in many industries — from aerospace to software development. But F1 has a particularity few industries share: the design — test — revise — retest feedback loop is compressed into the gaps between races, often just two to three weeks. Any delay in that loop translates directly into a delay in car development. With Bottas and Pérez, the team holds one of the most experienced driver pairings on the grid. Bottas spent several years at Mercedes, winning races at multiple circuits. Pérez won races with Racing Point and took numerous podiums with Red Bull before joining Cadillac. These are not two drivers learning how to finish a race — they have done it hundreds of times in their careers. That is why a 2-out-of-10 finish rate forces a pause. When a rookie driver hits reliability trouble, it can be chalked up to inexperience. When two race-winning drivers hit the same trouble, the probability of human error falls and the probability of a systemic problem rises. This is a basic principle of professional sports analysis that I learned — sometimes painfully — over years of covering English football clubs: when the best operators keep failing, audit the system before auditing them. By mid-season, the team sat bottom of the Constructors' Championship. The title race — what the project was said to be building towards in the long term — became a distant prospect. And precisely at that moment, Szafnauer appeared on High Performance Racing with an argument aimed at the structure, not at any individual. That is where the story becomes interesting. Szafnauer's argument, put as simply as possible: you can have the best car, the best engine, the best two drivers, but if you split the design, aerodynamics and simulation functions across three separate bases, you cannot synchronise the development loop. And in modern F1, where the gap between leading teams is measured in thousandths of a second, that desynchronisation — however small — is enough to drag you down. This is a technical argument with weight. But it is also an unproven one. Across the entire recorded content of Szafnauer's remarks, there is not a single figure on upgrade lead times, no specific comparison between Cadillac's development packages and their rivals', and of course no statement from the team itself. This is a hypothesis, not a conclusion. But the hypothesis has an important property: it can be tested — or at least, part of it can. The race-results data from the debut season provides some anchor points. Anchor point one: both cars finished the first three races. This matters because it implies the team's baseline capability — the ability to build a car that can run two consecutive races without failing — was established. The question is only whether that capability is sustainable. If the problem lies in the three-site structure, what should we expect? We should expect Cadillac's performance to decline relatively against other teams as the season progresses — because the others are upgrading at a normal cadence, while Cadillac, with a longer development cycle, falls behind in the update race. This is precisely the pattern the data shows. From Canada onwards, both Red Bull and McLaren showed clear improvement. Cadillac, in the same window, did not merely fail to improve — it lost reliability stability. Over the ten races after Canada, only twice did both cars finish. That is not random variation. It is a trend. But caution is required. Reliability and performance are two different matters. A slow but steady car can be the symptom of a conservative but dependable design. A fast but fragile car can be the symptom of an ambitious but unfinished design. A car that is both slow and fragile — Cadillac's case — can be the symptom of several things at once: an insufficient design, a quality-control system that is not tight enough, or a design — build — test feedback loop not fast enough to catch errors. This is where technical analysis must be careful. If we stop at reliability, we can offer two opposing explanations. The first: the car has durability problems in specific components — possibly materials, structural design, or a manufacturing process not yet standardised in a young team. The second: there is a data-correlation problem — meaning what is tested in the wind tunnel and in simulation does not match what happens on the actual track, leading upgrades to underperform, with new parts interfering with existing systems in unforeseen ways. Szafnauer's hypothesis leans towards the second explanation. When you separate aerodynamics from design, and simulation from both, you create a gap through which data must pass multiple conversion steps before being confirmed. Each conversion is an opportunity for error. In an environment where a millimetre on a wing surface can be worth a tenth of a second per lap, small accumulated errors can create large gaps. This principle — known in engineering as correlation failure — is not abstract. F1 teams have experienced it in various ways. When one team's wind tunnel calibrates differently from others, upgrades developed in it can underperform on track. When simulation models are not updated with real variables — track temperature, humidity, altitude, tyre characteristics — simulation output may fail to predict car behaviour. And when engineering groups across three time zones cannot sit in the same room to debate a problem, the speed of detecting and correcting error slows. I have seen this mechanism at a smaller scale while following Brentford FC through their Championship years. When the coaching staff and the recruitment department worked from two different locations, information about a prospective player had to pass through several layers of processing before reaching the decision-maker. Each layer was a chance for nuance to be lost. The result was deals that sometimes worked technically but failed on fit — the right player, the wrong moment, the wrong system. In F1, the system is the car. The wrong system means the car does not run as expected. Interestingly, Szafnauer was not only talking about performance. He was raising a bigger question: what is the goal of this project? In the recorded material, Szafnauer frames the issue as a dichotomy: if the goal is to make Cadillac a more American team — with facilities in Indianapolis and South Carolina, both symbolically significant to the American automotive industry — then building there makes sense. But if the goal is to do everything necessary to win the world championship, then the three-site structure is an obstacle. This is the most analytically important point in the whole Cadillac story, and it is not technical. It is governance. A sports project can be organised around two different objectives: an identity objective (building a brand, a symbol, a national story) and a performance objective (winning, championships, podiums). Both are legitimate. But they require different resource decisions. When a project tries to pursue both without declaring which takes precedence, every downstream decision — including the smallest technical ones — becomes a silent argument about who the project serves. In that framing, the three-site structure ceases to be a purely technical matter. It becomes the physical embodiment of an unanswered question. If the Indianapolis and South Carolina facilities exist for identity reasons — so the project can be presented as an American team — then criticising them on performance grounds is asking the wrong question. The right question is: are those reasons worth trading for car-development speed, and has the leadership acknowledged that trade-off transparently? Meanwhile, on track, the problem continues. Rivals are not waiting. Development in modern F1 happens on several fronts at once. The car needs to be faster in low-speed corners — where new downforce comes into play. It needs to be more stable on long straights — where engine performance and aerodynamic efficiency determine top speed. It needs better tyre management — because over a 300 km race, tyres degrade lap by lap, and the ability to sustain pace late in a stint determines finishing position. Each improvement requires its own development cycle. And each cycle must compete with the others within resource limits. In that environment, the speed of the feedback loop is not merely a competitive factor — it is the decisive factor. The team that learns faster progresses faster. That is why the central question is not whether three sites are good or bad, but whether three sites allow a fast enough learning cycle. Here we should consider a challenge that may not be fully addressed. In the mid-2020s, when simulation tools have developed far beyond where they were a decade ago, does a distributed structure still produce as large a disadvantage as in the past? When simulation is more accurate, when online collaboration tools let teams across three time zones work on the same engineering model, when data travels in seconds rather than days — has the margin of disadvantage narrowed? Szafnauer does not answer that question. But his assumption — that a championship requires a tightly synchronised design loop, and that loop is only achievable when core functions are co-located — is testable. And the debut-season data, while too early to conclude, does not appear to refute it. One further point: the Indianapolis facility is still under construction. This means the team is being judged on an unfinished model. The full version of the structure — with all departments operating in sync — does not yet exist. This could weaken the critique: if you criticise a model based on its transitional state, you may be criticising the build phase rather than the model itself. On the other hand, the transitional phase may be the worst phase. When an organisation is still forming, processes are not set, roles are not defined, and teams have not had time to learn to work together. If even in the transitional phase the team could get both cars to the finish in the first three races, then the question is: as the model matures, will performance improve, or will structural problems become clearer as teams are forced to interact more tightly? These are questions with no external answer. Only the data of the following season can answer them. In the meantime, one important source of information has been under-exploited: the drivers themselves. Bottas and Pérez are not young drivers needing protection from media pressure. They are veterans who have been through every register of the F1 environment. When a driver at that level says the car does not behave as the simulation predicted, that is a technical signal more important than any external analysis. In medicine, this is called a patient-reported symptom — it cannot be replaced by remote observation. But in the available material, there is no comment from Bottas or Pérez on the matter. This is a significant gap. Because if the drivers have spoken about a data-correlation problem, Szafnauer's argument — though unproven — would at least have an anchor from inside the team. Conversely, if the drivers are silent or only speak of reliability problems in general terms, we have less evidence to evaluate the competing hypotheses. The absence of internal voices is a striking feature of this story. In most sports crises I have covered, there is always at least one internal figure who speaks — a driver, an engineer, a technical director. Here, the only voice shaping the story is external. That makes the story more one-sided than reality may warrant. This is why I always look for the unnamed voices — the groundskeeper, the young engineer, the elderly supporter. Not because they have higher professional authority than experts, but because they are closer to events in a way data cannot replace. In Cadillac's case, those voices have not yet been heard. That is why I hold back my judgement. Another point to consider: the origin of the critique. Szafnauer is a former Alpine team principal — a rival team in the midfield battle. This does not mean his critique is wrong. But it means he occupies a structurally adversarial position relative to Cadillac, and criticism from that position should be treated as informed expert opinion, not neutral audit. In sports analysis, tracing the origin of a view matters no less than assessing its content. And there is a subtle point in Szafnauer's argument: he constructs a hypothetical. If you give them a Mercedes car, a Mercedes engine, two Mercedes drivers — that is a powerful rhetorical construction but not an empirical observation. It isolates one variable by holding all others equal. But in reality, variables are never equal. A debut team never has the same resources as Mercedes — not only in car, but in people, processes and organisational culture. So isolating the three-site structure variable by assuming other factors equal is not a fair test — it is a stipulated test. That does not make the argument worthless. It only means we must distinguish two different questions. First: is the three-site structure the sole or most important cause of the problem? The answer is almost certainly no — no single factor exists in a system as complex as an F1 team. Second: is the three-site structure a significant contributing factor? The answer may be yes, and this is the most valuable part of Szafnauer's argument. In either case, the right approach is not to accept or reject the whole argument, but to identify the tests that can discriminate between different hypotheses. Now is the moment to challenge the very argument I have just developed. There is a simple possibility structural analysis often overlooks: the Cadillac has mechanical problems independent of the three-site structure. This is the strongest competing hypothesis, and throughout the story it has never been excluded. Consider the logic. Across the ten races after Canada, both cars finished only twice. That is a poor rate. But it could result from one or two specific components having durability problems — a type of crankshaft, a cooling system, a suspension component — rather than from a synchronisation problem at the organisational level. In F1, even mature teams with perfect structures have endured failure streaks caused by a single detail. Every team has had a season wrecked by one weak part. In other words, the three-site structure can be a genuine weakness and simultaneously not the cause of the current failure streak. The two are not mutually exclusive. The problem is that Szafnauer tends to merge them into one. For F1 teams, there are three main categories of DNF — Did Not Finish — cause. First, mechanical failure: component breakage, overheating, loss of oil pressure. Second, operational failure: a botched pit stop, a wrong tyre strategy, a wrong time to change car configuration. Third, driver error: collisions, mishandling, track-limit breaches. Within that list, only a small portion can be attributed directly to a data-correlation problem — mainly operational errors relating to tyres or car configuration. But all three categories can be affected indirectly by a distributed structure. A data-correlation problem can lead to upgrades that underperform, which then create new mechanical problems. New mechanical problems can force drivers to handle the car defensively, which increases collision risk. That chain reaction — from faulty data to mechanical failure to driver error — is one of the ways a distributed structure can wreck a season without leaving a clear trace. But this is precisely what makes the hypothesis hard to falsify. Because a distributed structure can explain every kind of problem, it risks becoming a universal explanation with no predictive value. If the three-site structure is the cause of everything, it explains nothing specific. This is an epistemological limitation of the argument, not a fault, but it needs to be recognised. There is another way to read the story, which I find more interesting. Instead of asking whether the three-site structure is the problem, we can ask: if the three-site structure is the problem, why is it being kept? What makes an organisation with significant resources accept a structure that outside experts say is sub-optimal? The answer is not in the technical data. It is in the project's governance structure. And this is where the question of purpose — identity or performance — becomes important not only as a philosophical observation but as a practical analytical factor. If the primary goal of the Cadillac F1 project is to build an American sporting presence in F1 — a symbolic and commercial presence, not necessarily a title-winning one within its first few seasons — then the three-site structure has its own logic. It allows the project to serve local and national objectives, to attract media and fan attention on both sides of the Atlantic, and to build long-term infrastructure in locations strategically meaningful to the parent brand. Within that framework, accepting some short-term performance disadvantage may be a rational choice — or at least a coherent one. But if the goal is to win the world championship — as official statements suggest — then the three-site structure becomes a problem to be solved, not a feature to be accepted. And the bigger the stated ambition, the greater the pressure to solve it. The problem is that the two objectives are not compatible at the organisational level. The identity objective rewards presence in many places. The performance objective rewards concentration. A project trying to pursue both will be caught in constant negotiation between opposing pulls. And while it negotiates, rivals are moving forward. Throughout my career, I have seen this pattern at several scales. In Belgrade in the early 2010s, when I started as a young reporter, I watched several Eastern European clubs try to balance maintaining local identity with competing in European competitions. The most successful — such as Dinamo Zagreb in Croatia — typically had a single leader with clear authority and a publicly declared priority objective. The failures typically had leadership with multiple voices and no tie-breaker. The Cadillac F1 story so far has not shown that tie-breaker. And this, perhaps, is the real problem — not the office map, but the absence of a decisive voice in the debate over purpose. Bottas and Pérez do not need forgiveness; they need someone standing beside them, not judging them. That is what I keep in mind whenever I write about any driver in a situation like this. What I learned from those months at Brentford, when matches were suspended and I sat listening to hundreds of supporters talk about their club, is that the loudest applause does not come from the stands, but from the empty seats. What is unsaid is often more important than what is said. In Cadillac's case, the voices of Bottas, Pérez, and the engineers in Indianapolis and South Carolina — those are the biggest empty seats in this story. And a piece like this — if it exists to do anything — should not exist to assign blame to a structure. It should exist to ask the right question. Between now and the end of the season, there are certain signals I will be watching. The both-car finish rate over the final ten races. If the rate remains at 2/10 or worse, the persistent-systemic-problem hypothesis is strengthened. If it improves markedly, the story may be rewritten as a transitional-phase problem, not a structural problem. Any official announcement about changing the three-site structure. If design, aerodynamics and simulation functions are consolidated at one site, that would be indirect confirmation that the problem Szafnauer identified has been acknowledged internally. The emergence of a senior technical figure with cross-site decision rights. The presence of such a person would signal that the team is building a tie-breaking mechanism at the organisational level. Any public statement from leadership about the project's priority objective. This is the most important signal, and the hardest to see — because declaring a priority objective requires admitting there is a trade-off, and not every leadership team is ready to admit that. And finally — perhaps most important to me as an observer — listening to voices from within. Bottas and Pérez, the young engineers, the people working in Indianapolis and South Carolina. When those voices emerge, the story will become clearer. The rhythm of the track can only be heard when you put your ear to the asphalt. At Silverstone, three months ago, I had not yet heard anything clearly.