Cadillac F1: Three Bases, One Car, and a Contract Nobody Signed

The Canadian Grand Prix ended in Montreal, and across the ten rounds that followed, the two Cadillac F1 cars crossed the finish line together exactly twice. Two out of ten. For anyone who has spent time inside a technical garage, that ratio points in a single direction: this car is not breaking because of a component. It is breaking because of a system.

I stayed behind in the hotel cafe, reopened the results sheet, and read it three times. The first pass I looked for driver error. The second pass I looked for engine failure. On the third pass I stopped looking for fault and started counting two things: how many races this team finished with both cars, and how many times this team was described as a World Championship project.

Those two sets of numbers sit on different pages. The gap between them is where the story begins.

Context: a new team, three addresses, and a charge levelled from outside

Cadillac F1 is the Formula 1 team of TWG Motorsport, entering its debut season with two drivers who have won races at the highest level: Valtteri Bottas and Sergio Perez. That is a driver pairing most debutants in history could not dream of. Their experience is an asset. It is also an analytical trap. When two race-winning drivers both fail to finish, the probability of human error falls, and the probability of machine and organisational error rises.

On debut, Valtteri Bottas retired. Sergio Perez finished sixteenth. Over the next three rounds, both cars finished, a feat that Red Bull and McLaren, across the same stretch, could not claim. Then Canada arrived. From that point, Cadillac F1 lost consistency while Red Bull and McLaren showed evident improvement. The team sits at the bottom of the Constructors' Championship.

Then Otmar Szafnauer spoke.

He is a former team director who ran operations at Racing Point, Aston Martin and Alpine. On a podcast called High Performance Racing, he advanced an argument that shifted the story from a struggling newcomer to an organisational model tying its own hands. He said Cadillac placed design in Silverstone, aerodynamics in Indianapolis, and simulation in South Carolina, with the Indianapolis facility still under construction.

To sharpen the point, Otmar Szafnauer constructed a hypothetical: put Cadillac F1 on equal footing with Mercedes in every respect — same car quality, same engine, same drivers — and who wins? His answer: Mercedes. The reason he gave is not technical. It is that three engineering centres in three places leave no feedback loop fast enough to convert data into decisions.

That is a claim about process, not about speed. It forces the reader to dissect it on its own merits rather than nod because the speaker holds an impressive title.

The core: when three bases co-sign a contract nobody can see

In this file I am not trying to find a name to blame. I am looking for a structure.

Some contracts are signed on the pitch, and some are signed in the dark. The contract Cadillac F1 is executing exists in no document. It is an unspoken arrangement between Silverstone, Indianapolis and South Carolina: each holds part of the design loop, and none holds all of it. When the car does not behave as simulated, nobody walks out of a room to say their data was wrong, because each site's data is correct in its own terms.

The technical term is correlation failure. The wind tunnel, the CFD simulation and the driver-in-the-loop simulator return one result; the real car on track returns another. When that gap is large and persistent, upgrade packages are approved in the laboratory but fail to convert into lap time. You spend, you work, you deliver on schedule, and you are still slower.

The crux is the speed of the feedback loop. Each time the wind tunnel at Silverstone produces a result, that data must travel to Indianapolis, be reconciled with aerodynamic data, then return to South Carolina for simulator validation, before anyone with sufficient authority signs off a part for the track. Three time zones. Three queues. Three sets of internal standards that can drift apart.

Time is the one variable money cannot buy.

In modern Formula 1, competitive advantage does not come from what you discover. It comes from how fast you turn a discovery into a part on the car. A single-campus team can meet in the morning, revise in the afternoon, and send drawings out the same day. A three-campus team needs extra time for every loop. And time, across a long season, is a resource that cannot be topped up.

The track data sharpens the picture. After Montreal, only two of ten races had both cars finish: Britain and Italy. That is descriptive data, not inferential data. It does not prove a cause. It narrows the set of possible causes.

That set holds at least four possibilities. One, a car whose design is structurally fragile. Two, correlation failure between simulation and track caused by the split sites. Three, logistics and supply-chain strain from operating across three continents. Four, two drivers in a debut season making more individual errors than usual.

The source analysis picks the second. It has reason to: both drivers are race winners, which weakens the fourth. But it picks before eliminating. That is the single largest methodological weakness in the whole story.

There is a technical detail the original piece does not exploit, and it matters: the Indianapolis facility is still under construction. That means the team has never experienced the fully operational version of the very model it is being criticised for. Judging a model in a transitional state and then concluding about its steady state is a timing error, the kind of error my line of work calls concluding before the sample is complete.

Cost: the hinge nobody calculated

One dimension the original piece never touches, yet anyone who has worked with sports budgets sees immediately: a three-base model carries structurally higher fixed costs and greater duplicated headcount than a single campus.

Formula 1 operates under a spending limit set by the FIA. That mechanism constrains operational spending, while infrastructure spending such as buildings, wind tunnels and fixed equipment is treated differently. The consequence is a specific cost squeeze: you can afford to build three bases, but you must pay for three operating machines out of the same capped budget.

Put differently, you can buy bricks, but you cannot buy more people.

Alongside that sits the aerodynamic testing allocation tied to championship position. Lower-placed teams are allowed more wind-tunnel runs and CFD work. In theory, a debutant at the back of the grid has an advantage in converging faster than the front-runners. If that team still fails to converge, the organisational hypothesis gains weight, but only gains weight, not proof.

And there is a further point the original piece skips entirely: three bases on three continents may be a sporting cost, but they may also be a commercial asset. A US facility serves sponsor activation and market development in a way a purely British campus cannot.

If leadership designed that structure partly around commercial objectives, judging it purely by lap time misreads the design intent. This is not a defence. It is a missing variable in the equation being solved.

Governance: two objectives, one leadership, one unanswered question

This is the part I consider most important, and the part buried deepest in the story.

Otmar Szafnauer raises an issue he does not answer directly, but it opens up the entire governance question of the project. He says that if the goal is to make the team more American, then building in Indianapolis and South Carolina makes sense. But if the goal is to do everything necessary to win the World Championship, then that structure must be justified on performance criteria, not identity criteria.

The two objectives can coexist. They do not always point the same way. And when they conflict, an explicit order of priority is needed to adjudicate.

In an organisation with three engineering centres, ambiguous priority has a direct consequence: it strips away the tie-breaker when centres compete for resources. If Silverstone wants more headcount for aerodynamics, and Indianapolis wants more for structural validation, who decides? On what basis? If the answer depends on which centre currently has leadership's attention, the machine is being run by internal politics rather than performance data.

For a new project, the biggest risk is not choosing the wrong strategy. It is failing to choose.

Notably, the original piece names no technical director, no team principal, no chief executive at Cadillac F1. In an article whose entire argument concerns the separation of functions, the absence of any named technical authority is itself a signal. It suggests the project's command structure remains undefined.

The contrarian angle: a charge that may be right and still unproven

Now I have to say what a well-mannered commentary would not.

Otmar Szafnauer's argument may be correct. But the evidence presented does not establish that it is. There are at least six reasons to withhold judgement.

First, single-sourcing. The entire charge comes from one person on one podcast, with no formal response from Cadillac F1 or TWG Motorsport. A piece about governance that carries no voice from the party under scrutiny is one-sided, however senior the speaker. Investigative journalism does not discharge its duty by quoting one side and closing the page.

Second, the speaker's position. Otmar Szafnauer is a former director of rival teams. He holds technical authority. He also occupies a structurally adversarial position. There is no need to impute bad motive, only to record a fact: speaking from outside, you pay no price for being wrong. The corrective feedback internal decision-makers face is absent for outsiders. That does not make his argument false. It means the argument needs independent testing.

Third, the hypothetical. The claim that Mercedes still wins with everything equalised is a rhetorical construction, not a finding. It isolates one variable by stipulating the others are equal, and that equality is asserted, not observed. Rhetorically it is powerful. Evidentially it is hollow.

Fourth, the early data pushes back gently. Across the first three rounds, Cadillac F1 finished with both cars when Red Bull and McLaren could not. Over a three-race sample that is not proof of excellence. It is also not proof of impotence. It is a signal that the team's underlying capability is not the bottom of the problem.

Fifth, the timing error. Indianapolis is still being built. Judging an incomplete model and concluding about the model exceeds the data.

Sixth, and this is the point I want to stress most: two risks are being conflated. The first is car reliability, measurable, countable, with figures attached. The second is organisational coherence, structural, asserted by an outsider. The piece uses the first as proof of the second. That inference is not licensed by the data.

The second blood sample does not lie; only people lie. Here, the track data says one simple thing: there is a problem. It does not say where the problem is. People fill that blank, and each fills it differently depending on who they are protecting.

An undervalued asset: the driver pairing

There is a detail both the original piece and most surrounding commentary overlook. Cadillac F1 holds two drivers who have won races at the highest level. In the history of debutant teams, that is a rare tier of human resource.

Their value is not only in results. It is in diagnostic capability. In motorsport, when a veteran driver says publicly that the car does not behave as the simulator predicted, that is a classic early indicator of correlation failure. A young driver may doubt himself. A race-winning driver is less likely to doubt himself and more likely to doubt the machine.

The original piece quotes not a single word from Valtteri Bottas or Sergio Perez about the car's behaviour. That is a material evidentiary gap. If such comments exist, they are the most important piece in distinguishing a design fault from an organisational one.

Personnel risk: brain drain at the satellites

One dimension no analysis of a multi-base model should skip: retention risk.

Aerodynamics engineers in Indianapolis and simulation engineers in South Carolina may see themselves as peripheral to the "real" team in Silverstone. Being separated from the main campus is a standard exit catalyst in engineering organisations. In parallel, recruiting senior European engineering talent to US sites is structurally harder than recruiting them to the UK motorsport cluster.

In other words, a three-base model does not only consume time. It consumes people, quietly and cumulatively.

Why this matters to people who do not watch racing

I write about football, not about racing. But I read this file with the same pair of eyes I once used to read a club's financial statements.

The argument structure here is familiar to the point of being uncomfortable. A sports team declines. A credible expert identifies a single organisational variable. The public absorbs that variable as the cause, because it is tidy, easy to grasp, and delivered by a voice with enough weight behind it.

You have seen it in football. The multi-club model is why this team declined. Splitting the sporting director and head coach roles is why recruitment failed. Putting the training centre forty kilometres from the stadium is why academy players do not develop.

All of it may be true. But almost nobody ever produces the test that proves it.

The lesson worth taking from the Cadillac F1 file is not the conclusion about Cadillac F1. It is the way of asking: when an expert says a structure is the cause, demand three things — measurable evidence, an evaluation date, and a voice from the party under scrutiny.

In this file, all three are missing. That is why I write this piece to raise questions, not to convict anyone.

What to track

Five signals will decide whether the structural argument holds or collapses.

One, the both-car finish rate per round, measured against Cadillac F1's own trajectory rather than against Mercedes. Sustained improvement falsifies the structural thesis. A continuing two-in-ten rate strengthens it.

Two, any change to the three-base model. If design, aerodynamics and simulation are consolidated into one site, that is retrospective evidence the team itself identified the problem.

Three, the appointment of a single technical authority with cross-site decision rights.

Four, a public statement of the primary objective: identity or championship.

Five, further commentary from Otmar Szafnauer. If it escalates from "not ideal" to a specific structural remedy, that indicates a systematic critique rather than a one-off remark.

Takeaway: memory does not vanish like money, memory haunts

I keep a notebook, and it does not record victories. It records dates, names, account numbers, and answers that do not match each other. In this file the notebook holds a single line: not enough data to conclude on cause, but enough data to conclude on the existence of a problem.

That is not a satisfying ending. It is a correct place to stop.

What I want to leave behind is not a verdict on three bases in Silverstone, Indianapolis and South Carolina. It is a question every large-scale sports project — clubs, federations, multi-club groups — should answer before building the second house: which objective is primary, and who has the right to adjudicate when two objectives pull in two directions?

If the answer is "both", the project does not have an answer. It has an unsigned contract, quietly in force, with nobody who will step forward to take responsibility when it falls apart.