Honda, ADUO and the Price of Spec 2: Power Is Up, Driveability Is Back at the Start Line

Fernando Alonso reached for two words: “random downshifts”. Gear changes that arrive against the driver's intent, scattered across a lap, following no pattern a human can learn. In a modern racing car, that is the most corrosive kind of fault, because it does not break the car — it breaks the driver's trust in the car. The driver brakes half a metre early, feeds the throttle a beat late, selects a defensive gear for the next corner. Each precaution costs a few hundredths, and multiplied over fifty laps it becomes a race strangled from the inside.

What makes this worth dissecting rather than dismissing as radio grumbling is that the car had just been upgraded. Honda delivered its Spec 2 package to Aston Martin under the ADUO mechanism — a framework created precisely so that struggling power-unit manufacturers get extra development opportunity. The package hit its power target. It also dragged driveability back towards Spec 1, the version that tormented the team through the early part of the season.

One step forward and one step back, shipped in the same crate. That is why I spent three weeks logging every shift event of the AMR26 into a spreadsheet I built myself, rather than reading headlines about Honda's return. Read the results and you see a team climbing. Read the mechanism and you see a manufacturer paying for a loan of time it was never granted.

Context: a season in which the combustion engine became the protagonist again

2026 rewrote the entire power-unit rulebook. Electrical energy rose to roughly half of total output, the MGU-H disappeared, and 100% sustainable fuel replaced fossil petrol. Technically this is the largest change since the hybrid era began, and it inverted the development centre of gravity: the electrical side became a problem of control systems and software, while the hardest part — combustion of a new fuel, compression ratio, ignition strategy — returned as the frontier of performance.

Honda entered this cycle from an unusually weak starting position. They left their Red Bull partnership after 2026, then had to rebuild the entire development structure under HRC, rehire experienced F1 engineers, and return to the track roughly a year later than their rivals. HRC president Watanabe described the gap in a single line: they lost almost two years.

During that window, Mercedes, Ferrari, Audi and the Red Bull–Ford operation ran thousands of hours on the dyno. A two-year gap is not the gap of one upgrade. It is the gap of a generation of data.

Aston Martin is the recipient of that power unit, and this is where the story sharpens. The team has Adrian Newey designing the AMR26, Alonso in the cockpit, and a works partnership with Honda. On paper that is a formula for a step change. On track, the team scored 6 points across the last six rounds — a midfield scoring rate, no more.

One clarification matters, so that nobody mistakes ADUO for a favour. ADUO is the system that grants manufacturers classified by the FIA below a defined performance threshold additional development and upgrade opportunities. In other words, for Honda to use ADUO at all, the FIA had to formally acknowledge a measurable deficit. Being granted the upgrade is itself a technical indictment.

The core: change the combustion process, pay with driveability

Spec 2 was not a software polish. Honda changed the combustion process inside the internal combustion engine — fuel-air mixing, ignition, compression strategy — to lift peak output. That is the shortest path to extra horsepower when you are boxed in by regulation and by budget. It is also the shortest path to destabilising the chamber at part load.

The mechanism is concrete. Push combustion towards high power and you usually accept a narrower safety margin in the low and mid-load region — precisely the region a driver uses most across a lap: corner entry, part throttle, the handover between braking and acceleration. The chamber loses stability there, torque delivery becomes uneven, and the control system has to correct continuously. Those corrections surface externally as snatches, cuts, and gear changes nobody asked for.

Watanabe was blunt: they achieved the power target, but driveability regressed to something closer to Spec 1. That is a rare admission at leadership level in a sport where every statement is wrapped in the language of progress. It is also data. If the head of the racing division says so publicly, Alonso's complaint is not the subjective sensitivity of a demanding driver.

The second point matters more to me, and it is routinely skipped. Watanabe confirmed that battery and motor targets were broadly achieved, while development of the internal combustion engine fell behind. That inverts the familiar story of a Honda strong on combustion. Their historic strength is the area that is short; the newer discipline, electrification, is the one they got right.

The strategic consequence is severe. The electrical side is the easier area to update inside a homologation cycle, because it leans on software, energy maps and deployment strategy. The combustion side is far more locked: changing the chamber, compression ratio or burn strategy is a hardware change requiring dyno time, budget, and a valid upgrade allocation. Honda is short precisely where the fix is hardest.

I built my own tracking sheet per round: count of anomalous downshifts in braking zones, moments where the driver picked up throttle earlier than normal, cross-referenced against sector times. The sheet does not give me causality, but it gives me a pattern. The anomalies cluster in corners that require late braking followed by a slow apex, where the driver must move from brake to throttle in a very short window. That is exactly where blending engine braking with energy recovery is most sensitive.

Random downshifts: hardware failure, or software that has not finished learning?

Here I want to break from the crowd. The popular reading is “Honda's gearbox has a problem”. It is convenient, and it does not hold up well. A failing gearbox usually fails lawfully — same gear, same rev band, same load pattern. What Alonso describes is random, and randomness in a modern drivetrain almost always originates in the control layer.

In a hybrid architecture, downshifting is no longer the gearbox's decision. It is the output of a negotiation between three parties: the shift map, the engine-braking strategy, and the energy recovery system. When combustion destabilises at part load, actual torque no longer matches the torque the gearbox predicted. The controller must choose between holding the gear and dropping one, and for a few milliseconds it chooses wrong. One wrong choice goes unnoticed. Three in a single lap and Alonso notices immediately.

Read this way, the good news is that the problem is software-addressable: remapped calibrations, anti-stall logic, revised regeneration blending. Watanabe points in that direction, saying driveability can improve through hardware changes and control-system refinements. The bad news is that software can only fix what hardware permits. You cannot program your way around a chamber that is unstable in a load region you traverse hundreds of times per race.

This is why I classify “random downshifts” as an operational risk rather than a pure reliability risk. It will not stop the car at the side of the track. It will make the car slower in exactly the corners that decide lap time.

Six points in six rounds and the limits of reading results

A team scoring 6 points across six rounds is a team improving but still inside the midfield. I do not use that figure to conclude anything about absolute pace, because it depends too heavily on strategy luck, safety cars and how many cars finish. I use it to measure something else: the ability to convert potential into points, week after week, while the power unit remains unsettled.

Those six points came in races where the team did not need to push the car into its limit region. That is the signature of a team running below its potential to protect results. When a car does not tell the driver in advance what it will do, attacking strategies become far more expensive than usual. You cannot plan an undercut if your in-lap and out-lap are unstable. You cannot extend a stint if fuel and energy consumption fluctuate with every anomalous downshift.

Put differently, the cost of Spec 2 is not in the standings. It is in what the team does not dare to attempt.

Newey's car is wearing a mask

This is the part I think the forums misprice most. The AMR26 is an Adrian Newey design, and Newey has a record thick enough that any serious analyst should wait for data before concluding. We do not have that data. A car cannot be fairly benchmarked while its drivetrain keeps making decisions that contradict the driver's intent.

Picture aerodynamics as a map of intent. If the driver cannot place the car precisely at corner entry because torque delivery is uneven, every correlation between aerodynamic load and lap time is contaminated. The load curves an engineer plots in the evening will not match what happened on track in the afternoon. Nobody can say how much of the gap is chassis and how much is power unit.

That is why I use the phrase “wearing a mask”. Not to claim the car is good. To state that nobody has measured it yet.

The contrarian angle: Honda's real enemy is the calendar, not the engineering

The broadcast narrative is: Honda came back, they are a little short, and Newey is being wasted. That framing is easy to consume and wrong in its centre of gravity. Three variables decide Honda's season, and none of them live in the laboratory.

The first is timing. Returning a year later than rivals inside a regulation reset is a structural disadvantage, not one that effort can offset. When others already hold a year of correlation between dyno and track, you are still learning to read your own data.

The second is the cost cap. Watanabe was explicit: they pushed to catch up and ran into the budget cap and restrictions on operating time. This is often read as an excuse. It is not. It is a description of a mechanism: the rules were designed to prevent exactly the money-and-dyno-hours catch-up Honda says it was denied. What Honda needs is the one thing the regulations will not let them buy.

The third is people. After leaving Red Bull in 2026, Honda had to bring experienced F1 engineers back onboard. That is slow, and it explains the sequential development path — electric first, combustion second — rather than a parallel one. Under a staffing shortage that choice was rational resource allocation. It also front-loaded the slower-maturing technology and left the fastest-iterating area underdeveloped.

There is one more layer worth putting on the table: Watanabe's candour. A racing division president saying publicly that driveability regressed, that they lost almost two years, that they could not make up the ground — that is deliberate expectation management. Say it first to lower the reputational cost of the failures still to come. And possibly to strengthen the case for further upgrade allowances ahead of an FIA classification review.

As for “Newey is being wasted”, I file that under unproven hypothesis. It assumes a strong chassis, which the data does not yet support. A claim repeated often enough becomes an expectation, and an unmet expectation becomes pressure. With a two-time champion in the cockpit, that pressure will not stay quiet for long.

Data limitations

I should be explicit about what I do not have. No power figures in kW for Spec 1 and Spec 2. No lap-time deltas isolated to individual upgrades. No dyno numbers. Every conclusion above about the power-versus-driveability trade-off is qualitative, grounded in team leadership statements and observed car behaviour.

A reverse scenario also deserves a hearing: if the anomalous downshifts originate in sensors or the gearbox rather than the combustion chamber, my entire combustion trade-off argument loses its footing. I have not excluded that. I simply judge it less likely, given that the symptom appeared alongside a combustion-process change and alongside a reported regression in driveability.

What to watch

Three items go into my tracking sheet for the coming rounds. First, the count of anomalous downshifts in late-braking zones — if it falls while power holds, the control-layer hypothesis is confirmed and Honda has a far cheaper escape route than a hardware upgrade allocation. Second, straight-line speed and slow-corner exit speed — those two separate peak power from driveability better than any commentary. Third, how the FIA classifies Honda at the next assessment, because remaining upgrade allowances are the ceiling on their recovery.

Every tactical diagram begins with a shaky hand-drawn line on PowerPoint. I drew the AMR26's torque curve on a blank page, and it broke exactly where Alonso complains. Transition is not the run itself. It is the silence between two intentions that few people learn to read — here, the silence between the moment combustion stops pushing and the moment the electric motor takes over, where the car loses the continuity the driver needs.

What Honda lacks is not another upgrade. They lack time — the one thing the cost cap and operating-time limits will not let them borrow. And if Newey's car really is stronger than the standings suggest, the market is mispricing Aston Martin in the pessimistic direction. When the power unit settles, we will know who read that gap correctly from the start.