New Tesla Roadster in 2026: The Physics of SpaceX Thrusters and What Changed Since 2017
Hello HaWkers, almost nine years after the first prototype, Tesla finally showed the new Roadster. The presentation was streamed live on October 15, 2026, from SpaceX's test site in McGregor, Texas, and it had everything: a car driving upside down, a jump over other Teslas and a 0 to 60 mph (about 96 km/h) launch timed at 0.81 seconds. The topic blew up on Hacker News, with more than 1,600 points and 3,000 comments.
But, setting the show aside, what is really going on there? How does a street car accelerate harder than its tires can hold? And how much of what was promised in 2017 actually arrived? In this article I will break down the engineering, run the numbers with you in JavaScript and separate confirmed facts from what is still a promise.
What Was Announced on October 15
The new Roadster comes in three versions, and Tesla itself treats the two most extreme ones as "augments", meaning packages that add on top of the base car:
- Performance: two seats, active aerodynamics and a "multimode suction system", with 0 to 60 mph under 1.6 seconds. Available for reservation.
- Thrust: two seats, with cold gas thrusters mounted at the rear and 0 to 60 mph under 0.9 seconds. Limited release.
- Rocket: a single seat, thrusters spread around the entire perimeter, 0 to 60 mph under 0.59 seconds and a top speed above 230 mph (about 370 km/h). Invitation only and, according to Tesla itself, not approved for public roads.
The announced range is more than 300 miles, roughly 480 km. The car uses magnesium, titanium and carbon fiber, has a yoke with steer-by-wire steering, a center screen and two smaller screens for the driver. Tesla says the current design draws from the SR-71 and the F-35.
What was not announced matters too: there is no price for any of the versions. The production date landed on "2028, actually", in Elon Musk's own words. Reservations reopened with US$ 5,000 upfront and another US$ 45,000 within 10 days, adding up to the same US$ 50,000 from 2017. Some outlets reported higher reservation amounts for the Thrust version, but Tesla did not publish an official table.
And there was a line from Musk on stage that says a lot about the project: "This is not necessarily a safety-optimized car".
Why No Tire Can Hold 0.81 Seconds
Here comes the part I find most interesting. There is a physical limit to the acceleration of any car that relies only on its tires: the friction force. Simplified, the maximum acceleration is the friction coefficient (μ) multiplied by gravity. A sporty street tire sits in a μ range between 1.0 and 1.2, which gives something around 1 to 1.2 g.
Let's convert the announced times into average acceleration to see the size of the problem:
// Converts a 0 to 60 mph time into average acceleration (m/s² and in "g")
const MPH_TO_MS = 0.44704;
const G = 9.80665;
function averageAcceleration(timeSeconds, speedMph = 60) {
const finalSpeed = speedMph * MPH_TO_MS; // 60 mph = 26.82 m/s
const a = finalSpeed / timeSeconds; // a = Δv / Δt
return { ms2: a.toFixed(1), g: (a / G).toFixed(2) };
}
const times = {
'2017 promise': 1.9,
'Performance (2026)': 1.6,
'Thrust (2026)': 0.9,
'Live demo': 0.81,
'Rocket (2026)': 0.59,
};
for (const [version, t] of Object.entries(times)) {
const { ms2, g } = averageAcceleration(t);
console.log(`${version.padEnd(22)} ${t}s -> ${ms2} m/s² (${g} g)`);
}
// 2017 promise 1.9s -> 14.1 m/s² (1.44 g)
// Performance (2026) 1.6s -> 16.8 m/s² (1.71 g)
// Thrust (2026) 0.9s -> 29.8 m/s² (3.04 g)
// Live demo 0.81s -> 33.1 m/s² (3.38 g)
// Rocket (2026) 0.59s -> 45.5 m/s² (4.64 g)Notice the jump. Even the Performance version, with a 1.71 g average, already goes beyond what a street tire delivers on its own. The Thrust and Rocket versions go to 3 to almost 5 g, which is aircraft carrier catapult territory, not automobile territory.
That is exactly why the Roadster needs two tricks that do not depend on friction alone.
Trick 1: increase the normal force with suction
The friction force is μ times the normal force, which is how hard the car "presses" against the ground. If you can push the car against the asphalt without increasing its mass, the tire starts holding more. That is what the suction system does: fans create low pressure under the car and it sticks to the ground.
The idea is not new. The Brabham BT46B used a fan like that in Formula 1 in 1978, won the Swedish GP and was withdrawn shortly after. More recently, the McMurtry Spéirling, an electric car with a fan, broke the Goodwood hill climb record. Tesla claims the Roadster's fans can hold the car upside down while stationary, which, if true, means a suction force greater than the car's own weight.
Trick 2: push the car without using the ground
Cold gas thrusters do what no tire does: they generate force directly against the air, without relying on friction. Tesla says they produce four to five times the force the tires can transmit. That is why the Rocket version goes past 4 g.
How a Cold Gas Thruster Works
A cold gas thruster is the simplest form of rocket there is. There is no combustion. A gas stored under high pressure is released through a nozzle, and the reaction to the mass of gas leaving pushes the vehicle in the opposite direction. It is Newton's third law in its rawest form.
Satellites and spacecraft have used this kind of thruster for decades for small attitude adjustments, precisely because it is simple, reliable and has no flame. SpaceX uses nitrogen cold gas thrusters on the Falcon 9 stage, for example, to control orientation on the way back.
In the Roadster, according to Tesla, the gas is stored in carbon fiber wrapped pressure vessels, COPVs (composite overwrapped pressure vessels), the same family of tank used in rockets. Reports from the presentation mention around 600 liters of compressed air stored in carbon fiber spaces inside the structure. More gas means more "flight time".
The thrust of one of these thrusters follows a simple formula:
// Thrust of a cold gas thruster: F = ṁ * ve
// ṁ = gas mass flow rate (kg/s)
// ve = effective exhaust velocity (m/s)
function thrust(flowRateKgPerSecond, exhaustVelocity) {
return flowRateKgPerSecond * exhaustVelocity; // in newtons
}
// HYPOTHETICAL values just to illustrate the order of magnitude.
// Cold gas with air or nitrogen usually sits in the 400 to 700 m/s exhaust range.
const ve = 550; // m/s (assumption)
const carMass = 1800; // kg (assumption, Tesla did not disclose the weight)
// How much extra thrust would be needed to go from 1.2 g (tires only) to 3.38 g?
const G = 9.80665;
const extraG = 3.38 - 1.2;
const requiredForce = carMass * extraG * G; // F = m * a
const requiredFlowRate = requiredForce / ve;
console.log(`Extra force required: ${(requiredForce / 1000).toFixed(1)} kN`);
console.log(`Gas flow rate required: ${requiredFlowRate.toFixed(1)} kg/s`);
// Extra force required: 38.5 kN
// Gas flow rate required: 70.0 kg/sThe input numbers are my own assumptions, because Tesla disclosed neither the car's mass nor the tank pressure. But the order of magnitude already shows the challenge: to deliver a few seconds of strong thrust, you need to dump tens of kilograms of gas per second. With a 0.81 second launch, the math works. To "fly" for a long time, it does not.
How long does the gas last?
We can estimate the burn time from the tank volume and pressure. Since the real pressure was not disclosed, I will leave it as a parameter:
// Rough estimate of how much air fits in the tank and how long it lasts
// using the ideal gas law: m = (p * V) / (R * T)
const R_AIR = 287; // J/(kg·K), specific gas constant of air
const T = 293; // K, about 20 °C
function airMass(pressureBar, volumeLiters) {
const p = pressureBar * 1e5; // bar -> Pa
const V = volumeLiters / 1000; // L -> m³
return (p * V) / (R_AIR * T); // kg (approximation: real gas deviates at high pressure)
}
const volume = 600; // liters, figure mentioned in the presentation
const flowRate = 70; // kg/s, from the previous example
for (const pressure of [100, 200, 300]) {
const m = airMass(pressure, volume);
const duration = m / flowRate;
console.log(
`${pressure} bar -> ${m.toFixed(0)} kg of air -> ${duration.toFixed(1)} s of max thrust`
);
}
// 100 bar -> 71 kg of air -> 1.0 s of max thrust
// 200 bar -> 143 kg of air -> 2.0 s of max thrust
// 300 bar -> 214 kg of air -> 3.1 s of max thrustHere is the interesting part: at any reasonable pressure, the stored gas is enough for only a few seconds of maximum thrust. And the compressed air itself has weight, so a full tank means a heavier car. That explains why Tesla talks about "flight time" as a limited resource and why the Rocket version is invitation only and kept off the streets. It is not a car that flies, it is a car that makes short, very well controlled jumps.
Steer-by-Wire, Yoke and Optimus Actuators
Another big change is the steering. In steer-by-wire there is no mechanical link between the steering wheel and the wheels. The yoke reads the driver's intent, a computer decides the wheel angle and electric motors execute it. The Cybertruck already uses this setup, and the Roadster inherits the idea.
The biggest challenge of any "by-wire" system is redundancy: if a sensor lies, the car cannot turn on its own. The classic technique is to read the same value from several independent sensors and vote. A simplified example:
// Median voting among 3 steering angle sensors.
// If a sensor fails and sends garbage, the median discards the outlier.
function median(values) {
const sorted = [...values].sort((a, b) => a - b);
return sorted[Math.floor(sorted.length / 2)];
}
function safeAngle(readings, toleranceDegrees = 2) {
const m = median(readings);
// Counts how many sensors agree with the median within the tolerance
const agreeing = readings.filter(v => Math.abs(v - m) <= toleranceDegrees);
if (agreeing.length < 2) {
// No majority: enter degraded mode instead of trusting a single sensor
return { angle: null, state: 'FAIL_SAFE' };
}
return { angle: m, state: 'OK' };
}
console.log(safeAngle([12.1, 12.3, 12.0])); // { angle: 12.1, state: 'OK' }
console.log(safeAngle([12.1, 87.0, 12.2])); // { angle: 12.2, state: 'OK' } - sensor 2 ignored
console.log(safeAngle([5.0, 40.0, 87.0])); // { angle: null, state: 'FAIL_SAFE' }The real code in a car runs on dedicated controllers, with certification and many more layers. But the principle is this: never trust a single source. And this is where the line "not necessarily a safety-optimized car" weighs in, because redundancy and gas thrusters compressed to hundreds of bar share the same chassis.
Tesla also showed a "Track Mode", in which the car lowers itself and opens active wings driven by actuators from Optimus, the company's humanoid robot. The promise is downforce "approaching F1 levels". If you follow the blog, you have already seen how robotics is moving from the lab into real products; the case of the crab robot trained in simulation shows the same movement at a much smaller scale.
2017 Promise vs 2026 Delivery
This is the point that sparked the most discussion. The Roadster was unveiled in November 2017 with production promised for 2020. In 2026, the presentation was postponed several times: April, late April, August, October 1 and, finally, October 15. It is worth comparing side by side:
| Item | 2017 promise | 2026 announcement |
|---|---|---|
| Range | 620 miles (about 1,000 km) | More than 300 miles (about 480 km) |
| Top speed | Above 250 mph | Above 230 mph (Rocket version only) |
| Seats | 4 (2+2) | 2, or 1 in the Rocket |
| 0 to 60 mph | 1.9 s | Under 1.6 s (Performance) |
| Production | 2020 | "2028, actually" |
| Reservation | US$ 50,000 | US$ 50,000 |
The fair reading cuts both ways. On acceleration, the car delivered far more than promised. On range, seats and timeline, it delivered far less. And those who reserved in 2017 still have no firm delivery date.
The SpaceX Factor and the Merger Speculation
The presentation took place at a SpaceX facility, with rocket hardware in the car. That did not go unnoticed: part of the market read the event as yet another sign of the Musk companies moving closer together. Reports cited a rise of about 5% in Tesla shares in premarket trading the next day and rumors of a possible merger between Tesla and SpaceX, but no source confirmed any real negotiation. Treat it as speculation.
The context helps explain why the idea is going around. SpaceX has already absorbed xAI and is in the middle of a sequence of Starship tests; I covered that in the post about Starship flight 14 and the first orbital attempt. On Tesla's side, the EV market has become much more competitive, as shown by the turnaround in which BYD surpassed Tesla in electric car sales. A showcase car with rocket technology is also a marketing piece.
What Developers Can Take From This
It may look like a topic far from our daily work, but there are at least three very practical lessons here:
- Physics is the spec that does not accept pull requests. No software makes a tire go beyond 1.2 g. When the requirement breaks the physical limit, the way out is to change the architecture, as Tesla did by swapping friction for thrust. In systems, this applies to network latency, the speed of light and memory limits.
- Do the math before believing the slide. Twenty lines of JavaScript were enough to show that the stored gas lasts only a few seconds. Knowing how to make an order-of-magnitude estimate is a skill worth gold in architecture reviews and technical interviews.
- The deadline is part of the delivery. An amazing feature that is nine years late has a real cost for those who relied on it. Communicating reduced scope early is usually better than promising everything and delivering late.
Outlook for the Coming Years
The new Roadster is an impressive engineering demonstration and, at the same time, a product full of open questions: no price, no clear road approval for the most extreme versions and production scheduled for 2028. The combination of suction, cold gas thrusters and steer-by-wire shows how far you can push a car's physics when you mix the automotive and aerospace industries.
What is worth following from here on is the approval of the Thrust and Rocket versions, the final price, the real weight of the car with full tanks and whether the acceleration numbers hold up in independent tests. Until then, the numbers from the stage are a promise, and the Roadster's own track record recommends caution.
Let's go! 🦅
📚 Want to Keep Up With What Is Coming?
This article covered the engineering of the new Tesla Roadster and SpaceX's cold gas thrusters, but the ecosystem changes every week and not everything becomes an article here.
On X I share what I am testing, behind the scenes of my projects and the news that shows up before it turns into a post.
Follow Me There
💡 Daily content about development, career and the tools I actually use