If a motorcycle's tyres never touched the road, it would need something else to hold it up, push it, steer it and stop it. Holding it up on fans costs about 90 kW, a hundred times the 0.9 kW a tyred bike uses at 40 km/h. Holding it up magnetically over a special conducting lane costs only about 1.4 kW, which beats tyre rolling resistance above about 120 km/h. The catch: with no tyre friction, every newton of drive, braking and cornering force must also come from the track or from thrust.
Calculated in this post from momentum theory and the Inductrack efficiency of 2 N/W (Post and Ryutov, LLNL, 1996), with these example inputs: a 285 kg bike and rider, two 1.2 m rotors, a rotor figure of merit of 0.7, a rolling-resistance coefficient of 0.015 and, for the 0.9 kW at 40 km/h, an assumed aerodynamic drag area of 0.5 m².
- ~90 kW to hover a 285 kg bike and rider on two 1.2 m rotors, by momentum theory.
- 2 N per watt: the levitation efficiency of LLNL's Inductrack, so floating 285 kg costs about 1.4 kW.
- 120 km/h: above this, magnetic levitation loses less power than tyre rolling resistance.
- Earnshaw's theorem (1842) rules out stable floating on fixed magnets alone.
- 0.57 g of sideways force for a 60 km/h turn on a 50 m radius, which a tyre gives free.
Calculated in this post from a = v²/r, with these example inputs: a 60 km/h turn on a 50 m radius.
A tyre does four jobs at once. It carries the bike's weight, transmits drive, transmits braking and supplies cornering force, all through a contact patch about the size of your palm. Take the tyre off the road and you have to replace all four.
There are only two serious ways to float a vehicle: push air down, or push against a magnetic field. I work out what each costs for a bike, then ask how it would turn and stop.
How much power does it take to hover a bike on air?
About 90 kW for a 285 kg bike and rider, which is why hoverbikes carry car-sized engines. Momentum theory says a rotor makes lift by accelerating air downward, and the ideal power to hover is:
where \( T \) is thrust (equal to weight in a hover), \( \rho \) is air density and \( A \) is the total rotor disc area. Bigger discs move more air more slowly, which costs less power.
The worked solution
- Thrust. A 210 kg bike plus a 75 kg rider: \( T = 285 \times 9.81 = 2{,}796 \) N.
Calculated in this post from thrust equal to weight in a hover (momentum theory), with these example inputs: a 210 kg bike and a 75 kg rider.
- Disc area. Two rotors of 1.2 m diameter, about the widest that fits a bike's footprint (my assumption): \( A = 2 \times \pi \times 0.6^2 = 2.26 \) m².
- Ideal power. At sea level:\[ P_{ideal} = \frac{2{,}796^{1.5}}{\sqrt{2 \times 1.225 \times 2.26}} = \frac{147{,}850}{2.35} \approx 62.8\ \text{kW} \]
- Real power. Real rotors reach about 70% of ideal (a figure of merit of 0.7, my estimate):\[ P \approx \frac{62.8}{0.7} \approx 90\ \text{kW} \]
- Battery for a 30-minute hover. 45 kWh, about 280 kg at 160 Wh/kg (estimate), which is heavier than the bike itself. So the battery needs more lift, which needs more battery.
Calculated in this post from the 90 kW hover power above, a half-hour hover (an example input) and 160 Wh/kg, the low end of the 160 to 300 Wh/kg range given for lithium-ion cells (Wikipedia).
Compare a tyred bike at 40 km/h: drag plus rolling resistance come to about 0.9 kW. Hovering costs roughly a hundred times more power than rolling, before the bike has moved a metre.
Calculated in this post as 466 W of rolling resistance (0.015 × 285 kg × 9.81 × 11.1 m/s) plus about 420 W of air drag (½ × 1.225 kg/m³ sea-level air density, NASA × 0.5 m² × 11.1³ m/s), with these example inputs. The 90 kW is from the worked solution above.
On 26 October 2021, at the Fuji Speedway in Japan, the start-up A.L.I. Technologies opened orders for the XTURISMO Limited Edition hoverbike. It was 3.7 m long, weighed 300 kg, carried one rider of up to 100 kg and flew for up to 40 minutes at about 100 km/h. To do it, it carried a 170 kW combustion engine driving two big lifting propellers, plus four smaller electric stabilising ones. The price was 77.7 million yen, about US$680,000, for a run of 200. By January 2024 the company had filed for bankruptcy. The physics above explains the engine: lifting 400 kg on rotors that size takes well over 100 kW.
New Atlas: XTURISMO Limited Edition hoverbike
Why can't a bike just float on magnets?
Because of Earnshaw's theorem. In 1842 Samuel Earnshaw proved that a set of fixed charges, or by extension fixed permanent magnets, cannot hold an object in a stable, stationary float. You can balance a magnet above another, but there is always some direction in which a tiny nudge makes it fly off or snap down.
Every working maglev uses one of the known loopholes:
- Active control: electromagnets switched thousands of times a second by sensors, as in the Shanghai Transrapid.
- Diamagnetism and superconductors: materials that only repel. A superconductor cooled in liquid nitrogen floats a magnet stably.
- Motion: a magnet moving over a conductor induces currents that push back. This is the Inductrack principle, and it only works while moving.

How cheap could magnetic levitation be? The worked solution
Surprisingly cheap, if the road carries a conductor. In the Inductrack, invented by Richard Post and Dmitri Ryutov at Lawrence Livermore National Laboratory, the vehicle carries a Halbach array of permanent magnets that concentrates the field downward. Moving over a track of shorted copper circuits, it induces currents that repel it.
Their 1996 analysis gives a lift-to-drag ratio that grows with speed:
Here \( L \) and \( R \) are the inductance and resistance of each track circuit and \( \lambda \) is the magnet array's wavelength. The power lost to magnetic drag is \( P = F_{drag}\,v \), so the "levitation efficiency" is:
Speed cancels out. Post and Ryutov targeted K = 2 N/W, giving a lift-to-drag ratio of about 280 at 500 km/h and a transition speed of only 1.8 km/h.
- Levitation power for a bike. With \( K = 2 \) N/W:\[ P_{lev} = \frac{m g}{K} = \frac{2{,}796}{2} \approx 1.4\ \text{kW, at any speed above walking pace} \]
- Tyre rolling resistance power. With a coefficient of 0.015 (my estimate):\[ P_{roll} = C_{rr}\, m g\, v = 41.9\, v \ \text{W} \]That is 466 W at 40 km/h and 1,165 W at 100 km/h.
- Crossover speed. Set them equal:\[ v = \frac{1{,}398}{41.9} = 33.4\ \text{m/s} \approx 120\ \text{km/h} \]Above 120 km/h, a maglev bike loses less to levitation than a tyred bike loses to rolling.
Calculated in this post from the 1.4 kW levitation power (Post and Ryutov, LLNL, 1996, K = 2 N/W) and a rolling power of 41.9 v watts (a coefficient of 0.015, my estimate).
- Magnet size. The LLNL example levitated about 7,000 kg per m² of array at zero gap, with an array mass of about 900 kg/m². For 285 kg: at least 0.04 m² of magnets weighing about 37 kg, more in practice because force falls off with the gap.
Calculated in this post from Post and Ryutov of about 7,000 kg of lift per m² of array and 900 kg of array mass per m²: 285 ÷ 7,000 = 0.04 m², and 0.04 × 900 = 37 kg.
See the data as a table
| Tyre rolling resistance (Crr 0.015) (W) | Inductrack levitation (K = 2 N/W) (W) | |
|---|---|---|
| 20 km/h | 233 | 1398 |
| 40 km/h | 466 | 1398 |
| 60 km/h | 698 | 1398 |
| 80 km/h | 931 | 1398 |
| 100 km/h | 1164 | 1398 |
| 120 km/h | 1397 | 1398 |
| 150 km/h | 1746 | 1398 |
| 200 km/h | 2328 | 1398 |
Source: Levitation efficiency from Post and Ryutov, LLNL (1996); rolling resistance coefficient is my estimate · Aerodynamic drag is the same for both and is left out. Rotor hover would be about 90,000 W, off this scale.

How would a floating bike turn and stop?
Not by itself. Cornering at 60 km/h (16.7 m/s) on a 50 m radius needs a sideways acceleration of:
Calculated in this post from a = v²/r and F = m a, with these example inputs: 60 km/h, a 50 m radius and 285 kg.
A tyre supplies that from friction. A floating bike must get it from somewhere else:
| Job | Tyred bike | Fan hoverbike | Maglev lane bike |
|---|---|---|---|
| Support weight | Tyre, ~0.5–1.2 kW rolling loss | ~90 kW | ~1.4 kW |
| Drive forward | Rear tyre friction | Tilt the thrust | Linear motor in the track |
| Brake | Disc and tyre friction | Reverse thrust, slow | Linear motor regeneration, eddy currents |
| Corner | Lean plus tyre friction | Bank and tilt the thrust | Magnetic guidance from the track |
| Where it works | Any road | Anywhere, briefly | Only on the lane |
The maglev bike becomes a train car you sit astride: the lane levitates, drives, steers and brakes it. That also answers how it leans. On a flat guideway it can't use countersteering the way a tyred bike does (see countersteering physics), so the lane would need banked curves.
This is not only theory. In September 2022, engineers from Southwest Jiaotong University in China ran cars fitted with permanent-magnet arrays along about 7.9 km of road with a conductor rail laid in it, according to Motor1. The cars floated about 35 mm above the surface, and one test reached about 230 km/h. Without the rail, the magnets underneath do nothing at all.
On an ordinary road, a bike whose tyres never touch is possible only as a fan hoverbike, at about 90 kW, a hundred times the power of rolling at city speed and impractical on batteries. On a dedicated maglev lane with conducting circuits, it is not only possible but efficient: levitation costs about 1.4 kW at K = 2 N/W, less than tyre rolling resistance above 120 km/h. The law that limits it is Earnshaw's theorem, which rules out passive floating on fixed magnets, plus Newton's second law: with no tyre friction, every turning and braking force must come from the track.
How sensitive is the answer to the assumptions?
- Rotor size. Hover power falls as \( 1/\sqrt{A} \). Doubling the disc area to 4.5 m² cuts the 90 kW to about 64 kW, but makes a vehicle about 3.4 m wide.
Calculated in this post from the 1/√A scaling in momentum theory, with these example inputs: a doubled disc area of 4.5 m², as two rotors each about 1.7 m across.
- Altitude. Hover power rises as \( 1/\sqrt{\rho} \). At Bengaluru's 920 m on a 30°C day, NASA's standard-atmosphere fit gives about 1.04 kg/m³, so hover power rises about 8%.
Calculated in this post from the NASA density formulas with a 30°C day (example input) at the 920 m elevation of Bengaluru (Wikipedia): 1.04 kg/m³ against 1.225 kg/m³, and √(1.225 ÷ 1.04) ≈ 1.08.
- Track quality. Halve K to 1 N/W and levitation costs 2.8 kW; the crossover with tyres moves to 240 km/h.
Calculated in this post from the same two formulas, with K = 1 N/W as an example input: 2,796 N ÷ 1 = 2.8 kW, and 2,796 ÷ 41.9 gives 240 km/h (Post and Ryutov).
- Tyre choice. A low-rolling-resistance tyre at 0.010 moves the crossover up to 180 km/h. Tyres are very good at their job.
Calculated in this post from a rolling power of 0.010 × 2,796 N × v against the 1.4 kW of levitation (Post and Ryutov), with 0.010 as an example input.
- Aerodynamics. Identical for both, and above 100 km/h it dwarfs both levitation and rolling losses.
Where the model breaks: Inductrack figures come from train-scale designs. A bike-sized array has more edge losses, and a two-wheeler's narrow footprint makes roll stability on a magnetic lane a separate unsolved problem.
Use this to check what you just read. Change the mass, speed, rotor area or track efficiency and compare all three ways of holding a bike up.
Tyres vs maglev vs rotors: power to hold your bike up

A tyre is a levitation system, a motor mount, a brake and a steering rack, all in a patch the size of a palm.The working above
What does the research say?
A lumped-circuit analysis of passive maglev with Halbach arrays over shorted track circuits. Lift-to-drag equals \( \omega L/R \), rising linearly with speed; a design target of 2 N/W gives a 1.8 km/h transition speed and a lift-to-drag ratio near 280 at 500 km/h. For a 50-tonne car the levitation drag is about 250 kW, roughly 3% of the 8.3 MW needed to beat air drag at 500 km/h. Limit: analytical, at train scale.
Read the report (free)The follow-up reports levitation forces of about 40 tonnes per square metre of Halbach array with NdFeB magnets weighing a few percent of the load, a transition at walking pace, and fail-safe settling onto auxiliary wheels if the drive fails. A small model on a 20 m track ran at about 12 m/s, and a NASA-sponsored model aimed at 10 g acceleration was being built. Limit: model-scale tests.
Read the record (free)What's new
- A Chinese superconducting maglev test vehicle reaches 700 km/h in two seconds on a 400 m track China Daily, Dec 2025
- Volonaut's jet-powered Airbike flies, skipping rotors altogether designboom, May 2025
For how a tyre actually makes grip, see tyre size markings and traction control and cornering ABS. The Yantra page collects the rest of the engineering on this site.
What if next?
- What if you built a motorcycle that runs in space? No road, no air, so neither tyres nor rotors work.
- What would riding on the Moon feel like? One-sixth gravity means one-sixth the tyre grip for the same rubber.
- What if a petrol engine ran on Mars? Thin Martian air makes both breathing and rotor hover far harder.
Questions people ask
Can a car or bike levitate on a normal road?
Not with magnets. Magnetic levitation needs a conducting or magnetic track; Southwest Jiaotong University's 2022 floating cars ran over a conductor rail laid in the road. On ordinary tarmac, only fans or jets can lift a vehicle.
Why don't hoverbikes replace motorcycles?
Power. Lifting a 285 kg bike and rider on two 1.2 m rotors needs about 90 kW, roughly a hundred times what a tyred bike needs at 40 km/h. The XTURISMO needed a 170 kW engine and cost about US$680,000.
Is maglev more efficient than wheels?
At high speed, it can be. With an Inductrack efficiency of 2 N/W, levitation costs less than tyre rolling resistance above about 120 km/h for a bike-sized load. At city speeds, tyres win comfortably.
What is Earnshaw's theorem?
It shows that fixed permanent magnets alone cannot hold an object in a stable float; there is always an unstable direction. Working maglevs get around it with active electromagnets, superconductors or motion-induced currents.
Sources
- Wikipedia: Momentum theory (source for the ideal hover power formula)
- Wikipedia: Earnshaw's theorem (source for the 1842 theorem and its loopholes)
- Post and Ryutov: The Inductrack concept, LLNL UCRL-ID-124115 (1996) (source for K = 2 N/W, L/D 280, 1.8 km/h, 7,000 and 900 kg/m²)
- Post and Ryutov: The Inductrack approach to magnetic levitation, OSTI (2000) (source for 40 t/m² and the model tests)
- NASA Glenn Research Center: Earth atmosphere model, metric units (source for air density at Bengaluru)
- New Atlas: XTURISMO Limited Edition hoverbike (2021) (source for 300 kg, 170 kW, 40 minutes, ¥77.7 million)
- eVTOL News: A.L.I. Technologies XTURISMO LE (source for the six propellers and the January 2024 bankruptcy)
- Motor1: China tests magnetic levitation car (2022) (source for 7.9 km, 35 mm, 230 km/h)
- Guinness World Records: Fastest maglev train (source for 603 km/h on 21 April 2015)
- China Daily: Record set in high-speed maglev tech (2025)
- designboom: Volonaut Airbike (2025)
- Science Forums: Plasma cushion levitation cars (reader question)

