New York to London in under 90 minutes. No jet lag, no contrails, no airport. Just a maglev capsule sliding through an evacuated steel tube resting on (or under) the seabed. Elon Musk dreamed of the Hyperloop on land — what happens when we push it across 5,500 km of saltwater?

Problem 1: The tube is fighting the ocean. The North Atlantic averages ~3,700 m deep. External pressure at that depth is roughly 37 MPa — about 370 atmospheres squeezing every square centimeter. For a thin-walled cylinder under external pressure, Timoshenko's buckling formula gives:

P_cr = 2E / (1 - ν²) × (t/r)³

For steel (E = 200 GPa, ν = 0.3), surviving 37 MPa requires t/r ≈ 0.045. A 10 m diameter tube needs ~22 cm of steel wall — and that's just to avoid buckling, before safety factors. Triple it for fatigue, corrosion, and pressure cycling, and you're at ~70 cm. Mass per meter: roughly 130 tonnes. Across 5,500 km: ~700 million tonnes of steel, or about 35% of global annual steel production. One year of dedicated output. Feasible? Yes. Cheap? Roughly $500 billion in raw steel alone.

Problem 2: Vacuum is hard. Vacuum across an ocean is absurd. Internal volume ≈ π(5)² × 5.5×10⁶ m ≈ 4.3×10⁸ m³ — comparable to Lake Tahoe. To hit 100 Pa (rough Hyperloop spec), you must pump out 99.9% of the air, then keep pumping forever to fight outgassing and microleaks. Real-world ultra-high-vacuum systems lose roughly 10⁻⁷ Pa·m³/s per square meter of wall. The tube has 1.7×10⁸ m² of inner surface, demanding continuous pumping power on the order of tens of MW just to hold vacuum.

Problem 3: Thermal expansion will tear it apart. Steel expands 12 ppm per °C. Seasonal sea temperature can swing 5°C even at moderate depths. That's 5,500 km × 12×10⁻⁶ × 5 = 330 m of length change. You'd need an expansion joint every few hundred meters — each one a potential vacuum leak. Invar alloy (1 ppm/°C) drops it to 27 m, but Invar costs 10× more than steel.

The reward. In 100 Pa air, drag on a streamlined capsule is ~10,000× lower than at sea level. With magnetic levitation eliminating wheel friction, top speed is limited mostly by passenger comfort (acceleration) and tube straightness. At 5,000 km/h (Mach 4 in air, but trivial in near-vacuum), New York to London takes 66 minutes. Energy per passenger? A 20-tonne capsule accelerated to 1,400 m/s carries ~20 GJ of kinetic energy. With regenerative braking recovering 70%, net energy per passenger (~50 people) is roughly 120 kWh — comparable to a transatlantic flight, but powered by whatever's on the grid.

The verdict. The physics works. The materials exist. The killers are leak management, seismic events along the Mid-Atlantic Ridge (the tube would cross an active spreading zone gaining 2.5 cm/year), and economics. Floating the tube at neutral buoyancy ~50 m below the surface — tethered to the seabed — slashes pressure to 0.5 MPa and the wall thickness problem largely vanishes. That's the version that might actually get built someday.