The Strait of Gibraltar is 14 km across at its narrowest — Tarifa to Point Cires. That sounds tractable until you check the bathymetry: the strait plunges to 900 m deep in the middle, with steep walls and a brutal two-layer current (Atlantic flowing in on top, denser Mediterranean flowing out below). The longest suspension bridge ever built, the Çanakkale 1915, spans 2,023 m between towers. Gibraltar needs roughly seven times that.

Can a suspension bridge even reach that far? The limiting physics is the cable's self-weight. A steel cable's specific strength gives a theoretical free-hanging span limit around 4–5 km before the cable snaps under its own weight. High-modulus carbon fiber pushes that to maybe 12 km. So a single 14 km span with conventional materials is impossible. You need intermediate supports.

And here's where the seafloor fights back. Building piers in 900 m of water is unprecedented — the deepest bridge pier today (Russky Bridge, Russia) sits in about 50 m. So engineers have proposed a suspended floating bridge: tension-leg pontoons anchored to the seabed, with the deck and suspension towers riding on top.

Back-of-envelope on a floating tower: Suppose we need three intermediate towers, dividing the 14 km into four ~3.5 km spans. Each tower must support deck load plus cable tension. A two-deck road/rail bridge masses about 200,000 kg per meter (deck + cables + traffic). One 3.5 km span loads each tower with roughly:

F = 200,000 kg/m × 3,500 m × 9.81 m/s² ≈ 6.9 × 10⁹ N

That's 700,000 tonnes of vertical load per tower. To float that, you need displacement — about 700,000 m³ of submerged hull. A pontoon 100 m × 100 m × 70 m draft does it. Now anchor that against currents.

The Gibraltar current runs ~1 m/s in the upper layer, locally to 3 m/s. Drag on a 100 × 70 m face:

F_drag = ½ × ρ × v² × C_d × A
       = 0.5 × 1025 × 9 × 1.0 × 7000
       ≈ 3.2 × 10⁷ N (32,000 tonnes lateral)

Tension-leg anchors handle that, but now consider seismic loading. The Azores–Gibraltar fault runs straight through the strait. The 1755 Lisbon earthquake (M 8.5–9.0) generated a tsunami that scoured these shores. A floating bridge actually rides out tsunamis better than a fixed one — the wave passes under as a long-period swell — but the seabed anchors must survive lateral acceleration plus liquefaction.

Worst problem: thermal expansion. A 14 km steel deck across a 40°C summer-to-winter swing expands by:

ΔL = α × L × ΔT = 12 × 10⁻⁶ × 14,000 × 40 ≈ 6.7 m

Almost seven meters of seasonal breathing. Expansion joints at each tower must absorb meter-scale motion while carrying tram rails and pipelines.

Cost estimate, extrapolating from Çanakkale (~$2.7 B for 2 km main span): a Gibraltar crossing lands somewhere between $25–60 billion, plus a century of marine maintenance in saltwater that eats steel at 0.1 mm/year. A subsea tunnel — actually studied seriously since the 1980s — is cheaper at ~$15 B but faces the same fault-line problem.