A space elevator needs a cable stronger than any material we can make. A space fountain, proposed by Robert Forward in the 1980s, sidesteps the strength problem entirely: instead of a static structure supported by tension, it's a dynamic tower held up by the momentum of a continuously circulating stream of pellets. Turn off the pellet stream, and it collapses like a sand castle.

Here's the trick. A ground-based accelerator shoots iron slugs straight up at high velocity. At the top of the tower — say 100 km up, at the Kármán line — magnetic deflectors reverse the pellets' direction. The reaction force from that deflection pushes up on the tower. The pellets then decelerate on the way down, get caught, re-accelerated, and launched again. The tower stands on a fountain of upward-flowing metal.

The numbers. Launch pellets at v₀ = 2,000 m/s. Gravity bleeds them down: v_top² = v₀² − 2gh = 4×10⁶ − 2(9.8)(10⁵) ≈ 2×10⁶, so they arrive at 100 km with v_top ≈ 1,414 m/s. Deflect them into a downward path and the momentum change per pellet is Δv ≈ 2,828 m/s.

Say we want to support a 5,000-tonne tower structure (plausibly light — mostly thin magnetic guideway, no compression load). Required upward force: F = mg ≈ 5×10⁷ N. The mass flow rate of pellets we need:

ṁ = F / Δv = 5×10⁷ / 2,828 ≈ 17,700 kg/s

Nearly 18 tonnes of iron per second in continuous flight. The pellet stream mass aloft at any instant, averaged over the trajectory, is roughly ṁ × 2h/v_avg ≈ 17,700 × 200,000/1,700 ≈ 2,000 tonnes. There's literally a small ship's worth of iron hovering above you.

The power bill. Kinetic energy flux at launch: P = ½ṁv₀² = 0.5 × 17,700 × 4×10⁶ ≈ 35 GW. That's the entire output of ~30 nuclear reactors just to keep the tower standing. With regenerative catchers recovering the returning pellets' KE, and ideally recovering the deflection energy magnetically at the top, losses might drop to 5–10% — call it 2–4 GW steady-state, roughly a Hoover Dam's worth. Not free, but tractable.

Why bother? Because now you have a launch platform sitting above 99.9997% of the atmosphere. Ride an elevator up the guideway, hop onto the pellet stream (or a linear motor running along its length), and get accelerated horizontally at the top station. A 5-g horizontal push over ~160 km gets you to orbital velocity (7.8 km/s) without needing a rocket. The tower doubles as the first stage of every launch.

The failure modes are spectacular. Lose power to the pellet accelerator and the tower deflates in about a minute — the last pellets already aloft finish their arcs, and then nothing is pushing up anymore. You need triple-redundant power, and the ground station must catch a decelerating rain of iron slugs safely. A single deflection misalignment at the top could send a 1-kg pellet at 1.4 km/s (roughly a tank shell) off in an arbitrary direction.

Materials-wise, everything is soft magnetics, superconducting coils, and vacuum tubes — all off-the-shelf. Unlike a space elevator, no unobtainium required. The engineering is power management and control loops, not tensile strength.