Underwater (2020)

Underwater (2020) is a wet, claustrophobic creature feature set in 2050 at a deep-sea drilling station. The plot is simple: an earthquake ruins everything, the escape pods are gone, and a small crew walks across the ocean floor toward another facility. Kristen Stewart plays Norah Price, a mechanical engineer who solves problems, drags injured colleagues, and eventually makes the ultimate sacrifice. She is the clear protagonist and entirely capable. The supporting cast includes Black and Asian actors as crew members, and the film never pauses to comment on this. That naturalism is welcome, though it does not amount to a thesis. There is no LGBTQ content, no climate parable, no body-positive messaging, no neurodivergent characterization, and no revisionist history. The monsters are Lovecraftian, the water is cold, and the station is doomed.

The feminist component is present but modest. Norah is competent and self-sacrificing without a speech about patriarchy, which is the correct way to handle such things. The anti-corporate angle is similarly restrained. Tian Industries is negligent, profit-driven, and eager to keep drilling; the end credits headlines note its refusal to cooperate with investigations. This is a mild undercurrent, not a sermon. The film has almost no lecture energy. It wants to show a defective helmet imploding and a creature emerging from a corpse, not to instruct the audience on structural injustice. Race is never engaged as a theme, and the 2050 setting is not used to revise any historical record.

We are left with a survival horror exercise with a strong cast and a derivative story. Its social consciousness is ambient: a female lead doing engineering, a diverse crew treated as a crew, and a corporation behaving badly in the background. The film does not campaign. It kills people in the dark and moves on. For viewers seeking cultural instruction, Underwater offers almost none. For viewers seeking a giant monster and a countdown to a nuclear explosion, it delivers with reasonable efficiency. The final assessment is low on progressive signaling and moderately high on wet doom.