Crysis

The question was not whether the game was good. For most of a decade, buying a graphics card meant asking whether it could run Crysis, and the phrase became the standard unit of hardware ambition. Crytek built a game that no computer of its year could display at full settings.

2007: a game designed for machines that did not exist

Crytek had made Far Cry in 2004 and built a new engine for the follow-up. The design targeted hardware two or three generations ahead, on the assumption that processors would keep doubling in speed.

They stopped. Manufacturers moved to multiple cores instead of faster ones, and the game's single-threaded engine could not use them, so Crysis stayed unplayable at maximum settings for years after release.

What it displayed justified the demand: a jungle where every tree could be broken, foliage that moved correctly, and lighting that no competitor matched until well into the following decade.

A suit that gives the player four ways to solve everything

The nanosuit switches between strength, speed, armour and invisibility, and the game is built as open spaces with objectives rather than corridors.

That structure was unusual for a shooter of the period, which mostly moved players along a fixed path. Crysis expected improvisation and rewarded approaching a base from an unexpected direction.

The first game scored 91 on Metacritic. Crysis 2 moved to New York in 2011 and to consoles, which meant narrower spaces and a tighter engine. Crysis 3 followed in 2013 in an overgrown city, technically superb and commercially unremarkable.

The benchmark outlived the series

Crytek released a remaster in 2020, and the joke came back immediately: the remaster also ran badly, because it inherited the original's inability to spread work across cores.

A fourth game was announced in 2022 and has not appeared. Crytek's business moved to licensing its engine, which is where the technology had always been the strongest asset.

The series sold well enough, several million copies for the first game, but its reputation rests on a hardware question rather than on its own design.

What a technical demonstration costs

Building for future machines is a bet on how they will improve, and Crytek lost that bet on a specific point: speed per core stopped rising in the year the game shipped.

The engine itself succeeded where the games did not: CryEngine has been licensed to dozens of studios and is still in use. The lesson has been absorbed. Studios now target the hardware people own, and the games that push the most are built to scale down rather than to wait.

Crysis remains the clearest example of a work whose ambition was aimed at a future that arrived in a different shape.