Cosmoteer: Starship Architect & Commander

Most spaceship-builders lie to you. They let you bolt a cannon to a hull, attach a reactor somewhere in the general vicinity, and call it engineering. Cosmoteer doesn’t. Here, the power your reactor generates is a physical object that a physical person picks up, carries down a physical hallway, and hands to your gun. If that hallway is three tiles too long, your gun goes quiet at the worst possible moment and you lose a fight you designed yourself into losing.

That single design decision — resources as logistics, not as a number in a UI panel — is what separates this from every other tile-based construction game on Steam. It’s also why the game has an almost cultish retention rate among people who like being told exactly how they were wrong.

The developer is Walternate Realities, which is to say one person: Walt Destler. Cosmoteer existed for years as a free, publicly available alpha before it hit Steam Early Access on October 24, 2022, and that long gestation shows in ways both good and bad. The simulation systems are unusually mature for an indie release. The surrounding game — the campaign, the progression, the reason to keep playing after you’ve solved the puzzle — is not.

The enthusiasm is easy to understand, and the game’s reputation among its audience is strong for something this unfriendly to newcomers. The more interesting question is how that enthusiasm holds up over time: a player base that adores the sandbox has reason to grow impatient with an Early Access roadmap that has been open for a very long time.

Market positioning is unusual. It’s not really competing with Space Engineers (3D, survival-flavored) or FTL (roguelike, abstracted). Its closest relatives are Factorio and Rimworld — games about designing a system and then watching the system betray you. Cosmoteer just points guns at the result. It’s Steam-only, Windows 10/11 64-bit only, with no Mac, no Linux, no Deck verification, and it sells as a paid title at $19.99.

• Tile-by-tile modular ship construction: Reactors, crew quarters, thrusters, corridors, ammo factories, shields and dozens of weapon types. Every part has a footprint, a mass, a power draw and a position that matters tactically.
• Individually simulated crew: Anywhere from half a dozen to over a thousand little humans who physically ferry power, ammunition and repair materials through your ship. They are your supply chain, your damage control, and your single biggest design constraint.
• Physics-driven modular combat: Target and blow off specific modules. Cripple engines. Breach a hull and watch the atmosphere — and the crew — react. Ships don’t have a hit-point bar; they have anatomy.
• Open-galaxy campaign, solo or online co-op: Procedurally generated systems, trading, bounties, salvage, fleet upgrades.
• Competitive PvP, online and LAN: Where your blueprint stops being a personal opinion and becomes a testable claim.
• Creative Mode + deep modding: Unlimited resources for pure design work, plus a modding framework and Steam Workshop pipeline that supports everything from single ships to total conversions.

The Gameplay Loop

The loop is design → deploy → fail → autopsy → redesign, and the autopsy is the best part.

You build a frigate. It looks great. It has beautiful symmetry, a proud spinal railgun, armor where armor belongs. Then it dies in ninety seconds — not because it was outgunned, but because the ammo factory was aft and the guns were forward, and the crew spent the entire engagement jogging. The failure is legible. You can watch it happen, dot by dot, and that legibility is the game’s core pleasure. Very few systems-heavy games make their own failure states this readable.

This produces a specific kind of player behavior: obsessive iteration. A design session meant to take forty minutes can sprawl into hours, because a second corridor running down the port side would cut ammo transit time by a third, and now you need to know if you’re right. You are frequently not right. Then you find out why.

Combat rewards the same analytical instinct. Because damage is modular, fights are less about DPS races and more about targeted dismemberment. Kill the thrusters and a superior ship becomes a stationary target. Punch through to the reactor and the whole thing goes dark. Breach the crew quarters and watch a thousand-strong workforce evaporate into vacuum. It’s genuinely tactical in a way that “real-time” usually isn’t, and the pause-and-order controls give you enough room to think without turning it into a turn-based game.

Where It Thins Out

The campaign is the weak link, and it’s not close.

The open galaxy gives you the classic sandbox verbs — trade, bounty, salvage, upgrade — and for the first stretch they’re compelling because you’re still resource-constrained enough that design choices hurt. The problem arrives the moment you land on a dominant design. Once you’ve built the ship that wins, the campaign stops being a series of problems and becomes a series of confirmations. You fly out, you win, you salvage, you fly out again. The economic layer isn’t deep enough to generate new constraints, and the procedural galaxy doesn’t produce encounters that meaningfully punish a good build. The best strategy games force you to abandon your favorite solution. Cosmoteer lets you marry it.

Multiplayer partly patches this. PvP is where a dominant design actually gets stress-tested, because a human opponent will find the seam in your logistics that the AI never noticed. Co-op campaigns are the strongest way to play — dividing a fleet between two architects with different philosophies creates the friction the solo campaign lacks. But leaning on multiplayer to fix a single-player pacing problem is a patch, not a fix.

Onboarding Friction

This is a hard game to start playing. The tutorial teaches you the controls; it does not teach you the thinking. There’s a substantial gap between knowing that crew carry ammunition and internalizing what that means for hallway topology, and the game largely expects you to close that gap through repeated humiliation.

The UI is the culprit as much as the complexity. It’s dense, information-rich, and clearly designed by someone who already understands the systems intimately. Part placement, resource routing overlays and crew assignment are all there, but discoverability is poor and the visual language for “why isn’t this working” is weaker than the underlying simulation deserves. For a game whose entire appeal is diagnosing your own mistakes, the diagnostic tooling should be its showpiece. It isn’t.

Expect a punishing first stretch. Expect the Workshop to be the real tutorial — downloading a well-regarded community ship and dissecting it teaches more in twenty minutes than the tutorial does in two.

The art direction is smart and unpretentious: clean top-down 2D, readable at a glance, with a functional-industrial look that prioritizes telling you what a module does over making it look cool. It works. When forty ships are exchanging fire and hulls are shearing apart, the battlefield should still be parseable — which is a real achievement, and one that a prettier, more detailed style would have wrecked.

Performance is the honest caveat. The engine is simulating every crew member as an individual agent, and that bill comes due. Large fleets and thousand-crew capital ships are exactly the scenarios where frame rate degradation is the predictable cost of the design’s central conceit. This isn’t sloppiness; it’s the direct price of the simulation. But it means the game’s most spectacular scenario — the enormous battle a long campaign builds toward — is also the one most likely to strain hardware.

Steam is the only storefront, and it’s the right one. The Workshop integration is doing enormous structural work here: ship sharing turns a solitary engineering exercise into a genuine design community, and the modding framework is deep enough to support total conversions rather than cosmetic tweaks. That ecosystem is a real chunk of the game’s long-term value.

Controls are mouse-and-keyboard, and unapologetically so. Construction is a precision drag-and-drop affair; combat wants rapid module-targeting clicks. There’s no controller pretense and no reason for one.

The Windows 10/11 64-bit exclusivity is the sore point. No native Mac or Linux build means no straightforward Steam Deck story for a game that — turn-based-adjacent pacing, top-down clarity, pause-heavy tactics — would be spectacular on a handheld. That’s the single largest addressable gap in the product.

Cosmoteer is one of the more intellectually satisfying games on Steam, and it earns that with a single stubborn idea executed without compromise: make the supply chain physical, and let the player live with the consequences. The result is a building game where design is a real skill rather than a taste, and where losing is informative instead of merely annoying. Walt Destler has built something structurally more ambitious than most studio-sized teams attempt.

The reservations are real and they’re all downstream of the same issue — the game around the simulation hasn’t caught up to it. The campaign runs out of pressure. The UI doesn’t explain a system that badly wants explaining. Performance strains exactly where the fantasy peaks. And an Early Access tag approaching its fourth anniversary is no longer a promise; it’s a status.

Buy it for the sandbox and the Workshop, both of which are worth the price on their own. Treat the campaign as a scaffold for experiments rather than a destination. And if you’re on a Mac, or hoping to play this on a Deck, keep waiting — that’s the one problem no amount of clever hallway routing will solve.