Valve’s Linux push can look almost inevitable in hindsight. Proton lets SteamOS run Windows games, the Steam Deck turned SteamOS into something people actually buy on purpose, and Valve followed it with a new Steam Machine for 2026. Linux gaming still has plenty of rough edges, but we’ve come a long way from the days when installing Linux meant accepting that most commercial PC games simply weren’t coming with you.
More than a decade before Valve brought Steam to Linux, a small company in Tustin, California, tried to change that in the expensive way. Loki Software licensed Windows games from major publishers, ported them individually, boxed them up for Linux users, and hoped there were enough customers to keep the whole thing afloat. The company lasted barely three years before going bankrupt with less than $50,000 in assets and somewhere between $1 million and $10 million in debt.
Plenty of tech companies disappeared around the turn of the millennium, so bankruptcy alone isn’t what makes Loki’s story interesting to me. What caught my attention was the problem it burned through all that money trying to solve. Linux had users who wanted commercial PC games, but each major Windows release needed someone to justify building a separate Linux version. Valve would spend years wrestling with the same basic problem.
A lawyer decided Linux needed real PC games
Somebody had to make the terrible business decision
Loki Software was founded in 1998 by Scott Draeker, whose previous career makes this story slightly stranger. He was a lawyer working on software licensing deals when he encountered Linux and became convinced there was a commercial gaming market waiting to be built around it.
By the end of 1998, Loki had secured the rights to port Activision’s upcoming strategy game Civilization: Call to Power. The Linux version reached stores in spring 1999 as the company’s first major release. Draeker later recalled that Loki expected it might lose money on Call to Power, but he saw the project as a way to prove that a mainstream PC game could actually be sold commercially for Linux.
That gives you a pretty good idea of what Loki was signing up for from the beginning. It was willing to take a loss on its first major product because the market it needed barely existed yet.
Linux already had games at the time. id Software had released Linux versions of titles such as Quake, and there was an active community around free and open-source projects. Commercial releases were much harder to come by. Most major PC games stayed on Windows, publishers had little reason to spend money on Linux versions for a relatively small audience, and retailers had even less incentive to give those versions shelf space.
Loki’s plan was easy enough to understand. It would license games people already wanted, handle the Linux port itself, and sell a separate boxed release. Looking at that model now, I can see why somebody might’ve convinced themselves it could work. Linux had a passionate user base, PC gaming was growing quickly, and Loki only needed a small slice of the wider market to make each port worthwhile.
The problem was how much work lay behind each of those releases.
Loki briefly made commercial Linux gaming look viable
Quake, SimCity, and a dangerous amount of optimism
Over the next couple of years, Loki released around twenty commercial games and expansions for Linux. Its catalog included Quake III Arena, Heroes of Might and Magic III, SimCity 3000 Unlimited, Tribes 2, and Sid Meier’s Alpha Centauri.
These were major PC games that Linux users already knew and wanted to play. For a relatively short period, Loki gave them access to professionally packaged Linux versions of titles that would’ve looked completely ordinary on a Windows gaming shelf. The porting work also exposed a bunch of problems that went beyond the games themselves. Linux distributions varied; installation could be messy; hardware support wasn’t always predictable, and cross-platform development tools were still fairly immature.
Loki ended up building some of the infrastructure it needed along the way. The company released Loki Setup, an installer framework that other Linux software could reuse. Its lead programmer, Sam Lantinga, was also the creator of Simple DirectMedia Layer, better known as SDL, and Loki’s commercial ports gave the library a demanding real-world environment in which to develop. Loki also helped revive and develop the OpenAL audio project, which continued after the company disappeared.
I find that part of the story especially interesting because the tools aged better than the products Loki was actually selling. Old boxed Linux games eventually became tied to aging libraries, distributions, and hardware. SDL kept growing into one of the most widely used cross-platform development libraries in gaming, while OpenAL also outlived the company that helped revive it.
Some of Loki’s developers followed a similar path. Ryan C. Gordon continued working on Linux and cross-platform game ports for years after Loki closed. Lantinga would eventually provide an even more direct connection to Valve.
By then, though, Loki had already run out of road.
The market never became big enough quickly enough
The games arrived faster than the customers
Loki filed for Chapter 11 bankruptcy protection on August 3, 2001. The numbers are brutal: less than $50,000 in assets against somewhere between $1 million and $10 million in liabilities.
Once you look at the business model, it’s easy to see how those costs piled up. Each release could involve another publisher agreement, another porting effort, another testing cycle, another boxed product, and another gamble that enough Linux users would buy it to cover all of that work. Loki had to repeat that process every time it wanted another major game in its catalog.
Chapter 11 gave the company an opportunity to reorganize, but Loki eventually moved toward Chapter 7 liquidation and shut down in January 2002. Its online store stayed open until noon on the final day, so the company could process outstanding orders. Loki spent three years trying to create a commercial Linux gaming market, and right up to the end it was still shipping the boxed Linux games on which the whole experiment depended.
There were attempts to keep parts of the catalog alive. Michael Simms, who founded Linux game retailer Tux Games in 2000, created Linux Game Publishing in 2001 as Loki’s financial problems became obvious. The new publisher tried to continue selling and supporting commercial Linux games and hoped to pick up titles from Loki’s catalog, but ultimately secured only MindRover: The Europa Project.
Much of Loki’s commercial catalog lost the company that had ported and supported it. Some of its open-source tools and code continued to circulate among former employees and community mirrors, which fits the pattern of Loki’s legacy quite well. The retail operation disappeared quickly, and the technical work proved much harder to kill.
Valve eventually ran into the same wall
Steam Machines learned an old Linux lesson
The connection between Loki and Valve gets unusually literal here. Sam Lantinga joined Valve in 2012, right around the time the company was preparing its serious move into Linux. Steam arrived on Linux in 2013, followed by SteamOS and the original Steam Machines, which launched in 2015.
Valve had advantages Loki could only have dreamed about. It owned Steam, had millions of customers, published major games itself, and had enough money to keep treating Linux as a long-term platform project without needing every experiment to pay off immediately. Even with all of that, Valve still had to deal with the same catalog problem Loki had faced years earlier. Linux versions needed developers and publishers willing to make them.
Valve ported its own games and encouraged other studios to do the same. Steam eventually built up a respectable native Linux catalog, but Windows still had years of games behind it. Anyone considering SteamOS was looking at a platform that couldn’t run a large chunk of the PC library they already owned. That dependence on individual Linux ports is exactly the bottleneck Proton would later attack, to the point that native Linux game builds have become much less necessary.
The 2015 Steam Machines arrived with that problem still hanging over them. Hardware came from several manufacturers at wildly different prices, SteamOS was still young, and the native Linux library couldn’t match what the same hardware could play on Windows. By 2018, Valve had removed Steam Machines from the main Steam Store navigation due to low user traffic, although it continued to work on Linux behind the scenes.
Valve had more money, more reach, and far more influence over PC gaming than Loki ever did, yet a native-port strategy still left it dependent on convincing developers to support Linux game by game.
Then Proton changed the math
Valve found a much cheaper way around the problem
Valve introduced Proton in 2018, using Wine and other open-source components to run many Windows games directly on Linux through Steam. The broader ecosystem now gives Linux users several ways to run Windows games through Wine, Proton, and related compatibility tools. Suddenly, Linux didn’t need a dedicated native version of every game before there was any chance of getting it running.
Loki had spent its short life tackling the catalog gap one port at a time. Proton gave Valve access to years of existing Windows games that could potentially work on Linux without going through that entire process first. Compatibility still varies, and anti-cheat remains one of Proton’s biggest Linux gaming problems. Launchers, DRM, graphics bugs, and other problems can still prevent individual games from working properly. Even so, the scale of the opportunity changed completely.
Valve’s Linux-powered Steam Deck in 2022 finally gave that approach the hardware it needed. Valve controlled the operating system, compatibility layer, storefront, and device, while customers could buy the handheld because they wanted a portable gaming PC. They didn’t have to care much about Linux being underneath it.
That integration depended on far more than Proton alone. The Linux and open-source technologies behind the Steam Deck include Arch Linux, Vulkan, Mesa, Proton and Wine, KDE Plasma, and Flatpak, giving Valve an entire software stack it could shape around the hardware.
I think that’s one of the smartest parts of the Steam Deck’s success. Valve finally stopped needing Linux itself to be the sales pitch.
Loki solved the problem the hard way
Loki’s failure looks fairly predictable from 2026. Building separate Linux versions of Windows games for a small desktop audience was an expensive way to create a market, and its bankruptcy numbers eventually showed how unforgiving that approach could become. Notwithstanding, Loki had identified the problem correctly — Linux was always going to struggle as a gaming platform, and every major release needed someone to justify a separate Linux development project.
Loki spent its short life working through that problem one title at a time. Its developers also helped build tools that survived the company’s demise, and one of its most important programmers eventually joined Valve just as Valve began its own Linux push.
Valve had the money and time to keep experimenting after the original Steam Machine strategy fizzled. Proton eventually gave it a way around the bottleneck Loki never escaped.