Over the past few years Steam has become a much tougher test environment, and indie developers are getting bruised for it. It was always meant to be a democratizing showcase, but in many ways the platform has completely collapsed, with thousands of games launching each month into near-total obscurity. Now, thanks to generative artificial intelligence, it's even harder to cut through Steam's deluge, and the indie games that once defined the platform are becoming its latest victims.
For context, according to SteamDB, more than 20,000 games launched on Steam in 2025, and roughly one in five new releases now comes with AI-generated content published on their store page. Because the same tools that allow an indie developer to whip up dialogue, character art, and code overnight also make it trivially easy to churn out forgettable filler, the two curves converge. By the looks of it, there's a waterline that's rising, and while great indie games aren't going anywhere, there's a serious risk that worthwhile titles will become dramatically harder to find.

Almost half of the top releases on Steam in 2025 came from AA or Indie Studios
Indie and AA games look set to dominate the Steam charts this year as they are already struggling with big budget AAA titles.
A pre-existing Steam discoverability issue
To understand why AI is such a threat in this position, we need to start with the context of the platform itself. Introduced with Steam Direct in 2017, Valve's open publishing model replaced the gatekeeping of the old Greenlight era with a $100 fee and an open door. This democratization worked exactly as intended – especially as game creation became more accessible – and as such, releases more than doubled in five years, from 9,654 in 2020 to over 20,000 in 2025.
For consumers at least, more games on Steam is in many ways a good thing, but in practice it also creates an ecosystem where most titles disappear upon arrival. Steam's discovery algorithm rewards momentum, so games that launch without a pre-built wishlist or community tend to drown before anyone even notices they exist. The numbers provided by SteamDB and Gamalytic paint a brutal picture of how crooked the platform has become:
-
According to PC Guide, nearly half of the 2025 releases—roughly 9,370 out of more than 20,000 games—received fewer than 10 user reviews, and around 2,200 received none at all.
-
More than 40% of games in 2025 failed to break the $1,000 threshold that refunds the $100 submission fee on Steam, and some independent estimates put the failure rate as high as 66%.
-
Steam now averages around 350 new releases each week, which would allow players to sift through more than 50 games per day.
What's more, according to an analysis by Games-Stats, the average paid game launched in early 2026 earned roughly $350 over its lifetime. The actual revenue stream is wildly high, as only three games at the start of 2026 (Resident Evil Requiem, Crimson desertand Slay the Spire 2) received almost 43% of all revenue from the game on Steam. All things considered, it's clear that AI didn't create this particular flood, but as players and developers look forward, it's also clear that generative AI will play a big role in the flood yet to come.
AI lowers the last left barrier
While it's true that game development has become more accessible over time, it still requires real skill at its best – someone has to work with the engine, animate, design and write the code. But as the estimated 700 percent year-on-year increase in AI usage in 2025 already shows (originally reported by Ichiro Lambe for Totally Human Media), generative AI is actively breaking down each of these barriers at once, allowing a single person to create content from a text prompt. What used to take a small team and months of work can now be approached by one person in a weekend, and overall the average consumer on Steam has proven to be approachable.
GameRant Quiz
Easy (15s) Medium (10s) Hard (5s)
That further increase in accessibility is the story here, as the multidisciplinary frictions that once limited how many games one could make have largely evaporated. The same report says that visual asset generation alone accounts for about 60% of Steam's AI exposure, covering characters, backgrounds, and art that previously required an immense amount of skill development or an artist on the payroll. As the most expensive and time-consuming parts of actual game development become subscriptions, it's clear that the volume of what's being produced is growing accordingly.
Ultimately, the scope of Steam's AI data makes this level of acceleration impossible to ignore: roughly 8,000 Steam games flagged AI-generated content in the first half of 2025 alone, compared to around 1,000 in all of 2024—an eightfold jump in a single year. As noted earlier, by the end of 2025, approximately one in five new releases contained an AI disclosure; this number has risen to 25% in some months and is almost certainly an underestimate as disclosure is self-reported and loosely enforced. SteamDB will now automatically flag these titles for players to filter or filter out, which tells you all about how routine AI content has become.
Hidden gems will be harder to find
Of course, AI won't replace the indie games that made Steam worthwhile anytime soon. 2025 proved that a great idea can still explode –TOP moved more than 15 million copies at eight dollars apiece Schedule I and REPO turned sketchy concepts into a nine-figure phenomenon. Quality still finds its audience, and success stories haven't slowed down at all.
Hidden gems will always exist; it just might take longer to dig through the sediment to get to them.
The real danger is dilution rather than replacement, as AI-assisted asset flipping and rapidly generated fundraising will add another layer of noise between players and games that are actually worth their time. Steam's tax on a saturated store page was previously only paid by human effort, which naturally limited how quickly the catalog could grow. But as AI removes that cap, the trajectory is pointed somewhere inconvenient, even though most apocalyptic predictions have yet to come true.
Not quite the Apocalypse Some predicted
At the end of the day, the most hysterical version of this story is worth enduring. The more than 40,000 AI games per year that some analysts predicted during the opening shots of AI allegations never materialized; 2025's growth rate has actually slowed from previous years, and Steam's own frictions—the ten-review threshold, the wishlist-driven discovery queue—still make pure spam a losing strategy. Realistically, most people recognize AI as a tool with legitimate use cases, and as people continue to adapt to AI in their daily lives (whether we like it or not), developers can really tap into it in a smart way to make games they otherwise couldn't afford to make amid increasing difficulty.
However, that doesn't mean that consumers shouldn't be aware of the problems that indie games are facing on Steam, because it can only get worse before it gets better. A tool that allows one person to do the work of ten doesn't need to create 40,000 games to bury the catalog; they just have to push the line higher while the algorithm rewards anything that already has momentum. So as time marches on and the use of artificial intelligence continues to become a fixture in gaming, it's worth cementing one creed in your mind: a crisis doesn't have to be apocalyptic to be real – and likewise, a platform that's already drowning its best work doesn't need much more water.