Steam Next Fest is meant to make trying unfamiliar PC games feel easy. Developers put forward demos for unreleased projects, players download them for free, and promising games can earn wishlists, coverage and eventually sales. It is a valuable idea because a short playable build can answer questions that a trailer cannot: Does the movement feel good? Is the central premise clear? Is this the kind of game someone wants to follow?
But an event built around discovery has a practical limit. The current Steam Next Fest reportedly contains nearly 7,100 free demos. At that scale, the headline is no longer simply that players have plenty to sample. It is that finding a handful of genuinely interesting games may require more effort than many people can reasonably give.
The expansion is stark. The June event had roughly 4,300 demos. October 2024 had about 2,700, while October 2023 was closer to 900. Those figures do not establish that a larger event is inherently worse, and more submissions can mean more opportunities for small teams. Yet the jump from around 900 to nearly 7,100 changes what Next Fest is. It is no longer a manageable festival shelf where browsing itself is enjoyable; it risks becoming a warehouse where every visitor needs a map, a filter and luck.
The discovery problem is bigger than a big number
Seven thousand demos is not merely a daunting total. Demos take time. A player needs to find a listing, determine whether it looks appealing, download it, play enough to judge it, then decide whether to wishlist it or tell somebody else about it. Even a very brief first pass multiplied across thousands of games is impossible for any individual.
That creates a visibility bottleneck. Every creator has access to the same giant promotional moment, but attention does not expand at the same pace as the catalogue. The finite resources are player time, media coverage, creator streams and the number of games people are willing to remember. A festival can add thousands of additional demos, yet it cannot add thousands of extra hours to each visitor’s week.
This is especially punishing for the sort of project Next Fest is supposed to help: a game from a small team without an existing audience. Established developers may arrive with followers who already know what to search for. A lesser-known game must first be surfaced, then earn a click, then survive comparison with every other listing competing for the same moment of curiosity.
In a smaller lineup, browsing can provide some serendipity. A strange screenshot, distinctive mechanic or unusual genre blend has room to catch a player’s eye. In a lineup of nearly 7,100, the same qualities may never be seen by the people most likely to appreciate them. The problem is not that every demo needs to be played. It is that players need a plausible route to the right demos, and developers need a plausible route to the right players.
Why AI disclosures have become part of the frustration
Volume is only one part of the complaint. The current event has also drawn criticism over demos described as AI-generated, derivative clones or products of “vibe-coding.” Many listings reportedly disclose extensive use of generative AI in their codebases and assets. That disclosure can make the scale problem feel less like an abundance of creative work and more like an influx of material made quickly enough to crowd out projects developed through slower, more deliberate processes.
Generative AI refers here to systems that produce material such as text, images, code or other assets from prompts and patterns in their training. The term does not, on its own, tell players whether a game is functional, enjoyable or responsibly made. It is a description of a production method, not a quality score.
Vibe-coding is a loose term for building software largely by directing AI tools in natural language and accepting, revising or assembling their output. It can lower the barrier to making a prototype. However, that same lower barrier can make it easier to produce and submit a great number of small projects rapidly. In a storefront event already defined by scale, the concern is not simply that AI tools exist. It is that the cost and speed of producing submissions may be moving faster than discovery systems can distinguish worthwhile work.
There is an important difference between disclosure and curation. An AI-use disclosure lets a prospective player know something about how a product was made. It does not solve the more immediate browsing question: Is this demo complete enough to assess? Is it distinctive? Is it a clone? Does it work? Does it align with what this player wants? A disclosure is information. Curation is judgment and organization.
The debate also should not flatten every developer into one category. The available information supports concern about a large number of AI-heavy or hastily assembled demos, but it does not show that every one of the nearly 7,100 is low quality, nor that every use of an AI tool produces a bad game. The trouble is structural: when questionable work scales quickly, the burden of sorting moves to the audience and to legitimate developers trying to be noticed.
Algorithms can personalize, but they do not replace a festival program
Steam’s approach is described as relying heavily on user tags and algorithms that learn what individual users may like. Those tools can be useful. Tags can narrow a huge list by genre or feature, while recommendation systems can surface games that resemble a person’s past interests. For someone who already knows they want a particular style of strategy game, puzzle game or shooter, personalization may cut through some noise.
But recommendation is not the same thing as curation. An algorithm generally attempts to predict relevance from available signals. A curated program makes an overt editorial choice to foreground work for a reason, whether that is originality, craft, a new perspective or a particular theme. The first is optimized around matching; the second can help audiences encounter things they would not have thought to search for.
That distinction matters in a demo festival. Next Fest is not just a storefront category. Its value comes from the sense that this is a concentrated moment to look ahead at PC games. If the experience becomes indistinguishable from searching an enormous, constantly growing store catalogue, the event loses much of its identity.
More structure would not require Valve to declare most developers unworthy of participating. The central issue is navigability, not a demand for a tiny gatekept list. Better pathways could mean more meaningful thematic collections, clearer separation of browsing modes, stronger ways to find newly surfaced and less-followed games, or tools that help players make a deliberately limited personal shortlist. Those are examples of the problem to solve rather than claims about planned changes.
The stakes extend beyond one promotional week. The platform’s yearly release volume has already passed the total number of games added during all of 2025. If AI coding tools enable increasingly rapid production of games and mods, the supply of new software may continue climbing. A discovery system that felt adequate at 900 demos may not be adequate at 7,100, and it will be under even greater pressure if the number keeps rising.
What players and developers can take from the pileup
For players, the most useful response is to reject the implied obligation to browse everything. A giant festival can make completionism feel like the default: there are thousands of free demos, so surely a diligent visitor ought to sample as many as possible. That is the wrong frame. The practical goal is not to assess the event; it is to find a few games worth remembering.
- Start with genres, mechanics or themes you already know you enjoy rather than an unfiltered event page.
- Set a narrow demo budget, such as a short shortlist for one evening, and give each selection enough attention to make a fair first impression.
- Use wishlists as a follow-up tool, not as a substitute for deciding what actually interests you.
- Seek recommendations from people whose tastes you recognize, because trusted human filtering is particularly valuable when the catalogue is overwhelming.
For developers, the size of the event makes clarity more important. A demo cannot rely on the festival itself to deliver a patient audience. A store page and demo need to communicate the core appeal quickly, because visitors are making choices among thousands of alternatives. That is not an argument that every game must become identical or chase a trend; it is an acknowledgement that a crowded setting makes a clear first impression more consequential.
The larger industry argument is already visible in discussion about what it means for studios to say they are “using AI.” That phrase can cover very different practices, a distinction explored in the broader debate over how game studios define AI use. In the context of Next Fest, players are likely to care less about a single label than about whether they can reliably find games that are polished, distinctive and made with intent.
Steam Next Fest still rests on a compelling premise: let people play tomorrow’s games today. Nearly 7,100 demos prove that there is no shortage of people who want access to that opportunity. The grim part is that access without intelligible discovery can become its own kind of barrier. If the event is to remain useful to players and emerging developers alike, its next challenge is not attracting more submissions. It is making the signal possible to find amid the noise.
