Hi, it is Steam Next Fest. This blog post was written on Wednesday October 21st at 10AM PT and we are a couple days into the October Steam Next Fest 2026 (from here on out referred to just as SNF).
The first Wednesday of SNF is typically the day Valve switches the algorithm from “all games get a baseline of visibility” (see footnote about this) to “top performing games get visibility.”
These are my initial thoughts on what I saw in the first two days.
Also I updated my Steam Next Fest tracking tool so you can input your Steam Next Fest traffic and wishlist data and it will tell you how well your game compares. The comparison graphs are based on real data submitted from February and June 2026 Steam Next Fests.
Track your Steam Next Fest Progress
Sloppiest Fest
This was the big one. 7,240 games in October 2026 Steam Next Fest (**Footnote**). The number of participants jumped 65.2% over the last one.
Here is a history of Steam Next Fest. Data provided by this great blog post from GameDiscoverCo.
Yes, AI slop. But one thing underdiscussed is I think this is also developers finally learning that Steam Next Fest is a thing. Surprisingly, most games on Steam never participate in this 100% free opportunity to get wishlists.
If I do some crude math here you will see that traditionally only about one-third of games that release on Steam participate in Steam Next Fest. Basically I just count up the number games that participated in one of the 3 next fests and divide it by the number of games released that year. YES I KNOW, many games don’t always take part in Next Fest the same year that they release. But it gives you an idea of the proportion.
So now in October 2026 we are seeing the percentage jump from 38% – 51%. Couple thoughts here…
First: Maybe AI slopsters are jumping on the first SNF to test whether their game has “the juice” and will release in 2027. Basically in 2027 we are going to see even MORE games released.
Second: My denominator for 2026 is projected. Typically in October we see 76% of total releases. But maybe in the new AI world that number is going to change so instead of the usual 24% of games releasing betwween now and December 31, we get a late flood. MAYBE we use SNF participation to estimate total releases. If that is true, assuming 38% of games participate in SNF then 38% holds: 15,158 / 0.38 ≈ 39,890 games. Maybe 40,000 games will release in 2026. MAYBE
Who knows… back of the napkin here.
Who is winning this SNF?
I scraped the data for all the games on Steam Next Fest and here are the numbers for the games ranked by their Peak CCU so far. Note the
| Metric | Top 25 | Ranks 26–50 | Ranks 51–100 | Everyone else (101+, 6,942) |
|---|---|---|---|---|
| Peak CCU range | 419–20,505 | 217–418 | 94–211 | 0–94 |
| Median followers, 1d before SNF | 5,519 | 3,834 | 1,605 | 15 |
| Median followers gained, 7d→1d before SNF | 220 | 200 | 57 | 1 |
| Median % growth, 7d→1d before SNF | 7.0% | 4.0% | 4.1% | 12.5% |
| Avg days demo live before SNF | 18.8 | 26.0 | 26.9 | 36.6 |
Is raw wishlists or momentum better?
This is the eternal debate: should you wait wait wait to do SNF so you can build up a huge pile of wishlists. OR, should you keep your game secret and then spring it on the world just before SNF to get a sudden burst of excitement (and wishlists) to trigger the algorithm.
If we do a correlational analysis on the games in this SNF looking at CCU, the numbers are about the same. Slightly better to have a ton of wishlists (aka followers)
| Correlation with peak demo CCU | Spearman ρ |
| Total followers, 1d before | 0.55 |
| Follower gain, 7d→1d (raw) | 0.52 |
| Follower gain, 7d→1d (%) | −0.13 |
BUT, the best thing is to have a bunch of wishlists before SNF AND get a ton of wishlists in the weeks before.
But this is early data and based on demo CCU. I will be doing a bunch of analysis with total follower increase after the festival wraps up.
Let’s talk slop
I saw a meme like this on Reddit, but I couldn’t find the original so I recreated it from memory. If you find the original email me and I will update it.
I scraped the AI disclosures for all games in SNF. I know, most developers are not being 100% honest in their disclosures. All I can say is that I am just reporting on what people disclose. I am just the messenger. I didn’t do a game by game guess to see what was really AI or not. I can’t even tell myself most of the time.
| Metric | Top 25 | Ranks 26–50 | Ranks 51–100 | Everyone else (6,942) |
| AI disclosure | 8% (2) | 12% (3) | 26% (13) | 35% (2,427) |
So if we assume everyone is being honest here, the AI slop games really are the bottom tier of games. Only 2 games in the top 25 disclosed AI use. Again, maybe top games have more to lose if they reveal AI use and so chose to hide it. Who knows, we just have to trust.
One thought here, I know everyone is afraid of the Needle In A Haystack Simulator problem where their game is stolen and the stolen game gets all the glory. But that is a rare case here. AI Slop games don’t make it to the top on the backs of stolen games.
What are those two games with honest devs in the top 25?
BORE which said “This game was developed with help from an AI assistant under human direction and review. No content is AI-generated while the game is running.”
Path of the Yaoguai which said “部分游戏图片由AI生成并修改” with google translating that as “Some game images were generated by AI and subsequently modified.”
What is the most disclosed AI usage
AI disclosures are just free-text boxes that developers can write. So every disclosure is unique. But I used some common keywords and during the data scrape I tried to match them to some broad categories. Based on keywords here are the breakdowns for AI disclosure:
Edit: Some folks notified me that Valve does not require code disclosures. From Steamworks “We are aware that many modern game development environments have AI powered tools built into them. Efficiency gains through the use of these tools is not the focus of this section.” So it is low because you don’t need to disclose. That 8% is just people proactively saying they used code gen.
Overall, among fans and developers, AI-gen code seems to be the most permissive form of AI use. In a survey run by GameDiscoverCo they found:
On the ‘acceptable’ side, the clear winner is coding/programming: code helper tools like Claude Code had ~239 mentions as acceptable. This was way more than other categories like prototyping & and placeholder assets (119 mentions), and/or tedious or repetitive tasks (101 mentions).
Who is the sloppiest?
I then took the AI disclosures and cross-refernced them against the opt-in genre pairs. Here are the most and least sloppy genres:
RPG and Strategy games are the top AI disclosed. The least were Platformers. Here is the chart for this.
The genre pairs are based off the developer “opt-in” genres. When registering for the October Steam Next Fest, Valve added a new feature for developers to specify what genre and sub genres their games belonged to. If you go to the SNF Genre tab, you can browse the combinations:
Why are Strategy and RPG games so sloppy?
I think this comes down to Chinese developers being more comfortable with AI use in games, and Chinese developers more likely to make RPG / Strategy games. Here are the top 3 (by CCU) AI disclosed Strategy RPG games. All of them are Chinese studios.
- Path of the Yaoguai
- General Jeon.
- 灵劫变 (Chinese only but that translates to Transformation of the Spirit Tribulation)
There are a ton of Three Kingdoms ai generated strategy games out there.
Regardless of AI slop, what are the most popular genres this SNF?
I used the peak CCU count during the first 3 days of SNF and cross referenced it to the opt-in genres. Here are the most popular types of games on Steam:
So RPG, Strategy, Crafting, Building. Or as I like to say Crafty-buildy-strategy-simulationy games (I don’t make this stuff up).
And the lowest in demo CCU? Platformers (I swear, I don’t hate platformers, it’s just that Steam players don’t like them).
People are making slop in the right genres
The interesting thing here is that ai slop games are being made in the most played genres. I wonder if these developers are using AI to also do the genre analysis.
So I’ll do more analysis after SNF ends.
Now I need to go research whether you should give up game development because someone cloned that Needle in a Haystack game and that Drinking with an anime girl game. I promise I will discuss those 2 games that spooked the entire games industry. But please, I will tell you, don’t freak out. Please, just, don’t make any big dumb decisions. And join the mailing list to be notified when I write my “should you quit game dev because of a haystack” blog post.
Footnote
“7,240 games in steam next fest” – You will see many articles say 8,000. That is because there was a press preview 2 weeks before and it had 8000 something games there and that is when the reporting on it started. However as is always the case, developers pull their game for a number of reasons. The number I quote here is from scraping the live SNF steam page.
“Games get a baseline of visibility” – Many people misinterpret this. Valve does not give every game the SAME VISIBILITY on day 1 and day 2 of SNF. Valve just gives every game the same minimum via randomized front page placement.