Press start: every story, 2+ sources

Home / Indie

Steam Next Fest hit 7,240 games in October, and most demos barely registered

Early data on October's Next Fest: participation is up 65% from June, a few demos take the players, and about a third of games disclose AI use.

Indie 4 min read

Photo: Shixart1985, CC BY 2.0, via Wikimedia Commons / Wikimedia Commons, CC BY 2.0

Steam Next Fest in October ran from the 19th to the 26th with 7,240 games, according to a scrape of the live Steam page by a developer-marketing analyst who writes as Zukalous. That is 65.2% more than the June edition’s 4,382 and roughly eight times the 887 of June 2023. The numbers below come from that analysis, which covers only the first two days and was updated on October 23, so read them as early.

A festival that is now most of the market

The most striking figure is the share of new releases that take part. How To Market A Game puts it at 33.6% of Steam releases in 2024, 38.2% in 2025 and 51.7% in 2026, with a caveat: the 2026 denominator is projected, 15,158 of an estimated 29,300. If participation stays near 38%, the author estimates about 40,000 releases in 2026. Whether or not that estimate holds, the direction is clear. Next Fest is no longer a showcase for a selected few; it is the default way to be seen.

Who actually gets played

The cost of that growth shows in the tiers. The analysis ranks games by peak concurrent users on the demo:

Rank Peak players Median followers before
Top 25 419 to 20,505 5,519
26 to 50 217 to 418 3,834
51 to 100 94 to 211 1,605
Everyone else (6,942 games) 0 to 94 15

Those 6,942 games had a median of 15 followers a day before the event and a median gain of one in the week before it. The author’s reading is that existing audience matters most: followers before the festival correlate with peak demo players at 0.55 on a Spearman test, and the raw gain in the week before at 0.52, while the percentage gain barely matters (negative 0.13). A game that arrives with a following has an advantage no fest page can replace. This is early, peak-based data, and the author plans a later look at follower gains after the event.

What Steam’s own top 50 shows

Steam also published a most-played list, which Steam Deck HQ reproduced on October 28. It leads with Worst Class Ever, then Probably Stolen, with Melvor Idle 2 at 25, Tropico 7 at 29 and Warhammer Survivors at 41. The ranking metric is play activity, not peak concurrency, so the two lists will not match. The Steam Deck HQ writer noted four entries that carry AI disclosures: Never Be Alone (AI textures and music), Teach My Little Sister How To Drive, SLAM TANK (AI for programming, 3D model refinement and some marketing art) and BreathBound (AI-generated art assets), and said there were more than he had expected.

The AI question in numbers

Zukalous tallied developer-submitted disclosures and found 2,502 games, 34.6% of the festival, declaring AI use. Disclosure falls as success rises: 8% of the top 25, 12% of ranks 26 to 50, 26% of ranks 51 to 100 and 35% of the rest. Most disclosures concern visuals (75.7%), then text and localisation (37.7%) and audio or voice (37.6%). Code tools account for 23.1%, which the author thinks is understated because Valve does not require that disclosure. Only 0.3% (eight games) call themselves substantially AI-built.

Genre matters too. Strategy RPGs disclosed at 54.5% and real-time strategy at 50.4%, against 17.9% for 2D platformers and 15.3% for precision platformers. The author guesses that the gap comes partly from a higher share of Chinese developers in strategy and RPG, but labels that a hypothesis. Disclosures are self-reported, so the real figure could be higher or lower.

What to take from it

First, a demo is not a plan. With seven thousand pages competing for the same eight days, the long tail gets almost nothing, and this analysis says the reason is mostly what you had before the festival began. Second, the AI figures describe disclosure, not quality: the two disclosing games in the top 25 by peak players (BORE and Path of the Yaoguai) show that players are not avoiding them wholesale, but also that most of the weight sits in the bottom tiers.

For a small team, the practical lesson is dull but useful. Build a following months ahead, release the demo early and treat the festival as a megaphone for something that already has momentum. For readers, the takeaway is the reverse: the best demos are probably not the ones in the front of the queue, and the top 50 list is the faster filter.

Sources

  1. How To Market A Game: How sloppy is the October 2026 Steam Next Fest? howtomarketagame.com
  2. Steam Deck HQ: The top 50 demos from Steam Next Fest October 2026 steamdeckhq.com