Snapshot Every measurement here was taken on 10 September 2026 at 10:00 UTC, 21 hours after version 1.0 went live. This is not a full launch teardown: one day has passed, and there will be nothing here about sales. What we do have is a detailed picture of that first day, and the same game across the seven weeks before it.
01The short version
At 12:56 UTC on 9 September, Valheim left Early Access. Within six hours concurrents went from 17,260 to 168,093 — the fifth-highest count on Steam that day and the largest Early Access graduation of the past seven weeks: the next best is 37,190, by Scrap Mechanic.
A fivefold jump in concurrents looks like a sales event, but there is barely any money in it. Among the people who wrote a review after release, new buyers are 8%, and the median author had logged 160 hours before this week even started. The players who came back are the ones who bought the game long ago — while the fresh demand that day was not buying but shelving: Valheim climbed the wishlist-activity chart from #192 to #68 in twenty-four hours.
Hence the question of this issue: what is leaving Early Access, really — a sales event, or a free homecoming for an audience you already have? The answer comes from the reviews and from 412 other graduates.
There was no discount that day: the price stayed at $29.99. And that is the price that went up from $19.99 on 26 July — a month and a half before release, quietly, as the summer sale ended. The release date itself was announced on 7 June, 94 days ahead.
And then the date itself. At 12:56 Steam rewrote Valheim's release date: the store page stopped saying "2 February 2021" and started saying "9 September 2026". In the platform's own data a five-year-old game became a new release, and it entered the New Releases shelf in 41 countries. Section 08 covers that.
02The minute of release
The moment of release is known to the second: 9 September, 12:56:32 UTC. That is when Steam stamped the game with a new release date in place of 2 February 2021. A minute earlier, at 12:55, the studio had posted "Valheim 1.0 Has Arrived!" to the game's news feed — announcing the launch before the platform had processed it.
From there everything else took twenty-five minutes to follow, and the storefront shelves lit up before the store page did.
| Time UTC | Where | What changed |
|---|---|---|
| 12:56:32 | Release date | Rewritten to 9 September 2026; the earlier date, 2 February 2021, is no longer shown anywhere on the storefront |
| 13:18 | New Releases shelf | Valheim appears among new releases at #4, in all 41 countries at once |
| 13:20 | Steam front page | First position in the main featured block in 40 countries, and first in the new-releases block |
| 13:21:52 | Store listing | The Early Access genre is gone; 53 achievements appear, along with new trailers, screenshots, capsule art and a rewritten short description |
| 13:21:53 | Store page | The Early Access banner disappears |
| 19:29 | Concurrents | Peak of 168,093 players |
A Steam release is a dozen independent changes, and some of them flicker: at 13:37 the store page listed zero achievements, at 13:52 it said 53 again. The change of status also shows in the reviews themselves: Steam tags every review with whether it was left during Early Access, and all 10,537 of the recent months written before 12:56 carry that tag. Of those written afterwards, exactly one does.
Seven games left Early Access on 9 September, and one of them did so in the same minute as Valheim — Asgard's Fall — Viking Survivors, also Viking-themed, at 13:19:37.
03Five years and a flat line
This launch has a "before". For fifty-two days straight the daily peak stayed in a band between 19,000 and 35,000: 22,000–24,000 on weekdays in quiet weeks, a third higher at weekends, overnight troughs down to 7,500. The highest day in the whole stretch was 35,036, the last Sunday of the July sale.
For fifty-two days the line stays between 19,000 and 35,000, and its only visible bump is the 50%-off sale in late July. The last bar is 9 September: 4.8 times the previous maximum.
The run-up shows in five traces:
- 7 June — the studio announced the date ("Valheim Has A Release Date!"), 94 days out; the trade press picked it up the next morning.
- 19–26 July — a 50% discount ($19.99 → $9.99) and a slot in the front-page specials block in seven countries. The last sale before release.
- 26 July — the price rose to $29.99: €19.99 → €29.99 in Europe, R$79.09 → R$89.99 in Brazil. For six weeks the game sold in Early Access at the new price.
- 20 August — the studio's last dev diary before launch, "Almost there…".
- 8 September, one day out: the game enters the front-page top-sellers block in twelve countries at once and starts climbing the sales chart — from #100 on the evening of the 7th to #62 by midnight.
The last three days before the switch are no longer flat. Monday 7 September peaked at 33,968 — while every other Monday through the summer and autumn sits between 22,000 and 24,000; Tuesday the 8th did 27,022 against the usual twenty-two. In the sales chart the game travelled from #100 to #44 before the status changed at all. Part of the lift happened before the event, and that changes the arithmetic: measured against a quiet weekday baseline (21,742 on average through August) rather than an already-running week, the launch peak is not ×4.95 but ×7.7.
On launch day the game already held what a new release cannot: 1,002 active curator reviews (866 of them positive, 1,001 written before release), 524,796 community followers, six live bundles and 5,519 games whose store pages point at Valheim through Steam's "More like this".
04The first day
Release landed at 12:56 UTC — morning in the Americas, afternoon in Europe. Going in, the game had 17,260 players on it, an ordinary midweek number. An hour later: 84,144. Five hours after that, the peak.
On the left, an ordinary day of a five-year-old game with an overnight trough of 9,874 players. On the right, the same application a few hours later. There is no discount between them. There is the 1.0 update with a new biome, 53 new achievements, first place on the front page in forty countries, and a removed Early Access banner: which of those did the work cannot be separated on one game.
The curve climbs without a warm-up: growth from the very first hour, the peak on the sixth, no steps. Then an evening slide and a second day that says more about the launch than the first did. The daily peak on 10 September is 136,103, 19% below yesterday's — but the overnight trough held at 59,497 against 9,874 the night before: at the quietest hour the game was six times busier than a day earlier. That is the first sign that some of the arrivals stayed.
One thing not to confuse Every "×" in this issue is measured against the game's own baseline in the week before 1.0, not against its original launch. For scale: Valheim's record is 502,387 concurrent players in February 2021, per SteamDB. Yesterday's 168,093 is a third of that — leaving Early Access lifted the game a long way, but not back to its own record.
05What Valve provided
Storefront support arrived twenty-two minutes after release: first position in the main featured block of the Steam front page in forty countries, plus first in the new-releases block. By evening Valheim had slid to sixth in that block — front-page slots are shared out among the day's releases.
The climb started a day before release: #100 on the evening of 7 September, #62 by midnight on the 9th, #48 by morning. Release shows as the kink at 13:00 — from there the game went from #44 to #6 over the day.
| Shelf | Best rank | Countries | What it means |
|---|---|---|---|
| Front page, featured | #1 | 40 | Steam's largest shelf; #9 at the snapshot |
| Front page, New Releases | #1 | 41 | Gone from the block by 19:00 UTC |
| Search, New Releases | #4 | 41 | #91 a day later: the list sorts by recency, so later releases push it down |
| Country top sellers | #4 | 41 | Still inside the top ten in most countries at the snapshot |
| Global sales chart | #6 | — | The climb started a day before release; best rank came on day two |
| Most-played chart | #9 | — | Computed over full days, so it lagged: #72 on the first evening, caught up by morning |
| Wishlist activity | #39 | — | #192 before release, #68 that evening, #39 by morning: demand keeps accruing |
The last row deserves a note of its own. Wishlists are usually treated as a pre-release quantity — demand that later converts. Here the game climbed the wishlist-activity chart from #192 to #68 on the day it became buyable. What sits behind that cannot be read off one series: some defer the purchase until a sale, some just bookmark the game. But demand that day was visibly accruing somewhere other than the till.
06Who writes the 1.0 reviews
9 September brought 1,668 reviews against 303 on the 8th and 155 on the 7th. 1,251 of them were written after release — about 113 an hour against thirteen the day before, a ninefold jump, twice as hard as concurrents moved. By the morning of the 10th the rate had settled to 62 an hour: the review wave recedes faster than concurrency does.
The interesting part is not the score but playtime at the moment of writing. Its median is 160 hours: a 1.0 review here is mostly written not by a new buyer but by someone with weeks in the game. Those hours were not banked in a day — the game had been on sale for five and a half years. Across the fortnight around release the median author logged just one hour, and already had 154 before the wave began; the top tenth had over 600. The day's record holder posted at 13:06 with 5,254 hours behind them.
Grey bars are the previous month's baseline, coloured bars the launch window. The 200-hours-plus group grew from 36% to 43%, while the 2–10 hour group fell from 7.6% to 5.7%. The line is the positive share: it falls along with playtime.
New buyers are thin on the ground: 8.2% of the reviews come from accounts with under ten hours in the game in total, and 2.5% from accounts with under two. That is a lower bound, because a newcomer rarely writes on the first evening — but the order of magnitude is clear: the concurrency spike was made by returning owners, and 948 of the 1,251 authors had over fifty hours logged before this week.
The score falls with playtime: among those with under two hours, 59% are positive; in the 2–10 hour group, 72%; among veterans past 200 hours, 93%. On its own that is not a finding but a selection effect — whoever bounced off the game stopped playing and stayed on a low playtime. What is worth looking at is the size of the gap: thirty-four points between a newcomer and a veteran of the same game on the same day. The aggregate barely moved, though: the full day of 9 September comes out at 91.4% positive against 90.3% for the previous month. It only looks low next to the final pre-release days, when the people writing were the happy ones: 94.4% on the 7th and 95.2% on the 8th.
| Playtime at review | Share of reviews | Positive | Share in the month before |
|---|---|---|---|
| under 2 hours | 2.6% | 59% | 1.4% |
| 2–10 hours | 5.7% | 72% | 7.6% |
| 10–50 hours | 14.2% | 84% | 19.2% |
| 50–200 hours | 34.5% | 92% | 35.8% |
| over 200 hours | 43.0% | 93% | 36.0% |
One more detail: 280 of the day's 1,668 reviews come from key owners rather than Steam buyers, and those never reach the game's public counter. Steam's daily statistics count store purchases only: its figure for 9 September is 1,339, ours for purchases is 1,388, and the remaining 280 are keys. The fifty-odd between 1,388 and 1,339 is the usual gap between two recounts of the same day.
The rest in brief: under one percent of the reviews already carry a refund flag, and the median review is 43 characters long. Over half are written in English (87% positive) and one in six in Russian (90%); German authors have a median playtime of 251 hours, French ones 216.
07What leaving Early Access is worth
One case cannot size the effect, so we took the others. Between 19 July and 9 September, Steam graduated 412 games out of Early Access — between 22 and 78 a week: every week about fifty store pages change status.
The effect is worth measuring where there was real concurrency beforehand: 37 games. For each, the peak over the two days after the switch divided by the peak over the week before it. Valheim has not had its two days yet, and its ×4.95 covers only the first nine hours: that is a floor, and a fair comparison will only be possible on 11 September.
The diagonal is "nothing changed". The dashed line is the cohort median, ×3.03. Valheim sits in the top right: the largest of them, and above the median — but the multiple record belongs elsewhere. Garden of Witches grew twentyfold, from 60 players to 1,217.
| Measure | Value | Notes |
|---|---|---|
| Median uplift | ×3.03 | Quartiles ×1.38 and ×7.06; ninth decile ×10.02 |
| At least doubled | 24 of 37 | 65% of the cohort |
| Flat or down | 5 of 37 | Worst case ×0.46 |
| Valheim | ×4.95 | 33,968 → 168,093; between median and upper quartile |
| Largest before Valheim | 37,190 | Scrap Mechanic, 24 July; then Fields of Mistria at 27,092 |
Leaving Early Access moves concurrents on its own: the median game triples them. The spread runs from ×0.46 to ×20, and small games post bigger multiples simply because the denominator is small. "About ×3 on average" describes a distribution rather than predicting any particular game.
08A 2021 game on the New Releases shelf
At 12:56 Steam rewrote Valheim's release date, and from that minute the platform treats the game as having shipped on 9 September 2026. Valve does this for every Early Access graduate. The original publication date does not vanish — it sits in a separate internal field — but the store page, search results and new-release lists all show the new one.
The practical consequence: 43% of graduates land on the New Releases shelf — 178 games out of 412 across seven weeks. The rest never appear on it: getting there depends on what the shelf is busy with that hour, not on status. Valheim entered at #4 in 41 countries — a game with 486,855 reviews and five years of history has been listed among Steam's new releases for a second day.
For anyone counting release statistics this is a source of systematic error: the release date anchors every comparison — who a game is set beside, what day to count from. The size of the distortion can be measured. The "released 9 September" cohort holds 128 games, six of them Early Access graduates — one in twenty. By weight those six decide everything: the cohort's combined concurrency peak is 202,058 players, and 83% of it is Valheim alone. The median genuine new release that day drew two players.
Over seven weeks that has happened to 545 applications — Steam rewrote the date after they had already shipped, and for 131 of them it moved forward by more than a year. This is not a glitch and not a rarity: the platform works this way every day, and any shelf of new releases is half-built out of decisions about what counts as new.
So the release date cannot be one number. Three are needed: first publication, the date currently on the page, and the moment of graduation. For Valheim those are 2 February 2021, 9 September 2026 and 9 September 2026 12:56 UTC — and only the second is visible from outside. After the rewrite the earlier date is shown nowhere on the storefront, and recovering it means going back to archived snapshots of the page.
09What to take from this
Three things carry over from Valheim to somebody else's game, and none of them is about Vikings.
Burn the old price before you raise it. Valheim moved from $19.99 to $29.99 six weeks before 1.0, immediately after a week at 50% off. Anyone who wanted it cheap bought it cheap during the sale; by release the new price had had time to become the normal one. It left no visible trace of resentment in the reviews.
A graduate arrives on the New Releases shelf with a veteran's assets. On its "premiere" day Valheim held 1,002 curator reviews, 524,796 followers, six bundles and 5,519 games pointing at it through "More like this". A genuine new release has none of that, which is why the cohort median of ×3 does not transfer to games like this one: their peak is made by an audience they already had, not by the shelf.
A release date in your model is not one number but three. First publication, the date currently on the page, and the moment of graduation. For Valheim: 2 February 2021, 9 September 2026, and 9 September 2026 at 12:56 UTC. If your model has a single "release date" field, seven weeks have already driven 545 rewritten dates into it, 131 of them shifted by more than a year.
And the conclusion worth holding on to whenever a spike like this shows up: a rise in concurrents and a rise in sales are different events. Valheim's fivefold peak was made by people who had already paid — on average five years ago.
10What we do not know yet
One day does not show everything. What is missing:
- A full day of reviews. The 1,668 covers a day with only eleven post-release hours in it. The first ten hours of 10 September added 618, but the evening half of that day is still ahead, and its total will be known tomorrow. Lifetime, Steam's daily statistics account for 486,855 reviews, 93.5% positive, 57% of them written in 2021; the store page itself shows about 544,000 — the difference is those same keys.
- Sales. Steam does not publish them. On launch day the press reported 17 million copies over five and a half years — a publisher's claim, not a measurement. Earlier public estimates of the original launch — a million copies in five days, 6.8m by May 2021, 10m by April 2022 — come from GameDiscoverCo's write-ups — those are someone else's numbers, and they stand apart from everything else here.
- Wishlists. Valheim never appeared in the most-wishlisted chart: a game you can already buy does not go there. How many people were waiting for 1.0 we do not know.
11How we will check ourselves
Four predictions, recorded on the evening of 9 September. We will check them in the next issue, misses included.
| Prediction | Check on | Basis |
|---|---|---|
| The weekend peak of 12–13 September comes in above today's 168,093 | 13 September | For the median cohort graduate the peak lands on day two rather than in the first hours; and in ordinary weeks Saturday gave Valheim about +30% over weekdays |
| 10 September brings more than 1,800 reviews — the first complete day after release | 11 September | After release the rate held at about 113 reviews an hour; even a third slower, a full day lands near 1,800 |
| Valheim falls out of the top 100 of the New Releases shelf by 11 September | 11 September | It went from #4 to #91 over the day — roughly four places an hour — and that shelf sorts by recency, not by success |
| Two weeks out, on 23 September, peak concurrents stay above 40,000 — the new baseline holds above the old one | 23 September | The weak one: the graduate cohort has less than two weeks of follow-up, so this is extrapolation rather than measurement |
One of them is already visible at the snapshot: Valheim has slid from #4 to #91 on the New Releases shelf in a single day, so the prediction about dropping out of the top hundred is landing a day early. Day-two concurrency (136,103) holds at 81% of yesterday's peak, and reviews are running slower than two thousand would need: 618 in the first ten hours.
12Limitations
- The window is one day. Everything said about the shape of the curve and its decline rests on twenty-one hours. Yesterday may not turn out to be the peak.
- Concurrents are not sales. A large share of the peak is owners who bought the game years ago. We cannot separate them from new buyers.
- A cohort of 37 is small and mixed. It holds games with five thousand players and games with fifty, so a median over it does not transfer to an individual game.
- Steam's public counter and the review count are different quantities. The counter tallies store purchases only, so it reads 1,339 for 9 September while 1,668 reviews were written. The splits in section 06 run over all reviews; Steam's positive share runs over its own.
- 41 countries is the observation panel. "In 41 countries" means "in every country whose storefront was tracked".
Disclaimer. The material on this page is informational and exploratory. These are observations from open sources and estimates built on them: samples are small, the method is imperfect, calculations are approximate, and forecasts and interpretations may turn out to be wrong. We do not observe sales or revenue. Nothing here is business, financial or investment advice; any decisions you make on this basis are your own.
Observations: Steam storefronts across 41 countries, concurrent players, the store listing and store page, the sales, most-played and wishlist-activity charts, reviews, the recommendation graph, curator reviews, prices and bundles. Data snapshot — 9 September 2026, 22:10 UTC; storefront ranks are live and have almost certainly moved by the time you read this. The comparison set is the 412 games that left Early Access between 19 July and 9 September 2026, 37 of them with measurable concurrency. Sales estimates for Valheim's original launch come from GameDiscoverCo and cannot be independently verified by us. The all-time concurrency record comes from SteamDB. The game's page is Valheim on Steam, the dev diaries are in its news feed; the studio is Iron Gate, the publisher Coffee Stain Publishing; the date announcement was covered by GamingOnLinux. Cohort games mentioned: Scrap Mechanic, Fields of Mistria, Sephiria. Previous issues: Static Dread, Servant of the Lake, The Sorting Bureau, How to Fish.
The researches and datasets behind this issue were prepared on the PlayRecon v0.3 platform.
Figures from studio posts and third-party publications are cited with a link to the original and remain the property of their authors. If you quote or reuse this analysis, please link back to the original.