Power Laws

Game sales follow a power-law distribution. A tiny fraction sell exceedingly well and the vast majority sell very little.

We’re about to see a step-change in the number of games on all of the major distribution platforms. The mobile app stores are already showing signs of being overwhelmed with content - app review times are shooting up. This increase is coming to Steam. It seems likely that the number of games released will jump by between 2x and 10x. Even 5x sounds unfathomably high, but remember the vast majority of those games never sell even 100 copies.

This expansion of quantity could go in a number of ways. The two extremes are:

1. Slop - The content is bad and even the near-infinite monkeys never write Shakespeare. The tail of the curve stretches out more but games that generate meaningful revenue are totally unaffected.
2. More of the same - The new games follow the same distribution as the old but there are 5x as many fighting for the same revenue/attention. All games at all parts of the curve divide the same pie.

There is a 3rd extreme case , “Disruption” in which the vibe-coded games are actually higher quality than the previous generation. If we’re assuming 5x-the-content, this economically maps out roughly to be the same as “More of the same”, but with incumbents getting kicked out. Good topic for a future post but if you’re an Ai-agnostic publisher or developer, you’re still facing a content explosion, just a slightly smaller one.

Publisher Top-Grossing Fragmentation

One lens through which we might analyze the likely outcome is by examining what happened to the top-100 grossing games on Steam over time. We’ve seen step-changes in the quantity of games on Steam a couple of times before. When game engines became broadly available and as Steam gradually opened up. A quick timeline of Steam’s history:

• 2005 - Steam begins distributing 3rd party titles. Value hand-picks the games.
• 2012 - Steam “Greenlight” allows community voting for what games will get to be listed on Steam.
• 2017 - Steam Direct launches allowing any publisher to distribute games on the platform

Immediately prior to Greenlight, the weekly top-100 grossing spots were divided between just 51 publishers. A year after Greenlight 63 unique publishers were in the weekly top-100 and by the time Direct launched it reached 70. At the end of March 2026, it was 72. It’s not necessarily true that as there are more entrants the number of winners goes up but that’s what happened over the course of “Greenlight”.

The notion of a “publisher” is really just a “label” in this chart. When Valve was acting as a gatekeeper, bound by forming relationships with a small number of publishers, those publishers maximally leveraged being inside of the gate by putting lots of content there. What is perhaps more interesting is that from 2017 to today, the number of unique publishers in the weekly top-100 has remained fairly flat, suggesting that perhaps the power laws are fully in effect and there’s not going to be much change even if we 5x the number of games.

Power Law Eats Quantity for Breakfast

The top-100 view of the ecosystem hides what happens in the “More of the same” extreme as we need to look at revenue rather than ranking in the face of more games. Games are constrained right now by customer attention - I like to call it “eyeball-hours” and absent some sort of platform change or fundamental way in which players engage with games, I don’t expect we’ll see a step-change in eyeball-hours or revenue as a result of Ai-assisted development. This influx in quantity won’t increase revenue of the ecosystem.

Some data to emphasize the steepness of the curve.

• Out of ~19,000 games released on Steam in 2025, just 12 grossed an estimated >$125M (GameDiscoverCo)
• Roughly 40% of releases didn’t recoup the $100 Steam listing fee (OpenCritic)
• Of ~41K games released over a 3-year window, only 5.6% grossed over $200K, while 76.5% earned less than $5K (Gamalytic)
• Across all ~71K games on Steam at the time, 67% had lifetime revenue under $5K and over 50% never exceeded $1,000 (Gamalytic)

I asked Claude to make a little toy to explore these distributions. α ≈ 1.3 seems like a reasonable fit for the part of the curve we care about. The tail of the distribution on Steam doesn’t fit a power-law cleanly.

The power laws curve is unintuitive. When you add 5x more games there is almost no impact on the part of the curve publishers care about. When you add a ton more games, you end up with tons more games making < $1M and almost the same number of games making > $1M. As much as I’d love the resulting 2nd indie golden age, we won’t have more, smaller winners. Instead, the chances of having a hit are about to decrease dramatically. It’s possible some innovation in discovery or new marketing channel that highlights games currently sitting in the top-1000 upsets this.

Quality

In product management, we talked about revealed preferences - listening to players’ actions rather than words in order to understand what they care about. The best designers, game directors and product managers in the world are the ones who have the intuition to get this right but as evidenced by the number of high profile failures and cancellations from even the best developers this is still a highly fraught endeavor. We all remember Borderlands from Gearbox, but forget Battleborne. Fortnite was about to leave an Epic-sized crater before they brilliantly iterated their way to the game we play today.

The opposite is also true - time and again, we find janky, buggy games end up with high customer quality ratings. If you played PUBG in the early days (I was ranked in the top 1k globally at one point…by hiding from better players in a bathroom) or enjoyed the Elder Scrolls games, you know exactly what we’re talking about here. Great developers have a feel for what players will care about and that intuition develops the longer they work on a genre or franchise.

If things that developers strive for: quality, production values, polish, are unpredictably interpreted by players and there are 5x as many games out there, what is the revenue impact of all that polish? Doug Shapiro put it concisely in Pareto to Creato: “quality sets a floor.” If you look at Steam data, there is no correlation between review score and wishlist count, suggesting that “quality” isn’t the determining factor in success. Apply this thought experiment and I think you’ll see why: Pre-e-book, do you believe that all the best novels got publishing deals? Do you think that there were fantastic stories which never found their way to bookstores? What % of great books do you think saw the light of day? [Credit: GH] This analogy doesn’t pass the sniff test when a credible game cost $3M to make, but when it costs $300k everything starts to line up.

Externalities Overpower

If the impact of “quality” in the success equation is reduced, non-development factors end up having a larger impact on the chance a game will become a hit. A cynical view says that externalities and luck - market timing, “going viral”, app store algorithm changes, etc will take over. Indeed, the power of the distributors over producers will probably to increase. I’ve read some folks hoping to see Ai disrupt the Steam monopoly but it is more likely that Ai-augmented game development will be a sustaining innovation for game distribution.

There’s reason to hope that publishers will retain agency over their success, however. On Steam factors like store pages, screenshots, and marketing copy are inherently bottom-of-funnel and have little impact on success. More games will reduce the efficacy of Steam as a source for finding new games. There’s no reason to believe that paid marketing’s efficacy will be reduced. If we look towards mobile where there are ~210k games, roughly 2x Steam, we can see that paid marketing is still impactful thought this isn’t apples-to-apples, due to mobile having much more effective attribution than console/PC. Most importantly, players already discover games through their online and offline networks and publishers have control over that factor by injecting their games into those networks through existing methods like IP-licensing, community management, PR, influencer-marketing, etc.

We’re already seeing this start to happen with Hooded Horse and Coffee Stain. They’re great at identifying or developing games that cross the quality threshold but once there, being on their label appears to become self-fullfilling. Annumpurna Interactive is an interesting counter-example. Putting aside the recent controversy they rely on having exquisite taste rather than building and catering to a network of players. Outstanding execution in publishing without a network of players who have an affinity to the titles, exposed them to increasing market capriciousness.

Reduced Costs

The cost to produce equivalent scope and quality is going to go down as a result of Ai-augmented development. The two extremes this could lead us down:

1. Budgets remain flat and the increased productivity will go to making larger games with higher production values.
2. Game scopes will remain flat and the cost to produce those games will be lowered

Given the “quality floor” thesis, it seems likely that the market will settle closer to the lower budgets end of the spectrum except perhaps where AAA is concerned. Reaching the quality floor is incredibly challenging for developers but once you cross that threshold, marginal increases in production value are less likely to yield dividends than investing in go-to-market factors.

Great Publishing Wins

All of these changes point towards an increased criticality in activities bucketed under “publishing” in our industry. The traditional thinking in console/PC games (as well as film and TV) has been that developers create hits and publishers magnify them. With this next step-change, that rule seems likely to fall. Publishers (both pure-play publishers and developer-publishers) and their networks of players, have the potential to become king-makers just as distributors (Valve, Walmart, etc) were when ecosystems were closed.

Increased randomness means increased risk, but this will be almost exactly offset by a comparable reduction in development costs - we have N-times the number of games in part because we’re able to make that many with the same inputs. The benefits of those reduced costs will accrue mostly to the capital-allocating publishers in console/PC games because competition among developers for publishing dollars will become more intense as the number of developers skyrockets. Smaller, pure-developers get punished post-Ai. Their budgets go down which has unpleasant implications to their operations, they face more competition, and the risk of each title goes up. That’s a nasty combination. Maintaining a portfolio of games remains the way to stability navigate a climate like this one.

It’s really a perfect setup for the publishing side of the business to not only have greater impact but also capture the new value created. App store optimization, building a network of players, marketing, PR, CRM, price-management, geo-targeting, etc are all about to become between N-times more impactful. The quality-per-dollar-spent and the ability to have greater impact are multiplicative and then the resulting network of players is compounding with each title.

This is a personal blog, so let me conclude with a personal angle. I’ve always been a developer at heart - a maker of stuff and never really interested in the publishing business. What is about to happen strikes me as incredible exciting, though. I suspect game publishing will also go through a massive disruption in the process - new entrants will more quickly position themselves to leverage the potential and pre-Ai publishers will get run over by their more agile competitors. I hope the innovation will also happen inside of self-publishing developers and not just pure-play publishers. With a little luck, we’ll see cartels of developers form publishing arms to achieve the necessary risk diversification rather than having the mega-pubs reap all the upside.

Thank You

• Ichiro Lambe for reading and providing feedback. Check out https://weloveit.io/.