AI Game Factories Will Not Work

In recent months, I’ve come across a few individuals and teams building something I refer to as an **AI Game Factory** operation. The idea is basically to develop AI systems that can rapidly generate many new games — either reskinned clones or slightly modified versions of trending games — and push them out to “ride the wave” and make money.

While I think AI opens up quite a few new opportunities in our industry, this is actually NOT one of them. I’ll explain why.

AI certainly has a huge impact on our industry. It allows smaller teams to create and maintain high-quality games that previously required bigger teams, makes it easier to prototype new gameplay and themes, and allows people who couldn’t enter the industry before to do so much more easily.

What AI did NOT change is this basic business reality:

**Marketing games on mobile is SUPER EXPENSIVE** (and it’s only getting worse).

The basic reality is this:

1. The audience hasn’t grown significantly. There are still a lot of players, and they’re still spending more or less the same amount (through IAP and ad revenue), but while many more games are being produced (and faster), the audience is still roughly the same size.

2. Except for a few rare exceptions, there is almost no organic growth in the mobile market. When a game grows, it’s because of paid user acquisition. It can be ads, playable ads, influencer campaigns, whatever. But it’s almost all paid advertising.

3. More games + the same audience size (+ harder targeting) = even more expensive user acquisition. And we entered this AI era with very high marketing costs to begin with.

I won’t get into why advertising on mobile is so expensive these days. That’s an issue for a different post. But the reality is that even if you could create a gazillion new games a week, it means nothing unless you have a lot of money to test and optimize them, and then push each one with a substantial advertising budget.

How much money? Assume a cost per install (CPI) of about $1 to $3 in cheaper countries and at low volumes (which means it will get even more expensive if you target Tier 1 countries and higher volumes). To properly test D1 retention, you need a minimum of 250–500 installs per day (below that, the margin of error is too big). So even an initial test can cost between $2,000 and $7,000.

So while AI has made it easier to make games, publishing them (and monetizing them) has become even harder.

In fact, I would argue that for new studios today, unless you’ve raised millions of dollars, you’re better off working with a big publisher rather than publishing independently.

## About Trends

But wait, what about trends? What if you spot a rising hit game early, create a similar game very quickly, and ship it? Can’t you “piggyback” on the success of that trend and make money?

Probably not.

You need two things to properly ride trends: identifying the trend early enough and acting on it quickly.

As I mentioned, discovery on the app stores has been mostly dead for a few years now, and the main way players discover games is through marketing channels. It doesn’t matter if you develop and publish another Meowdoku, Cannon Shooter, or Rings game — if you can’t put serious money into marketing it, you won’t have players. Also, unless you have really strong (and expensive) marketing tools, you’ll probably miss the real trends.

I’m not saying there are no trends, but big publishers look at trend opportunities differently. Big publishers (with big marketing budgets) follow other big publishers closely. What they’re looking for is which games suddenly gain traction. New releases are an interesting signal, alright, but the real signal is when a game, even one that’s a year old, suddenly starts gaining users fast.

This indicates that the publisher behind it is seeing good enough numbers to justify pushing large amounts of paid traffic into it. It means the KPIs are good enough and the advertising costs are low enough to make economic sense. This is when other publishers identify the game and start producing their own versions to compete for market share.

If you also have the money to push marketing and compete for traffic on this game — excellent. You’re in a good position. But there’s probably no point in releasing a competing game without that budget. As mentioned earlier, organic discovery is practically nonexistent, and users discover games mostly through paid channels.

This also makes producing a lot of games and testing them until you find a hit even more expensive.

Remember the cost of testing a single mobile game we discussed earlier. It doesn’t matter that producing a game has become fast and cheap if testing a single game costs $2,000–$7,000 per test (assuming a $1 to $3 CPI).

The numbers I’m hearing from developers and publishers suggest it takes between 30 and 50 failed games to find one hit. It’s true that a hit game, managed correctly, can cover the costs of all those failed games, but you need to be able to finance those 30–50 failed games first.

That’s a lot of money. If you’ve raised it and you have it — congrats. You have a good chance of making it. If you don’t, you’d better partner with someone who does.

In this reality, the fact that you can generate many games quickly basically means nothing unless you have big bucks to back up all that spending.

## Some Words of Encouragement

I’m aware I sound a bit discouraging here. But this is the reality of the mobile games market.

AI can certainly help us make incredible games quickly, and this is a truly exciting time to be alive if you love making games. But mobile is a super-saturated, super-expensive platform.

Hit games still make a LOT of money. And a good game can live and generate revenue for YEARS. But it takes a big effort to get to that big reward, AI or not.

If you don’t have that kind of cash, what you should consider is either partnering with those who do or exploring other platforms where the cost of acquiring users is lower and organic discovery and growth are still possible.

Good luck!