Link: https://pivot-to-ai.com/2026/10/12/ai-game-demo-is-going-broke-on-just-430-players/

all my apes gone

https://www.youtube.com/watch?v=SwZ6_Q2lnsc&list=UU9rJrMVgcXTfa8xuMnbhAEA - video
https://pivottoai.libsyn.com/20261012-ai-game-demo-going-broke-on-just-430-players - podcast

time: 4 min 47 sec

That's at the investor subsidized rate as well.

Epic idiocy.

They’re spending about $70 per month per player.

The developers have taken out a bank loan to keep the demo running. A free demo where the whole way the game works relies on a paid service.

The financial end game here is not clear. Who will pay for this in the release version? The developers think the players will.

Can't wait for these jokers to learn the hard way that they don't understand The Market.

Absolutely incredible. Easily the funniest thing I read today. It's like a perfect micro-example of the AI bubble.

Man, every time I hear about the cost of an actual AI application, it is always something obscene. At least this isn't as bad as Steve "Gas Town" Yegge spending $30k/year for a dozen subscriptions (that cost Anthropic ~$1M/year in token costs) to develop his game with 6 active players.

At this point I have zero patience for my colleagues who think that the AI companies could make the economics all work out. Doesn't help that mathematicians are the least cost-conscious people on the planet, and the AI companies are massively subsidizing their subscriptions in particular.

Huh, this video got flagged as mature content. Unusual move for this channel...

I think the headline is a mistake. I believe that's peak simultaneous players, not total.

You know the funny thing? They're all competing hardcore right now selling below cost so they can become monopolies, the more they get used the more data so the better their product becomes, so the prices they're spending are lower than they should be.

Literally ai investors are eating the cost of people using a shit ton of AI for dumb questions or being lazy and letting ai do stuff like summarize cooking recipes that users could just read on their own.

It's cheaper now than it'll ever be, and whatever comes back after bubble bursting will be more expensive likely, something where you don't want to spend that much to code a feature that you can just figure out on your own.

whaaat? I'm not seeing that in my channel interface and it's not showing as such in a private window. What are you seeing?

Not convinced that makes any substantive difference.

It depends on how it plays. If thousands of people keep trying it for short periods of time the business model is less stupid than if a few hundred people play it nearly 24/7. Like are they spending 70 dollars per potential sale or 3?

Is an arc of play 3 hours or 60? etc.

Also just generally when people are being stupid it's funnier to describe what they're doing accurately.

The current version is a demo, so they are making $0 per sale. The SteamDB graph for the demo shows that the daily maximum player count has been around 300 for the past few days, and there are around 1000 followers for the unreleased full version, so I assume that a large fraction of those players are consistent (for now). Unfortunately, I could not find any data about the total number of downloads, and I don't think that information is public.

But setting that aside, every game has a player base that follows a power law. Yes, most players are casual, but the most dedicated players spend a disproportionate amount of time and money. In mobile games, these people are called "whales", and they are the backbone of the business. For example, 50-70% of a mobile game's in-app purchase revenue comes from 1-2% of the players. Even if most players spend a couple hours on the game, the developers must account for the people who will sink a ton of time into it.

The big problem with the AI business model is that this dynamic is now working against them. These whales now cost tons of money. They need to somehow charge these players at least $70/mo, with a crappy novelty game that would barely scrape by for $10 on Steam. Good luck with that.

Looking at the bigger picture, this is why AI subscriptions are not a viable business model. Whatever morsels of profit you might get from your regular users are more than cancelled out by Steve Yegge spending $2.8k/month for $87k/mo worth of tokens. Whatever revenue the AI companies get largely comes from convincing executives of a few rich companies to fork over absurd amounts of money, which produces a massive concentration risk in their revenues.

I have restricted mode turned on one of my devices because I often use it to watch stuff with my kids:

I hope so. But I have a feeling the local LLMs will just pick up the slack, running on hardware from liquidated "ai data centres"

Did you read the article? Getting more eyes on a game where every active player loses you more money is a bad thing.

“Teach My Little Sister How To Drive” is an early-release game demo on Steam. You speak to the game and tell the sister character what to do, and you see if you can get her not to crash this time. The whole interface is voice. It’s a cute little idea.

So the entire game is based around the fact that it is difficult to get LLMs to follow instructions and do what you say?

@Nouvellalia @flowerysong did you get lost on the way to LinkedIn?

I’m so honored that one of the fucking tools behind teach my little sister how to drive has decided to join us

please lean basic arithmetic before you float another bank loan

@self @Nouvellalia this seems like the epitome of "play stupid games, win stupid prizes" 😂

@NouvellaliaAre there, or will there be in the near future, backend AI models that would make all this stuff profitable?If not, that seems bad, business-wise.@flowerysong

I'm not a big business boy, but I think they should have considered the costs of AI before getting to the point of spending $1000/day and taking out bank loans.

@ozzelotWhat do you think? If there is a way to make the models perform better by let's say factor 10 (and for running a game in parallel you need more than factor 10), would all the companies build massive datacenters with massive power and cooling water needs and hog all the hardware available on the market so that the only RAM available for consumer systems is fucking DDR3 modules?Or in short: Hahaha, no!@Nouvellalia @flowerysong

Is it laziness or impossibility that they don't actually use a local model for this?

Have to admit, it is a cute idea. If this used a local model, it could probably function just fine. The only downside, then, is the plagiarism that went into the model.

This is so baffling to even hear about but imagine someone coming up with this and thinking that it’s a good idea.

Local models are astoundingly slow compared to online ones and you can't really run anything else while they are doing inference.

Probably the second, in the vast majority of PCs local inference should be painfully slow and resource intensive.