AI War and the hidden cost of indie games

It makes more sense than it looks.

Earlier this month Chris Park revealed that his company could be bankrupt by November . His company is Arcen Games , the developer of popular space strategy game AI War . Despite that game's excellence, it wasn't a surprise to find he was struggling: most indie games developers do.

Prompted by the tale of trouble on his site, I sent him an email asking for more information. What are the costs involved in independent game development? What are the profits of a game like AI War? Chris was willing to talk numbers - something no developer ever does - and he kindly replied with a breakdown of the profits and costs of AI War and his latest game Tidalis . I'm reprinting his email in full below with permission.

Regarding the figures on costs and income, that's a bit of a tricky topic -- not because I'm unwilling to discuss it, but because the best answer is "depends on how you look at it." For two reasons:

First, that there was a lot of free labor on my part for AI War, and even that could be quantified multiple ways if you count the years of engine development or just count the 7 months of actual development before the game came out (and actually, I kept none of the profits for the first 6 months after the game came out, too, and instead just reinvested those in the game -- was working another job at the time).

Secondly, in the case of Tidalis, we had three guys who were doing all of their work purely for royalties -- which if you just look at Arcen's expenditure it obviously misses that entirely, but I think it's worth counting in the sense that they invested heavily of their time and efforts; we've got a pretty tight team here.

So it's a complicated discussion, to say the least. I can say the following:


  • The general engine behind AI War and Tidalis, etc, was in development off and on in my spare time since about 2003 -- maybe 800 hours into that.
  • AI War itself was developed over the course of 7 months, and was conservatively 1000-1200 hours of my time. So, you could call that around $50k of work, though it was obviously speculative.
  • My out of pocket expenses prior to AI War's 1.0 launch totaled around $2.5k.
  • I then continued working on AI War for another 6 months without pay, at around the same rate, so another 850-1000 hours there. So, maybe $40k of work, something like that.
  • At that same time, I can tell you that Arcen spent $28,884.30 on other staff and contractors in 2009, all of that going back into AI War.
  • Arcen's total income in 2009 was $77,026.76, but most of that came in the last two months of the year. Through October 31st, revenue was $18,121.72.
  • Given other miscellaneous expenses, Arcen showed a profit of $38,299.91 and ended the year with a bit less than that in our bank accounts (before any taxes, which are about 35% for us, and were not first due until January).

It was in December 2009 that I decided to go full-time, despite some of the risks involved. For health insurance reasons I believed I needed to as my wife and I were trying to conceive, and I wanted the pregnancy to be covered, etc. I also felt that the quality and volume of my output would be increased enough by going full time with AI War, Tidalis, etc that it would help increase profits. Until recently, I was right.


  • Total income for Arcen during this period was $101,401.38
  • Total expense for Arcen during this period was $91,314.76, including paying staff and myself, acquiring new licenses and technologies (Unity 3D, and some sound software, etc), legal fees for contract negotiation, payroll taxes, health insurance, site hosting, all that stuff.
  • Things were going so well during this period that Arcen was able to pay me for a little under 1/5 of the hours I'd put in for AI War in 2009; I'd never intended all those hours to be for free, they were just extremely deferred.
  • Profits would have been around $8,000.00 higher, but one regional publishing partner [for AI War] was slow in paying us, and eventually declared insolvency. That was a moderate blow, but with Tidalis nearing completion I figured it was water under the bridge.

In general, in early 2010 Arcen was hitting a new level of being. Popularity was up, profits were obviously way up, and things were going better than I ever expected. I'd expected to struggle more. We did run into some bumps with having to upgrade a number of our assets, from our development technology to our website hosting, to our increased legal fees for dealing with the regional publishers, to our need to actually buy some used Mac OSX laptops so that we could test the Mac versions of Tidalis (and, as it turns out, develop them at all - long story).

In short, we had around $20k of expenses that were basically one-off growth expenses during that time, which we're still seeing the benefits from now, and which make our net profits lower than they would have been in the second two quarters of this year [if our income had included at the same level].

After April is when things started to change. I sensed it coming, so I'd cut my salary, royalty payments were down, and we weren't acquiring a bunch of new assets or doing any contract negotiation. As far as income goes, looking at my info on that, I see that we made, month by month:

  • May $19,085.06 income, $12,684.28 expense
  • June $17,233.70 income, $26,434.89 expense (a fair chunk of this was taxes)
  • July $6,956.48 income, $18,243.52 expense
  • August $13,847.02 income, $18,682.68 expense
  • September $13,395.42 income, (not full data yet, but expenses looks to be about the same but a little lower than August)

Bear in mind that our income is on a 30 to 90 day delay, depending on the partner paying us, so we have a lot of forward visibility into what future income will be. The figures above are all based on when we received the money, [not when the money was earned] . So far, [projected] earnings for October looked to have been about $5,000.00 at best [this figure has improved since Chris went public with his company's troubles]. Also, the September income also includes the part that went to Child's Play .

Also bear in mind that, while those figures for August and September might look pretty okay, they are actually based on both AI War and Tidalis. So, as you can see, when you look at them, the AI War numbers dropped in July and stayed constant at this new, low value. Then the Tidalis numbers came in and were only about the same, bringing the overall totals up to what you see there.

That, in turn, might not look like all that much lower than our actual income from before, but bear in mind that part of what we set up with Tidalis was royalty arrangements with staff who worked for no immediate money during the Spring. This means that the amount that Arcen actually directly brings in on Tidalis sales gets split back out to the staff immediately, minus about 40% which goes to the longer-term Arcen reserves.

These guys worked really hard - trust me, it's fair from Arcen's perspective. It would have worked out well for everyone if Tidalis had sold even just as well as AI War. Given the larger appeal of a game like Tidalis, we figured that 30k units sold would be trivial, whereas it had been an uphill battle with AI War. That said, it wasn't like we were sitting around waiting for it to sell hundreds of thousands of copies - there's a lot of luck involved in that - but we had all the business partners and positioning we thought we needed to earn back what we spent on making it, to give the staff fair income from the royalties, and then hopefully to have an ongoing small stream of money from it for at least a while.

You know how that's turned out so far.

In summary, how would I characterize the costs of making AI War or Tidalis?

  • I suppose I'd say that Arcen invested about $80k directly into AI War, plus about another $70k of unpaid time on my part, and we then made about $180k on it before taxes and expenses. That had still been counting up until recently. It was a very successful game, and only becoming more so, financially speaking.
  • I'd also say that we invested about $25k in general stuff for the business. This is stuff that's benefited Tidalis first, but which is also what is powering all the improvements behind AI War 4.0.
  • For Tidalis, I'd say that Arcen spent about $50k or $60k directly to make it, but that's ignoring another $50k or so of hours for which staff received royalties instead of immediate pay. (At their request, not my insistence, I should add - they wanted to share in the risks and rewards more directly, after seeing AI War, while at the same time being more full partners in the development of the game). At any rate, I say $50k to $60k because it's hard to quantify exactly, as some of us were splitting our hours between AI War and Tidalis work.
  • And then of course there is taxes. For 2010 the payments so far have been about $32k, with another $7k due later in the year.

All in all, that leaves us with about $15k sitting in our bank account at the moment, whereas I'd at one time had our buffer up to $60k, and in general had kept it north of $30k since last October, until everything fell apart.