Yes, and that is the error. The number of copies is not static, and might as well be considered infinite. With digital goods there is effectively no limit on the number of copies which can be produced at near-zero marginal cost, so cost per copy is a meaningless metric. Unlike physical goods where you need to recoup your variable costs for materials, labor, and capital on each item sold before you can even think about offsetting the fixed cost of R&D, every sale of a digital good contributes to the bottom line regardless of the price.
The single most expensive game to develop to date was apparently Star Wars: The Old Republic at $200 million[1]. If each of the 7.7 billion people on the planet had a copy, however, the development cost per copy would only be about $0.026, affordable even for the most limited budgets. Say that 5% of the population would be interested in trying it out at a price of just $0.25 per copy: that's $481M in revenues, or $281M profit. Or they could choose to make it a monthly subscription service at prices attractive to just 1-2 million players and lose money overall while charging subscribers $100 apiece. Same development cost, different distribution models. There is no predetermined cost per copy below which they are selling at a loss and above which they are selling for a profit.
[1] https://en.wikipedia.org/wiki/List_of_most_expensive_video_g...
This isn't pedantry, this is discussing the actual topic of the conversation.
It’s perfectly valid to talk about overall profitability when accounting for non-marginal costs, but that’s not the phrase you use to talk about it, and saying that some units are sold at a loss while others are not is nonsensical in that context.