As a percentage of potential revenue, costs have gone down.
It doesn't matter if per team costs have tripled. If your potential revenue/expected sales have outpaced that, your costs have actually gone down per unit. The only thing that went up is your risk. If you miss by a wide margin now, it will hurt more.
Companies are not shoveling more money into the fire for a lesser reward. They are releasing fewer risky games and shooting for games now that can turn a good profit, even if they come in under sales expectations, via microtransactions. They are making 15% more per digital sale, even on platforms they don't own, which is a massive chunk of change spread over millions of copies. They make 45% more if they own the platform like Origin. The large publishers are doing very, very well.
The narrative that they need these MTs to survive is false. They certainly don't need greedy ass monetization models like the one attempted with BF2.
People keep saying this but I've never seen any proof, and it's directly contradicted by just about every developer whose talked about it. The problem isn't something you can boil down to "development costs have tripled, but the market has increased by five times, ergo costs have gone down." It doesn't work like that. We have a few facts. Costs have skyrocketed, in all dimensions of making games. The market has also expanded significant, in all genres and at all levels. The problem is costs cannot be calculated on a linear level, and markets are different in every genre and at every level. At the same time as it is much easier to make a lot of money, it is also much more difficult in different ways and much riskier. As a result you're not really defining costs properly. What's important is how much you have to sell to break even, and that number is much much higher than it used to be. While the market has increased, the value brought back by each sale is proportionally less than it used to be, because of inflation and higher costs. A game like Battlefront (2015) probably cost in the 150-200,000,000 to make when all costs are calculated (Star Wars license, development, marketing, testing, production), and EA defined a success to be 13 million copies. Even today, that is an astronomical amount of copies, and with the lukewarm reception of the game EA must have been very worried that the second would be a similar success. If you do not raise the price of games, you need recourse and additional sources of revenue.
I don't think you're wrong, I think you're just defining some of the concepts improperly. Risk is the most important factor here, because although the "reward" for a success is much bigger than it used to be, the risk is equally larger. The stagnation of prices has a lot to do with them, as if games have followed inflation, the amount a game would have to sell to be a financial success would be a lot less. The problem is now that games haven't had incremental increases, so raising the prices now is very difficult.
Developers don't need microtransactions to survive at all. But they are made more prolific by the costs in the industry.