Telling the future is hard, but ABK's profits have historically been $4-6B. At that pace it will take about 15 years to profit what Microsoft has paid for ABK. But, ABK still has a value. So if they choose to keep them for 5 years, they could sell them for 50B and still have made a profit.
Paying a special dividend doesn't have a potential for profit. Making an investment does. And buying a thing (as opposed to spending it on a dead end research project) is a fairly safe investment.
Keep in mind that the ABK entity itself has value (as you've noted). So for people not paying attention,
if, for example, ABK is worth say 60 billion USD, then at 4-6 billion USD a year, you'd break even in about 2.5 - 3.5 years. Of course, the value of something could appreciate or depreciate over time, so the calculations are never that simple.
Basically, they key here is that Microsoft isn't losing much if any "value" with this purchase. It's not likely they are taking out a loan in order to finance the deal with interest on the loan being a deficit that has to be made up.
A lot of people fail to take into account the value of what is being purchased.
As well, it's important to note (as you did) that this is an investment that pays dividends to both the corporation (potentially increased revenue and profits) as well as shareholders (potentially higher stock value, potentially larger dividends in the futures).
As noted, is there a better way to invest 70 billion USD? At a profit of 4-6 billion USD a year, that'd be a yearly ROI of between 5.7 and 8.6 percent a year.
Also, it's important to note that Microsoft's plans for ABK include expanding the market for their products onto other consoles and devices. That potentially increases future profits generated by ABK above and beyond what they have historically generated. If that happens, then the ROI gets better and better.
Of course, as with everything, nothing is guaranteed. It's certainly also possible that ABK could start failing in a massive way after the acquisition.
Regards,
SB