It's only fair that I insert a disclaimer here. The cost effectiveness studies I've seen have not usually gone down to 23 weekers but have included slightly more mature babies. If anyone knows of cost effectiveness studies encompassing 23 weekers, I'd love to see them.
The second issue raised in the comments was the idea that, instead of spending money on NICU costs, we should spend more money on prenatal care so we can prevent premature births and NICU costs. This is a very appealing idea and a darling of the public health community. Unfortunately, though, it's probably not true. The data supporting the notion that prenatal care saves subsequent postnatal dollars is pretty scant. It makes sense, too, that it is not true. We don't know how to prevent preterm birth. If we did, the rate of preterm birth in the U.S. would be falling, but in fact it is rising. More than 10% of births are preterm now. Don't get me wrong: I think it's good to provide prenatal care, but let's not subscribe to the too-good-to-be-true idea that every dollar spent on prenatal care will save five dollars in NICU costs. (For reasons not clear, that concept always reminds me of Wimpy in the Popeye cartoon: "I'd gladly give you a nickel on Tuesday for a hamburger today.")
If you want to think about whether NICU dollars are justified, here's some food for thought. What if we took money spent on NICU care - about $2,ooo per day per baby - and spent it on food, clean water, immunizations, and medicine for kids in the developing world? There, the money probably could save more lives. But that's another big subject in itself, so I'll stop here.
P.S. If we're going to send NICU dollars to developing countries, we'd better send dialysis and coronary bypass dollars there too.