Because the pundit class in America is related to so few people who want to enlist in the military, there's negligible media awareness of how hard it has become to join up. A major hurdle is scoring high enough on the AFQT cognitive test. 

The Pentagon isn't in any hurry to make its intelligence requirements explicable to the media.  The conventional wisdom is that intelligence testing is a racist hoax or it just applies to academia, not the real world, or whatever. The fact that the military is obsessive about cognitive testing is something that simply isn't in the reigning worldview, and the military is fine with that. It likes testing and it dislikes outside interference, so the more convoluted its jargon for talking about its intelligence requirements, the better.

For example, the entrance exam is, in one sense, the ASVAB, a 9 or 10 part 3-hour test. But a 4-part subset of the ASVAB called the AFQT determines whether you'll be allowed to enlist or not. (The non-AFQT ASVAB subtests influence assignments, such as to vehicle repair.)

Are you losing interest in this topic already as you try to keep ASVAB and AFQT straight? The military doesn't mind if outsiders are baffled and bored. In fact, it kind of likes it that way. And if potential recruits can't keep this stuff straight in their heads, well maybe they aren't military material.

The AFQT is a verbal and math test kind of like the SAT or ACT. AFQT scores are so highly g-loaded that they are pretty much interchangeable with IQ scores on a non-culture free IQ test like the Wechsler, according to a retired head of psychometrics for one of the major branches of the armed forces whom I interviewed at length in 2004. Much of The Bell Curve was based on the military's AFQT data that was normalized on the 1979 National Longitudinal Study of Youth. 

The Wikipedia article on the ASVAB gives the AFQT minimum scores to enlist as of December 2012:

AFQT scores are not raw scores, but rather percentile scores indicating how each examinee performed compared with all other examinees. Thus, someone who receives an AFQT of 55 scored better than 55 percent of all other examinees. Maximum possible score is 99 as a person can do better than 99 percent of those who took the test, but he cannot do better than himself, so the high percentile is 99.

From Wikipedia:

Standards for enlistment

AFQT required minimum scores for people with a high school diploma as of December 2012 (unless otherwise noted) are as follows:

Also, if you get a GED and complete a certain number of college credits, that lets you use the HS Diploma column. 

So, the lowest percentile you can get into the military with is the Marines at the 32nd percentile (if you have a high school diploma, plus the Marines have plenty of physical and other requirements). With 315 million residents in the country, 31% percent aren't smart enough to join the Marines, so that's over 97 million.

But, with the recession and the winding down of the Iraq meatgrinder, the military now often won't let in kids who just barely make the minimums.

So, the current situation is actually worse for decent young people with 2-digit IQs than the Wikipedia table suggests. 

If you look around enough online you can find PDFs of statistical reports for the government that add even more insight. For example, the Congressional Research Service reported that the military branches have both quantity and quality goals in recruitment. The basic Department of Defense quality goals are that 90% of recruits have high school diplomas and 60% score above average on the AFQT (i.e., have a 3 digit IQ).

In FY 2011, all branches met their quantity goals and exceeded their quality goals. The Army had 99% high school graduates (i.e., not GEDs) with 63% scoring above average on the AFQT. The Marines had 100% grads with 72% scoring at the 50th percentile or higher. The Navy had 99% grads with 89% scoring above average. The Air Force had 100% grads with 99% above average in intelligence ("this would represent the highest “above-average AFQT” accession cohort of any service since the inception of the All-Volunteer Force in 1973," according to the CRS).

So, 150 million people, maybe more, couldn't join the Air Force in 2011 because they aren't smart enough. The Air Force is now Lake Wobegon and the other services are trending in that direction.

The various Reserves and National Guards are also exceeding their quality benchmarks.

The AFQT scores are normed versus the National Longitudinal Study of Youth 1997, the follow-up to the NLSY79 highlighted in The Bell Curve. I haven't looked into, but perhaps there is a Flynn Effect going on, which would making scoring a little easier than it was a decade-and-a-half ago. Also, people who want to join the military no doubt test prep on ASVAB/AFQT, while I presume the NLSY cohort didn't. So, the IQ hurdle to enlist is probably not quite as daunting as these numbers suggest, but still ...

I knew a kid who was working an McDonalds, dealing a little weed, and then he resolved he was going to turn his life around by joining the Army. He started working out, developing a good attitude, and he impressed the recruiters. But he flunked the AFQT. The recruiters liked him so much that they sent him to a six-week AFQT boot camp where they lived in barracks, wore uniforms, followed military discipline, and studied to pass the AFQT. At the end, the day before they all took the AFQT, the sergeants picked this kid as the best example of the military virtues in the camp.

And then ... he still flunked the AFQT again.

In summary, every effective institution in America works hard to to select better people. But, the fake "immigration debate" going on right now has ruled out all discussion of just what is the quality of illegal immigrants.

Moreover, we have a whole bunch of our fellow American citizens who aren't of the cognitive quality currently necessary to fight for their country. Shouldn't we be worrying more about what kind of living they'll be able to earn before we care about solving Mexico's problems?