I have to take exception to the notion that an AED on a dive boat is not much of a deal. It's long, because I think it bears explaining.
The reality is that, in a medical situation away from immediate advanced life support, such as on a dive boat, where an AED is useful, CPR is not a substitute and rarely even an effective stopgap. Note I said where defibrillation is medically indicated. In other words, when conditions are such that defibrillation can normalize cardiac rhythm, if you don't have that capability very quickly, the chance of a good outcome is very poor. Cardiac arrest produces an array of bad effects, mostly chemical, that become more difficult to reverse or even to precisely address without elaborate tests the longer the arrest persists. CPR does not effectively stave off those effects. The patient becomes and remains very, very ill. Drowning may produce additional problems of metabolic imbalance.
Defibrillation is useful to convert ventricular fibrillation. V-Fib is a gross, disorganized, random firing of multiple nodes. If you could see the heart during V-Fib, it would just quiver, and no blood is moving. Defibrillation is intended to induce a brief period of no electrical activity that hopefully will allow one of the cardiac nodes to take up a normal, rhythmic rate. Optimally, that would be the normal atrial node, so that the impulse propagates through the heart in the correct direction, but even other nodes lower in the heart can sustain life with reduced output. Since typically young, relatively healthy people who have suffered interference with their air supply are the patients in dive accidents, and this class of people have undamaged hearts, successful defibrillation will most often restart from a normal atrial node, and normal cardiac output will rapidly return. (Some effects, such as loss of blood vessel tone may take longer to recover from, but decent working blood pressure is likely.)
The ongoing metabolic imbalances induced when the heart is not moving blood also make it more difficult to terminate fibrillation. The longer the blood is not moving, the more the patient deteriorates and the more difficult defibrillation become. One factor is simple exhaustion of the cardiac nodes. Successful defibrillation is most likely when the electrical activity, disordered as it may be, is vigorous. One commonly observes the amplitude of traces on an ECG to slowly diminish as the incident progresses without success. When full advanced life support is on scene, a variety of drug interventions can be undertaken to achieve or maintain a "shockable" rhythm. The movie scene of using a defibrillator on a "flat-line" (asystole) is bogus. You need fibrillation in order to defibrillate.
From one study, it seems likely that at least 29% of drowning victims with cardiac arrest present with ventricular fibrillation and some 55% in asystole. And 16% with a heart rate so low as to be ineffective. In talking about AED's, we are talking about those 29% and any of the others who may develop V-Fib at some time during the incident. The best and only real chance of a good outcome is with defibrillation in the first minutes. Patients in V-Fib are routinely converted multiple times in hospitals with little bad effects, and that is often in patients who already have sick hearts. Patients prone to V-Fib may be converted frequently by their own implanted defibrillators. But chances of survival are dismal after more than a few minutes without advanced life support and not all that wonderful after many minutes with ALS. Do you recall reading of a dive accident where the victim required CPR for a cardiac arrest and was successfully converted and survived after a boat trip back to shore or when the airborne medic arrived 20 to 40 minutes later?
Obviously, there's no real expectation that any single dive boat will ever have to deal with a cardiac arrest. Whatever the real number of dive fatalities may be, given that there are an awful lot of dive boats making a lot of trips, they don't set out each day expecting it to happen. But our experience with AED's since they've been places in arenas, offices, airports, and on airliners makes it clear that it's 100% certain that there is some number of dead divers who would have lived had early defibrillation been available. I don't know that cost analysis in the developed nations is even necessary. An AED is 1,200 USD. For comparison, a DAN two bottle oxygen outfit is 750USD. There are places where I don't expect to find an AED due to cost. There's a lot of stuff I don't expect to find in those same places. But I would like to think that a dive operator with the resources would choose to carry the only thing that might save someone.
A dive operator considering purchase of an AED should always inquire of the additional costs and responsibilities. Most reputable AED dealers provide information and medical direction and keep track of grant opportunities. DAN has (or had) a grant program to cover half the purchase cost for operators. The program places three AED's per year. Failure to have an AED can be a liability issue. (At least one gym had to settle a very large suit on account of not having an AED when a client died on site. Another gym has been sued multiple times and has settled most of the suits with payments.) Medical control is not always required. (Florida has no pysician oversight, for instance.) But laws vary among the states:
http://www.americanheart.org/presenter.jhtml?identifier=3024006