I think hybrids will win out. Fuel cells are indeed WAY out in the future and may not offer improved efficiency over a high-output IC engine. I think a committed effort to produce an "extreme" hybrid -- flywheels, free-piston IC linear alternator, connected in series to power-dense DC brushless motor(s), will beat a conventional diesel due to no idling and regenerative braking, as well as no transmission, crankshaft, or connecting rods. The very high torque and short-term power of the electric motors will easily beat the performance of an IC diesel, and the linear alternator can be run at peak efficiency at all times, and is possibly up to 20% more efficient than a diesel anyway (runs at at least a 30:1 compression ratio, for instance). I belive a test alternator runs about 500cc or less, with an output of 40 kW, and a total weight of 60 lbs.
The point of a hybrid is to reduce the power needed from the engine (APU), and thus its weight, and to use some sort of power storage mechanism for peak power needs such as accelerating, passing, and hill-climbing. I don't believe existing schemes to do this, as exhibited by the Prius and Insight, commit to this philosophy to a sufficient degree to make them viable. Basically, their engines are too big, their electric motors are way too small, and their batteries are far too heavy and not power dense enough to serve their stated purpose of high-power bursts for short durations.
To do a hybrid right, I believe you have to put your money where your mouth is. That is: one of the major reasons, if not the major reason, for a hybrid, is to utilize the fuel-consuming APU at maximum efficiency at all times. A parallel hybrid can't really do this -- electric motors can "smooth," but not eliminate, fluctuating power demands, and the engine must spend a considerable portion of its time running at non-optimum speeds and outputs. An "extreme" hybrid commits to this principle, and runs the APU at peak efficiency at all times, or not at all.
The only way to make this work is to have a power storage system capable of handling very high power demands for short runs, capable of near-instantaneous, low-loss recharging from engine and regenerative braking, and able to handle an almost infinite number of deep discharges. This power system must also be fairly light, or it will defeat the purpose of the hybrid. I am pretty sure batteries will never, ever fit the bill. Their energy density is decent, but their power density is really poor, and their charge/discharge times are just unacceptable for a vehicle application.
Clearly, I am a flywheel enthusiast. Flywheels can deliver essentially unlimited power for the durations needed for almost all small vehicle applications. They recharge almost instantly, can discharge very rapidly, and can withstand essentially infinite deep discharges. Outfits like AFS Trinity, Lawrence Livermore, FESI of Ottawa, Canada, and the University of Texas have already built and tested flywheels of a suitable size and power output for hybrid vehicle applications, complete with sophisticated electronic contols for managing power input and output. Coupled with Sandia's free-piston HCCI linear alternator, and something akin to UQM Technologies' INTETS DC brushless motor, you have, in theory, a very high-efficiency series hybrid with performance significantly better than most sporty cars presently on the road, with twice the mileage and half the emissions.
The remaining problem is the body and, more significantly, flywheel ballistics containment in the event of a rupture caused by overspin or an accident. This is not trivial: particles of carbon fiber and epoxy fly off a flywheel in a rupture event on the order of Mach 5 -- a single flywheel sufficient to power a hybrid vehicle contains enough energy to blow a Corolla-sized car a mile straight into the sky.
And yet, a gas tank full of ten gallons or so of highly refined, volatile gasoline is a pretty scary proposition, too. You can do it wrong, and produce an unacceptable risk, as in the Ford Pinto -- or you can do it right, such that your car will probably not explode unless hit by a train.
In doing it right, it sure seems to me that enough smart people working on the problem can nail it down. We have Kevlar. We have lots of ballistics containment folks in military pursuits. And we have Burt Rutan, who knows enough about lightweight composites and structural engineering to build a plane that flew around the world without refueling, and to build the GM Ultralite concept vehicle, far and away the lightest, strongest, and stiffest monocoque car body ever created.
An extreme commitment to a true hybrid, I believe, will produce something truly remarkable, rather than the half-hearted and ultimately doomed efforts we see now. The thing must be very light, and very strong, both to handle flywheel containment and to take advantage of the true hybrid concept of light weight and high short-term power. This means a composite monocoque body, probably with flywheel containment integrated into the design for increased strength and stiffness. It means a commitment to the original vision of a hybrid -- run the fuel engine at peak efficiency at all times, and use only electric motors for propulsion and regenerative braking. This committment gives the added benefit of doing away with the need for a multi-gear transmission, as well as connecting rods and crankshafts -- the linear alternator uses neither of these. Moreover, it uses the innate capability of electric motors to generate very high low-end torque, which they do much better than all IC engines.
While this setup is designed for efficiency and low emissions, it would seem to me to have an added, and perhaps unintended, benefit: namely, really incredible performance, particularly in terms of acceleration. A commitment to light weight could result in a race car of perhaps 1,000 lbs, with perhaps 300 hp for ten seconds or so. Distributed to four wheelmotors, using automatic torque control to prevent burning rubber, that should result in 0-60 times that are, well, neck-snapping. The world record for autos now seems to be under 2 seconds -- this hybrid, applying all the smooth power and torque its tires can handle, and weighing about a third as much -- could very conceivably beat that. Handily.
Granted, it won't win LeMans, or Indy, and it won't pull very heavy loads up long, 6% grades in the Andes. But it WILL turn heads, in the most elemental of high-school teen fantasy kind of ways -- by taking off from a stop sign like a rocket and blowing the doors off of anything and everything misguided enough to try to match it. To me, there is little point in this -- the real world doesn't need rocket acceleration. But gearheads seem to like it, and some gearheads, like Jay Leno and Paul Newman, like it a whole lot and have a whole lot of surplus cash to indulge their fancy. Ditto many testosterone-charged CEOs who would welcome an extremely fuel-efficient and envirnmentally friendly car (great PR!) that just happens to have world-beating performance, at least in the world they actually live in -- tearing out of stop signs and clobbering Ferraris and Lamborghinis going up short climbs on the coastal highway.
That is to say, certainly, the research and development costs of an "extreme" hybrid are prohibitive -- in the short run. But much of the whole point of racing and making a splash in the automotive press is PR and image. You make extremely expensive, extremely high performance cars to tickle the imaginations of your buying public, and sell them vastly cheaper and less high performance vehicles of the same name that they can actually afford. By going after the extremely high-performance, expensive market, you build up demand for your product, instead of relying on committed environmentalists willing to put up with poor performance, high cost, and high maintenance expenses, as is presently done with the Prius and Insight. A successful hybrid can't make such extreme compromises -- it must be composite, it must use flywheels, it must have a radical, experimental IC engine, and it must commit to big, powerful electric motors. Way too expensive now, but not so when people begin packing the wailing lists for the amazing screamer that everyone's abuzz about -- 0-60 in 1.5 seconds and 200 mpg. THAT is why hybrids will win.