Carbon composites aren't necessarily my day job per-se, but take up about 50% of my research here.
Carbon fiber products have changed a lot in the last 20 years, but it's not the actual carbon fibers that have changed. In fact, they have changed relatively little. The changes have happened in the theroset resin systems and the effect on layups that simulation has had. I'm highly doubting that many bike companies do much with finite element modelling on their layups, I'm sure it's mainly empirical testing.
Given a perfectly produced carbon fiber composite structure with a modern resin system that has been properly stressed (i.e. layer orientation management) , you can expect a carbon tube bike chassis to be about 3x as stiff as a typical aluminum chassis.
By weight alone, a bidirectional carbon tube will be just as stiff if not slightly stiffer than high grade 4130 steel. The advantage here is that you can make the bike much lighter, as the stength is far greater than aluminum or even titanium.
The problem that you see with people breaking carbon bikes is that the companies end up going lighter and lighter, fewer and fewer layers of carbon ... and the bike is still stiffer than an aluminum equivalent but prone to failure. The best advice here is really to just go with a trusted company that has experience here. Modern resin systems are incredibly tough, and any good manufactured should be using a prepreg resin system with adequate toughness and additives for UV resistance and galvanic corrosion resistance. Carbon in contact with aluminum (which is what the bearing bores will be made from most likely) caused a galvanic reaction. It is an incredibly slow galvanic reaction compared to what you normally think of, i.e. zinc and whatnot.. but it is there, and therefore the use of a bonding epoxy between the bearing bores and the carbon that resists that is necessary if you want to be riding the same bike for more than 5 years. Lotta guys out there making their own bikes aren't doing this.
Another issue is again with the bonding between the chassis and the bearing bores. A lot of failure come when the bores are bonded in improperly, usually because the carbon and aluminum interface surfaces are not prepped properly for a good bond as required by most epoxy systems (most use a Henkel product like Loctite Hysol 9460). This is another reason to stay away from the cheap stuff... since a lot of this stuff is hand made, you can't be sure how well they're prepping the stuff.
All the top bikes are surely prepreg laminates, but some of the small company stuff can be a traditional 'wet-layup' bike... starting with pure bi-directional fibers and wetting out by hand with a two part resin system (Aeropoxy or West Systems are common names). More sophisticated methods are by vacuum assisted resin transfer molding, or VARTM. The key to strength and stiffness with wet methods is an appropriate volume fraction, or amount of resin compared to amount of fiber, and also the resin distribution in the fibers. In most cases, a wet layup bike will not be able to be as light nor as stiff as an aerospace prepreg equivalent due to the inherent inaccuracies in the resin delivery system. The use of computational flow solvers in recent times have made manufacturing VARTM components with properties close to that of prepregs possible for a cheaper price, yet still... I doubt any bike companies are using such simulation. It's not something that is particularly easy.
I digress. The real thing you need to know is that a good quality bike from a solid reputable company should be able to perform just as flawlessly as any old steel or ti bike. It starts to come down to rider preference. The one thing about the carbon bikes is due to the stiffness increase, the natural frequency of the bike is several times greater than aluminum and the damping characteristics of the material are very different. On a fixed suspension road bike, this appears in the form of a slightly harsher, less damped ride. This would be barely noticeable to any joe schmoe, but an experienced rider will feel it.
Fatigue in the carbon itself is a non-issue. A correctly designed carbon fiber structure will literally never fatigue if the loads it sees are kept within design limits. A stress strain curve of a carbon laminate is nearly a straight line. There is not clear yielding and plastic deformation like any engineering metal exhibits. Fatigue in the resins used to be an issue, but not with any modern resin system. For bikes that are not painted, just clearcoated, be damn sure that the resin system has a UV inhibitor or else sunlight exposure over the years will age the resin (resulting in a yellow-hue to develop) and can weaker the outer surface slightly. This usually isn't a structual problem on say, carbon fiber formula car monocoques due to the sheer thickness of the material, the outer layer is all that is exposed. But on a road bike that has as little as 4 layers of fiber thick in some areas, the UV permeability can be a problem.
So, that's my technical opinions. It should be noted that I'm not a bike rider by any real means anymore, so there may be more that you would want to focus on when purchasing. Such as the price tag. And the color.
