As with all such pipeline relationships, there is a tension between YC and Super Angel world on the one hand and the VC world on the other hand. In business there are rarely outright wars down a vertical or horizontal integration path (this one is "vertical"). Instead you get a mixed war: cooperation on some fronts, competition on others.

Cooperation-Competition Dynamics

Some dealflow behaves as though the YC-Superangel-VC pipeline is all harmonious, with everybody in love with everybody else. Enter the queue via YC, take your 150k from Yuri Milner, get some other Super Angel cartel to take you up to about $1 million, and then let the VCs take you the rest of the way to the approximately $20 million that, conventional wisdom (i.e., Fred Wilson) has it, it takes to get to an exit.

Last I heard, people thought that that kind of harmonious dealflow, with all interests aligned, would account for maybe 30% of YC and similar program alums. That means 70% will be stranded without growth capital at around the $1 million mark. I think Paul Graham wrote at some point that the $1-2mm range is the valley of death: too big for Super Angels, too small for VCs.

Of the 70% that the VCs cannot absorb, lacking any other reliable indicator, let's pretend that 35% is junk that should die and is allowed to die, and 35% is good stuff that deserves to be seen through to a decisive market outcome.

That latter half creates the "early exit" game.

That's where the tension comes in. YC and other seed-stage pipelines have to place bets: do they develop their programs for early exits post-Angel, or for the 5-year/20mm long-haul path? It's basically a question of which markets to serve. YC and similar programs have to basically decide whether to serve the Super Angels or the VCs.

The Super Angels, to a lesser extent, also have a decision to make: optimize for quick flips/early exit, or a feeders to the VCs at Series A and beyond. The feeder relationship is somewhat fraught with tension because the value of the Angel stakes will depend on how the VCs drive the company once they take over funding. So there are now uncomfortable bucket-brigade type dynamics starting to appear where old investors at various stages are bought out by new ones, etc. This is because the power does not yet reflect value of contribution. Super Angels are starting to (correctly) believe that follow-on VC funding does not pay off for them at levels commensurate to the risk they squeeze out. They want more. They may invest much less (like 5%) than the VCs over the lifetime-to-exit, but they also take out far more risk than the VCs like to admit. In fact, they squeeze out so much risk that VCs can enjoy the luxury of investing for growth instead of investing for traction.

Optimizing for Early Exits vs. "Going Big"

So the overall answer is that the YC world is not yet seriously disrupting the VC industry, but are in a position to do so, by tuning their programs to produce more/less early exit type outfits.

Early exits being often talent development or "feature, not product" efforts lead to programs looking more like University summer school type programs catering to a talent acquisition and innovation-outsourcing market. "Full startup" programs will look less like universities and focus more on serious market risk management instead of talent development.
Another way to think of it is as follows: should YC type programs attempt to squeeze out technical and talent risks or market risks. If they do more of the former, they'll be throwing their lot in with early-exit favoring Super Angels (and may indeed, merge with them, welding the pipeline; quite possible because attrition can be pushed down if market risk is avoided). If they do more of the latter, they'll be throwing their lot in with VCs.
I don't track this stuff too closely, but if you're interested, keep an eye on the average age of the accepted applicants, and event programs -- see if the programs focus more on personality skills and technical/product mentorship than on providing decisive market-risk-mitigating advantages (and I don't mean crap like training in A/B testing, I mean things like privileged relationships with big acquirers, or proprietary marketing databases). Maybe it's already become clear which way PG & co. are leaning.
The Prediction

Long-term, I suspect they'll lean towards early exits. Running a de facto education program, glorified and branded as an entrepreneurial program, is simply much lower risk, given the increasing reliance of big companies on acquisition for talent and IP. There is also a lot more volume in that market.
The "full service" VC path will likely diminish slowly. A lot of people are starting to think that a capital market based on a 10 year cycle, and the premise of a couple of 10x exits determining the overall return, is getting too shaky. Too few people are putting too many eggs into too few baskets based on too little data.
Data to mitigate risk is getting cheaper by the moment, and the smaller-scale investors seem to be better at using it.
The final factor is the entrepreneurial community itself. One effect of YC has been that the type of people entering the pipeline is increasingly young, inexperienced, rent-and-ramen types whose main investment-worthy attribute is their low cost, malleability, energy and optimism. The strategic thinking and careful consideration of risks is increasingly shifting from the entrepreneurs to the investors. They may still have CEO titles, but they are increasingly acting like grad students whose advisors play conjuring tricks on them, making them pick the cards they want picked, like canny PhD advisors.
This brings up an interesting question. If the early-stage world optimizes its operations around early, small-multiple exits, and the VC world is losing the competence to fund the 20mm type paths to 10x exits (as is increasingly the case), where exactly are the "next big thing" type companies going to come from?
Who'll do the Big Ideas now?
There are two possible answers.
The first candidate is bootstrapping. Today, bootstrapping is considered primarily a lifestyle business strategy, but this is not necessarily all it can do. You can Go Big with bootstrapping. The difference is that you do so on your own timeline, by getting inside the market tempo, rather than on a rigid timeline of 5 years or whatever.

Two of the biggest Robber Barons, Cornelius Vanderbilt and John D. Rockefeller, built their vast empires while mostly avoiding the world of professional institutional investors. For Vanderbilt it was because the capital markets and investment banking industries had not yet matured. For Rockefeller it was simple suspicion of people like J. P. Morgan, so he preferred to use loans, leverage, cash cows and other mechanisms (down to controlling entire banks) to grow his business. Late in his career, he could actually throw his weight around a lot more than J. P. Morgan himself, even though the latter is viewed as the banker of the Gilded Age.

2012 is looking increasingly like the Robber Baron era. Google sounded the first warning when it chose to do its IPO using the unique auction format. That sort of thing is going to increase, as players who really want to go BIG start to avoid the professional, institutional capital markets (of which the VC market is one little outpost). Also look at how Facebook and other companies have managed to delay taking investment and IPO decisions by using secondary mechanisms.

The second candidate is more depressing: nobody. There may be no more big, bold plays for a couple of decades, as the economy gets restructured around the Internet.