Civilization could be facing a big problem in the next few years. Oil could get considerably more expensive soon.
"Peak Oil" is not when we "run out" of oil; it's simply when our collective ability to extract enough oil is exceeded by our demand for oil. There's disagreement about when it will happen; pessimistic estimates say it could be in the next few years, while official estimates put it a couple decades out. Some people, including Alan Greenspan, say it won't happen at all: that new technology will create new supply from non-oil hydrocarbon reserves like tar sands and oil shale.
In evaluating these opinions, it's worth considering that global demand was within 1% of supply a couple of months ago. Demand keeps growing, while the oil production rates of most countries are already maxed out. This doesn't look like a very good safety margin. Another clue: In 1998, a report by the International Energy Agency listed "unidentified unconventional" oil as supplying 17% of the world's needs in 2020. In other words, they didn't know where it would come from.
For an overview of the Peak Oil argument, see Jack Zagar's presentation, especially charts 10 (vs. 9), 14 and 15, and 23. (It looks to me like Zagar is a bit pessimistic, as (if I read him right) he backdates advances in extraction by assigning the total oil in a field to the discovery date. But it looks like he's right that the standard discovery curve is far too optimistic.)
If oil supply ever does fall below demand, we can expect prices to rise steeply. At this point, it could take years for alternative technologies to come online, no matter how much economic incentive there is--and meanwhile, since oil demand is relatively "inelastic," the price of oil will be bid up until it slows the global economy enough to reduce demand. That's an ugly picture.
So how does this relate to molecular manufacturing? Well, to avoid a "Peak experience," some new technology will have to come online and grow quite rapidly. It will have to support rapid research and development (meaning, rapid prototyping of industrial-scale projects). Then it will have to support rapid trillion-dollar-scale building of infrastructure.
How rapid is "rapid"? It depends on how far in advance people are willing to admit they need a new technology. It's worth noting that after journalists noticed the 1998 "coded message" of scarcity the IEA reports transferred the "unidentified unconventional" oil to other categories. And in 2000 the USGS published a report (see "Chart 31" of Zagar's presentation) which appears to grossly overestimate the yet-to-find oil; the "Mean" curve is a sharp departure from history.
If corporate and national politics delay admissions of trouble until the signs are unmistakable, or if the Peak Oil camp is even close to right about a peak as early as 2008, then we could have just a few years in which to design and ramp up a new trillion-dollar infrastructure. The only way to do this without severe economic pain is large-scale, general-purpose, exponential manufacturing. And the only technology that appears to fit the bill is molecular manufacturing of the mechanosynthetic variety.
This appears to be a fairly strong argument for early development of advanced molecular manufacturing.