Any industry where the capital costs are low, or where a small group of niche specialists can add a significant multiplier to the value chain are candidates for this sort of effect. As more industrial tasks are farmed out to large companies that specialize in flexible production it is possible that you will see this show up in industries that are currently have strong vertical integration.
For example, ten people can't mass-produce a car. Ten people can provide spcialized bodywork and personalization of a mass-produced car. Ten people could probably design and oversee the outsourced manufacturing of discrete components in the car. Ten people could probably manage a company that deals with other small firms to create bespoke vehicles based upon mass-produced frames and engines. As you increase the specialization and make more parts interchangeable it is possible to decentralize and distribute a production chain to the point where a large collective of ten person companies can provide a higher-value good at a lower per-unit cost than a single large company.
Certain activities will always need large, high-capitalization organizations, but it seems that more and more of these companies and factories are being forced to become more flexible and nimble which is leading to a point where a larger and larger part of your world can be serviced by smaller teams than by large enterprises.
How do these things change with time, though? Do they always, invariably drop? Might computing conceivably go through another phase where things get consolidated into some kind of big servers that cost a ton of money? Doesn't sound likely to me, but I'm not very good at predicting the future. Actual research into these trends would be interesting to see.
Retail: just went through a mass-consolidation. The more dynamic firms, the "places to be" if you wanted to do something interesting in retail, were the ones doing the consolidating. Think Wal-Mart, Home-Depot, Walgreens, Best-Buy etc... While it may not be interesting to the crowd here, innovation in things like supply-chain has been stunning.
Consumer Goods: P&G has been unstoppable, and that's where most people in that business want to go because that's where the interesting things were happening.
Consumer Electronics: Dell, Apple, Sony etc... have crushed everyone.
Banking: Ditto. Where did all the ambitious young financial engineers want to go? Easy: Goldman-Sachs. Not that innovation here was a good thing... but that's another story.
Agriculture: ADM, Monsanto and the agri-business giants have almost completely taken over. Small farms are near-extinct.
This list could go on for a while, but I'll stop. I think you see a similar winner-take-all pattern playing out in most industries. It'd be interesting to see the demographics of all this: the percentages of people employed by institutions of different sizes.
Perhaps pg is right that it's starting to happen though. Maybe the web world is at the vanguard of what will be a broad counter-trend. I suspect that won't be as strong a force as he thinks though - the advantages of being big are still too great in too many fields.
Then again, maybe the mass-consolidation was artificial. Maybe all the M&As owe as much to artificially cheap money as to natural advantages of scale. I guess we'll see.
show me real examples outside of the web industry that this theory is even remotely correct