I think what Paul describes here is certainly a trend, but it has its limitations.
If each drug takes $1 billion to reach the market and 10 million people use it over its patent protected lifetime, then each patient contributes, on average, $100 to the development of that drug. If we keep shrinking the denominator, then the economics become more difficult. Taken to the extreme of personalized medicine, with one specific drug for each person, we cannot expect that one person to cover the $1 billion development cost. Even if the development cost drops to $1 million per new drug, the economics won't work.
I think the average development cost would need to drop to $10,000 per drug to be reasonable. To reach this price, we would need to exclude the FDA completely--allow drugs to be marketed without prior FDA approval--or allow the FDA to approve the process of drug development instead of each specific drug. With this arrangement, the FDA would evaluate and approve the process of developing personalized medicines, but would then stand aside and let the drug companies deal directly with patients, physicians, and managed care organizations.
Just like drug companies which built their business models on blockbuster medicines, the FDA will need to rethink its approach in the era of micro-market and truly personalized drugs.