Nature Biotechnology has a little news article profiling some of the companies driving the cost of sequencing down. There's ambition in spades:

Companies have also started to win bids under the NHGRI $1,000 genome program. Unlike the $100,000 technologies, which focus on refining and improving existing methods, the conception of a $1,000 genome requires an entirely different paradigm-a discontinuous innovation. Helicos' technology, unlike the cluster-based approaches of 454, Agencourt and Solexa, could provide such a leap: in the first commercial award under the $1,000 program, it received, in October 2006, a $2 million grant to further develop its single-molecule approach.

According to Steve Lombardi, senior vice president of Marketing at Helicos, "If you had perfect chemistry, and each step was 99.99%, the instrument would generate 100 billion bases a day. The instrument is being designed for that throughput, but the first-generation chemistry will have a smaller yield-around 600 megabases per day." Improvements in chemistry could move Helicos to the $1,000 genome "in the first few years," he claims-well ahead of the NHGRI goal of 2014.

Everyone talks about personalized medicine and whatnot, but my first reaction is: wow, are microarrays already almost obsolete? With 600 megabases per day (the human genome is 3300 megabases), SAGE on a huge scale is possible...