Here is a short article by Emily Singer at Technology Review on commercial gene synthesis, "DNA Factories".  The article contains a couple of tidbits I haven't seen before:

[Codon Devices] plans to use itsenhanced synthesis capacity to find better enzymes for industrial processes. Since nature hasn't always come up with the most effective proteins, scientists often design a more effective enzyme by tweaking the DNA code used to make it. But it's difficult to predict in advance which tweaks will produce the best enzymes. Codon is now using its synthesis technology to carry out that process en masse--it makes millions of copies of the same genetic construct with slight variations--and then tests them to figure out which does its job best.

For example, scientists are now hot on the trail of the ideal cellulase--an enzyme that can break down cellulose in plants. More-efficient cellulases are important for producing cellulosic ethanol, which is ethanol derived from waste biomass rather than from corn starch or sugarcane, and therefore more cost-effective. "We can take the sequence for the cellulose enzyme in, say, a termite's gut, use a computer program to figure out different ways to optimize the sequence, churn out a million different versions, and then test them to find the top ten forms," says Brian Baynes, chief scientific officer and cofounder of Codon.

...The company is planning to open an expanded production facility, which will operate much like any other mass-production facility, except its product will be DNA. Codon intends to build a facility, slated to open this summer, that's much larger than current needs warrant to prepare for the DNA-synthesis boom.

That isn't quite direct confirmation that Codon has an internal effort underway to compete in the biofuels sector, but it is close enough.  There is a heading for "bio-energy" on the company web site, but little more info, and no word anywhere I can find on whether or not they have partners in this effort.  The site doesn't give any hints about the way they make libraries of variant genes, save that it is "proprietary technology".

Anybody from Codon care to comment?

Finally, as we explained in Genome Synthesis and Design Futures, at the present pace of price reductions constructing a genome the size of E. coli should cost less than US$10,000 within 10 years.  It isn't at all clear to me how many orders that size are going to come in, and Codon and its competitors may have to subsist on a large number of orders in the $10-1000 range.  So Codon's new gene foundry better guarantee a path towards a couple of orders of magnitude improvement in cost, or their margins are going to be seriously squeezed.