On the Demise of Codon Devices

Nature is carrying a short news piece by Erica Check and Heidi Ledford on the end of Codon Devices, "The Constructive Biology Company".  I am briefly quoted in the discussion of what might have gone wrong.  I would add here that I don't think it means much of anything for the field as a whole.  It was just one company.  Here is last week's initial reporting by Todd Wallack at the Boston Globe.

I've been pondering this a bit more, and the following analogy occurred to me after I was interviewed for the Nature piece.  Codon, as described to me by various people directly involved, was imagined as a full-service engineering firm -- synthetic genes and genomes, design services, the elusive "bio-fab" that would enable one-stop conversion of design information into functional molecules and living systems.  Essentially, it seems to me that the founders wanted to spin up an HP of biology, except that they tried to jump into the fully developed HP of 1980 or 1990 rather than the garage HP of 1939.  Codon was founded with of order $50 million, with no actual products ready to go.  HP was founded with ~$500 (albeit 1939 dollars) and immediately started selling a single product, a frequency standard, for which there was a large and growing market.  HP then grew, along with it customers, organically over decades. Moreover, the company was started within the context of an already large market for electronics.

The synthetic biology market -- the ecology of companies that produce and consume products and services related to building genes and genomes -- still isn't very big.  A very generous estimate would put that market at $100 million.  This means the revenues for any given firm are (very optimistically) probably no more than a few tens of millions.  (The market around "old style" recombinant DNA is, of course, orders of magnitude larger.)  Labor, rather than reagents and materials, is still likely to be the biggest cost for most companies in the field.  And even when they do produce an organism, or a genetic circuit, with value, companies are likely to try to capture all the value of the learning that went into the design and engineering process. 

This leads to an important question that I am not sure is asked often enough by those who hope to make a living off of emerging biological technologies: Where is the value?  Is it in the design (bits), or in the objects (atoms)?  The answer is a bit complicated.

Given that the maximum possible profit margin on synthetic genes is falling exponentially, it would seem that finding value in those particular atoms is going to get harder and harder.  DNA is cheap, and getting cheaper; the design of genetic circuits (resulting in bits) definitely costs more (in labor, etc.) than obtaining the physical sequence by FedEx. That is the market that Codon leapt into.  If all of the value is in the design process, and in the learning associated with producing a new design, not many companies are going to outsource that value creation to a contractor.  If Codon had a particular design expertise, they could have made a go with that as a business model, as do electronics firms that have niche businesses in power electronics or ASICs.  There are certainly very large firms that design, but do not build, electronics (the new AMD, for example), but they didn't get that way overnight.  They have emerged after a very long (and brutal) process of competition that has resulted in the separation of design and manufacturing.  Intel is the only integrated firm left standing, in part because they set their sights on maintaining scale from day one (see the recent Economist article on the semiconductor industry for a nice summary of where the market is, and where it may be headed). 

In another area of synthetic biology, I can testify with an uncomfortably high degree of expertise that costs in the market for proteins (a very different beast than DNA) are much higher for atoms than for bits.  It is relatively easy for me to design (update: perhaps better phraseology would be "specify the sequence of") a new protein for Biodesic and have Blue Heron synthesize the corresponding gene.  It is rather less easy for me to get the actual protein made at scale by a third party (and it would be even harder to do it myself).  Whereas gene synthesis appears to be a commodity business, contract protein manufacturing is definitely not.  Expression and purification require knowledge (art).  Even if a company has loads of expertise in protein expression, in my experience they will only offer an estimate of the likelihood of success for any given job.  And even if they can make a particular protein, without a fairly large investment of time and money they may not be able to make very much of the protein or ship it at a sufficiently high purity.  Unlike silicon processing and chip manufacturing, it isn't clear that anyone can (yet) be a generalist in protein expression.  Once you get a protein manufacturing process sorted out, the costs quickly fall and the margins are excellent.  Until then: ouch.

So, for DNA bits are expensive and atoms are cheap.  For proteins, bits are cheap and atoms are initially very expensive.  Who knows how much of this was clear to the founders of Codon several years ago; I have only been able to articulate these ideas myself relatively recently.  It is still very early in the development of synthetic biology as a market, and as a sector of the economy.