Companies are acquired all the time, and Thermo Fisher is a serial acquirer, as was Life Tech and Invitrogen before that. So Affymetrix cashing in wasn't a total surprise. But what stunned me was the price: $1.6B seems like a lot, but consider that Illumina is worth $24B. The charts below are showing stock price, and not market cap as I'd prefer (I couldn't find a good site), but the gist of the issue is clear: the last decade has not been kind to Affymetrix (which, as I keep forgetting, is $AFFX -- $AFFY is the original parent Affymax, which is apparently bankrupt).
|AFFY (top) vs ILMN stock price since ~2003|
Note not only the shapes of the price plots, but the needle plots below them of trading activity; Illumina has often been very active and has a greater area under the curve lately -- though at the beginning of the time period it was a minor player.
Stock price isn't everything by a long shot, but it does illustrate that as a business Affymetrix had fallen from great heights. I hadn't been tracking them closely, but the few times there was Affymetrix product news it just seemed to be small iterations on the existing platform. Affy never seemed to grok that sequencing was taking over their markets, or at least crowding them out from where Affy might try to grow their markets into.
Illumina still does a lot of array business; arrays ain't dead yet. But Illumina also can play both the array and sequencing markets, and in particular get the action as projects bounce from one to the other: a salesperson selling one can entice for the other.
When I was a grad student, several of us wondered who could possibly compete with Affy in the array space, as their semiconductor-like approach seemed unbeatably clever. NimbleGen sneaked in partly on very clever technology and partly by finding an IP loophole (at least some of Affy's patents weren't enforceable in Iceland, so NimbleGen did their array work there). Illumina seemed like a bit player when I first heard about them, and at Millennium we ran a lot of Affy arrays. But it was designing a custom array to cover our novel transcriptomic content that really brought home Affy's potential weakness: it is so expensive to make the masks for an array that you really need to obsess over your design. Illumina was a small player, though the first time I saw one of their devices I thought it beautiful -- 96 pins tipped with opalescent BeadArrays.
Affy only rarely showed any interest in sequencing, with a small press release here or a small acquisition there. Just imagine if Affy had gone and bought into a sequencing technology back when they were flush with cash and customers?
But ultimately, Affy never saw the problem. Fat on existing business, they didn't worry that RNA-Seq would steal away business -- or largely solve the problem of customizing arrays for novel content or novel applications. Mind share went to sequencing approaches: why work within the confines and limits of an array when a sequencing assay could potentially find so much more? Illustrating this particularly well is chromatin immunoprecipitation for studying DNA-binding proteins: ChIP went very quickly from ChIP-Array to ChIP-Seq, as the latter could give far better resolution and be far less biased. It took a lot of work to find analytical solutions -- perhaps in many cases the work isn't finished -- but XXX-Seq became the dominant approach to querying biological systems in new ways, not XXX-Array. Worse, if you heard about Affymetrix in relation to sequencing, it was some Affy alumnus starting another company.
Affymetrix got out before their markets collapsed, and join a collective of mature technologies housed at Thermo Fisher. Affy had over 20 years of presence in genomics, and the brand will be around a bit longer, but the innovative golden age of the nineties and early aughts is a distant memory.