Everything Wrong With: "Illumina: The Measurement Monopoly" Part 8/8
A couple of people suggested that I must be independently wealthy recently. I’m not exactly sure where this misconception comes from. Possibly it’s due to the fact that I’m often fairly direct with my opinions. Well… unfortunately I’m not. And I remain in need of money to fund this substack and my other activities.
So, I’m going to continue to moan about the Century of Biology having an order of magnitude more subscriptions than I do and commenting on what I think is wrong with the article.
Keith over at Omicsomics also posted his thoughts there and it’s well worth a read.
This final piece, where we’re covering this paragraph:
While Illumina represents a category-defining horizontal biotech outcome, the company has always had a relentless focus on producing their own form of vertically integrated product. Affymetrix, Illumina’s initial competitor in the DNA microarray market, attempted to follow the Intel playbook. They wanted their chip in every other product on the market, and were content with other companies emerging to produce instruments, reagents, or analysis tools. Illumina wasn’t. The goal was to own every single aspect of the product experience in order to move faster and deliver superior results. This is much closer to the Kodak business model—sell the camera and the film.
AffyMetrix
I don’t know much about Affy. By the time I entered genomics they were already boring. But they made their own microarray readers right? And the later microarrays were much more than just regular slide glass format chips:
Did these even fit in other instruments? So… they seem pretty vertically integrated to me.
It’s true that there has generally been a bigger ecosystem around Microarray primary data analysis. My general feeling is that this wasn’t entirely intensional. Microarrays have a much simpler datatype, and are more prone to bias1, so benefit from researchers fiddling with the analysis algorithms.
I doubt they were “attempting to follow the Intel playbook”… just going with the flow2.
Illumina
I don’t understand what a non-vertically integrated Illumina sequencer looks like. At least from the acquisition of Solexa until ~2 years ago.
Would they just produce flowcells and reagents? To run on what?
Even the first Genome Analyzer isn’t a simple instrument. It’s not like you can just produce flowcells and stick them on a microarray reader. You need to make a custom instrument. At the very least you needed to make:
Reversible Terminators and other reagents.
A oligo lawn coated flowcell.
An instrument integrating:
A complex fluidic system with temperature control.
A (sometimes TIRF) fluorescence microscope.
Analysis software to produce basecalls.
Without any of these components you don’t have an MVP3. So they had to build all this themselves. They just didn’t have a choice, nobody was making this stuff4.
The problem with the Century of Bio article is that it makes it sound like there was a Machiavellian business plan where they said “yes we must vertically integrate in order to move faster”. The reality is that (deeptech/bio) startups are generally a mess and decisions don’t often get made that way.
You can reasonably say that Illumina have pushed towards more vertical integration in recent years a few examples would be:
Pushing into applications
Acquiring Verinata Health (NIPT)
Acquiring GRAIL (Cancer Diagnostics)
Pushing into analysis
Dragen
Basespace
Results have been mixed. And we’ve seen many calls for Illumina to return to their “core competency” of just building DNA sequencers recently.
Personally I’m hoping for a “less vertically integrated” future. Where we can just fill sequencers with reagents like qPCR machines and run experiments. But I’m not holding my breath.
Conclusion
Overall I think Elliot’s article was pretty good! There’s clearly a lot I disagree with, but it covered a lot of ground. The fact that Elliot has so many more subscribers than me is of course a fundamental injustice. You can help redress this by subscribing, or encouraging others to subscribe.
But overall I may have to accept that Elliot does a much better job of making these topics accessible to those outside the sequencing industry than I do and try and learn some tricks from him.
Temperature, Ozone levels, phase of the moon etc.
Perhaps some post-rationalization happened?
Minimum Viable Product for those of you who don’t read hacker news far too much.
And nobody would make it either as it wasn’t clear that the complete platform would be viable anyway.