Illumina's Boring New Sequencer is Boring

Boring doesn’t have to be bad dear reader. Boring is… dependable, reliable, like… a rock.. like a big heavy rock.

Clearly… DNA Sequencers should be big. And heavy. And boring.

And the NovaSeq X is one of the heaviest sequencers yet, with a crated weight of 722Kg! That’s nearly an additional 100Kg of value over the NovaSeq 6000. Hey you’re paying $1.25M for this. Nobody wants to buy a small sequencer for $1.25M do they? They want one that weighs as much as a small truck!

To be fair, this is still less than the market defining Pacific Biosciences RS which weighed in at 1091Kg, but it’s still an amazing effort from Illumina. I assume we can expect to see even larger instruments in the future when the logistics of shipping multi-ton crates have been worked out.

In fact, if we runs the numbers on this and prior Illumina sequencers, we can see a clear trend:

Based on this, I’d expect we’ll see instruments that weigh more than a metic ton from Illumina in ~2035. In doing so, Illumina will avoid the sad fate of mainframe computers, where IBM made the critical mistake of trying to make them smaller… rather than making them increasingly larger until their crushing weight requires purpose built facilities…

In any case, awesome work, I recommend everybody invests in steel reinforced flooring at their earliest opportunity¹.

For a more serious take on the NovaSeq X paid subscribers can read on below, or you can ping me for a copy (that is unless you work for an entity that prohibits me from purchasing their products). For those who don’t get the joke, the NovaSeq X seems like a solid instrument and the scientists and engineers behind it have done great work. The Illumina marketing on the X however suggests that the platform is a bit more transformative than it actually is...

Ok, so now that I’ve got that out of my system we can take a slightly more serious look at the NovaSeq X and other announcements.

NovaSeq X

First, let’s be clear. The NovaSeq X is an incremental development of the existing NovaSeq 6000 platform. I’m sure they’ve made signfiicant incremental developments in every area, but nobody is going to be throwing out their existing platforms to use this.

Both the transition from the Genome Analyzer to the HiSeq and HiSeq to the NovaSeq (with patterned flow cells and Isothermal amplification) were much more significant from a user perspective.

So Illumina’s marketing on this is pretty off base. The headline PR statement is that the NovaSeq X will enable $2/Gigabase (or $200/genome). Illumina could do that today if they wanted to, by reducing margins to ~60%. But they don’t want to.

So, their argument is I suppose that the NovaSeq X will allow them to maintain margins at this price point. Part of this is enabled by new technologies, but more significantly they’ve also announced that they’re moving more of the manufacturing process in house, by building out their own flow cell manufacturing facilities.

They’re using nano-imprinting to pattern flow cells, as stated in the presentation, and shown in recent patents, a relatively simple manufacturing process when compared to a full photolithographic semi-conductor fab process.

This could be a pretty significant, because they can now take a larger cut of what they were likely previously paying out to an external flow cell manufacturer.

Overall, things feel like they’re generally in line with my previous thoughts. We’re starting to see a race to the bottom, with Illumina trying to find new ways to maintain high margins. The problem is, with multiple new players all willing to sacrifice margin to gain market share (Element, Ultima, Singular, MGI and others) this could be difficult.

One more thing(s)

Chemistry X, sorry, X-LEAP SBS seems like a bit of a disappointment. Talk of doubling read length was notably absent. In fact, the NovaSeq X currently lists shorter maximum read lengths than the 6000 (2x250 kits are not shown in the X product page).

Run time is a little faster. The shortest 2x150 run times on the NovaSeq X are 21h, which is only slightly quicker than 25h on the 6000. This seems odd given that they said they can now image a lane in ~1s, and suggests that imaging is not necessarily the limiting factor in run time. Data quality appears to be equivalent, with both the 6000 and X stating > 80% Q30 at 2x150bp… so… no clear user benefit from Chemistry X?

Reagents can now be shipped at room temperature, which they say makes accessing a 700Kg instrument more accessible, but doesn’t seem like a huge deal to me. It’s possible this another factor that helps reduce their costs, and maintain margins.

Notably, nothing was announced for their most popular instruments (Miseq-class). Like other vendors they seem to somewhat reluctantly sell these instruments… which is unfortunate as I find them some of the potentially most exciting. The issue of course is that the bulk of their revenue comes from larger NextSeq and NovaSeq-class instruments…

Finally, they talked a bit about Illumina Complete Long Reads, which I’ve written about before. It still seems like anyone who is seriously interested in long reads will use a true long read platform (PacBio or ONT). But for users who just want to do a “little bit” of synthetic long read sequencing Illumina’s offering might be enough to keep them happy. Overall there’s still currently no clear large, non-research, market for long reads so I doubt Illumina feels that this is a hugely significant market for them.

Summing Up

There are lots of things that Illumina could have done that I would have found exciting. They own the MspA nanopore IP, so they could have come out with their own nanopore sequencer to augment their existing short read offerings. That would have been exciting!

They wanted to buy PacBio for $1.2B if they spent an equivalent amount of R&D couldn’t they develop something with similar performance characteristics in house? A real optical long read sequencing platform, possibly even compatible with the NovaSeq X, would have been exciting!

If they really wanted to address pandemic scale sequencing applications as they suggested in the presentation. Couldn’t they have come up with a small, cheap sequencer that would be a better fit? Rather than largely ignoring this market segment? A small sequencer I can buy for $1000 based on the iSeq technology? That would have been exciting too!

Unfortunately I think all these options are too risky for Illumina and we’ll continue to see well executed, but fundamentally risk-adverse, iterative development of the platform. Any fundamentally new technolgoies will likely only come though acquisition…

1. This is not investment advice.