This is London Calling.
I used to like The Clash. I suppose their working-class revolutionary spirit appealed to me. Of course then you find out that Joe Strummer went to a school “where thick rich people sent their thick rich kids”… and had significantly more opportunities than most.
I guess revolutionaries are not always what they appear to be… but perhaps this doesn’t make the music any worse.
Anyway! Oxford Nanopore have made some platform announcements at London Calling 2023! The most interesting of which is the development of a new line of ASICs.
Slides suggest that the ASICs are fairly old (if only someone had imaged these!). They have therefore has been due an update for quite some time, and it seems that time has come.
The most interesting slide presented shows the SmidgION ASIC:
What’s particularly interesting is the diagram on the top left showing the routing of the well array down to the ASIC. Previous iterations of the flow cell had well electrodes routed directly down through the PCB to the ASIC. This was to “ensure a short connection path with minimal parasitic capacitance”. However it means that both the nanopore array and the ASIC have to be exactly the same size.
These new ASICs seem to have solved this problem, meaning you can potentially use a smaller ASIC with a larger array. The slides show a ~6x6 die which can support 400 channels. On a modern process estimates suggest this would cost ~$10 an older process might be as low as $1. For the new MinION ASIC they’re looking at 2048 channels, so $5 to $50.
The bigger concern is that this implies that the nanopore array itself hasn’t shrunk. So while you might be able to save money on the ASIC, costs associated with the array itself remain. Fabrication of that array is likely costly both due to the use of novel materials (Platinum or perhaps now Silver) and likely significantly lower yields than standard CMOS processes.
So while these ASIC changes are welcome, I personally wouldn’t expect them to have a massive impact on margins/COGS. So…
“The ice age is coming, the sun's zooming in
Meltdown expected, the wheat is growing thin”
Or is it? Find out after the break!
Overall the ASIC stuff is a welcome incremental development but I don’t expect it to have a massive short term impact for the reasons listed above. It also suggests that approaches to reducing the well/array size have so far not panned out (it seems to have been a long time since we heard about “voltage sensing”). So I would be concerned that we are reaching the density limits…
There were a few more images of the ASIC floating around, including this one from Oxford Nanopore advisor Ewan Birney which is probably a SmidgION prototype?:
The new ASICs will likely have the most impact on these lower end platforms (MinION, SmidgION). However reports suggest that the majority of their income comes from the larger instruments (PromethION). So overall, unless these lower end instruments segments grow significantly overall impact on revenues may be limited.
Moving these ASIC improvements to the PromethION could provide challenging. I suspect there is a limit to how far along the PCB you can route the electrode traces without it impacting data quality. So as the array gets larger you may run into issues using a single ASIC.
From what I can tell, slides didn’t show a new PromethION ASIC. So I remain curious as to what will happen here. A potential solution is to use two smaller ASICs on the PromethION. But the associated increased packaging costs here would suggest that the impact on PromethION COGS would not be as high as for the MinION…
New ASICs should however give a significant throughput boost. The 4x multiplexing on older ASICs was presumably specified at a time when pore yields were much lower. With 2x multiplexing and more concurrently active channels throughput should increase across all instruments!
This throughput increase could help make them more competitive with Illumina on a cost-per-base basis (if Illumina and ONT run costs remain static). However I suspect if you look at the COGS cost-per-base, Illumina will still be significantly cheaper, and therefore has room to compete on cost if they wish to.
There were a few other interesting points discussed, like a potential shift to Silver electrodes (which we’ve discussed in other contexts before). But as noted in the slides they need to explore fabrication methods for scale up… (Silver is in fact the obvious choice for electrodes, but I assume the current fabrication approach using Platinum is easier to work with at scale):
Aside from this, there were various other product facing announcements, none of which were hugely exciting to me (but feel free to reach out and chat about them).
Overall, the announcements seemed like solid incremental progress, but nothing that dramatically changes the landscape.