Oxford Nanopore has a serious case of user community unrest, triggered by the announcement that the P2 Solo (aka P2s) will be phased out starting this summer. P2 Solo is the two flowcell version of PromethION which required external GPU for basecalling; P2i is the sibling with onboard GPU. P2 Solo clearly has a very passionate constituency, one which feels betrayed by its discontinuation. This ongoing episode - ONT could always change their mind and restore the P2 Solo - is a cautionary tale of the challenges of changing corporate strategy in the context of a brand which has built a near cult-like loyalty.
As an aside, I always found the P2 Solo branding a bit confusing. Sure, the "P2i" with "i for integrated" would de-confuse me, but why did the "Solo" unit require a partner? By the same logic, I just celebrated 34 years of being "solo"?
Pundits such as yours truly have criticized ONT for having a product lineup more complex than their support and distribution infrastructure can reliably service. ONT is under increasing pressure to break even in a few years, and it isn't clear that their current sales growth will achieve this without more spending cuts. So it's not surprising that products have been marked for the knackers. But each of those has proponents in the user community who have bought into a vision and its real world applications. Longtime user Rasmus Kirkegaard posted on X/Twitter the image below lamenting ONT hardware that have been terminated - and on the ASeq Discord a user lamented they had tried to implement the now-discontinued Cas9 enrichment protocol into a clinical setting.
P2 Solo is being cut, according to the company, because it had a higher support burden. Higher support also could translate into more unhappy customers, since nobody likes their expensive gadget not working reliably out-of-the-box and every time. Connecting P2 Solo to external GPU involved some software trickery, which not everybody is really up for. While I've never tried the Solo, I have come to dread configuring AWS EC2 GPU instances, as getting CUDA and PyTorch configured and compatible (woe be to thee who installs the wrong version of CUDA!) can often consume both a good part of a day and all of my good cheer for that day.
GPU isn't cheap and sometimes is hard to come by. I've never seen a statistic, but it would seem certain that the ONT platform - particularly when run in the highest accuracy modes - requires the most FLOPS per gigabase of finished data. The P2s proponents eye the substantially greater price of P2i and smell a markup far beyond the guarantee that the device has the right libraries fully configured. And it could well be that ONT is getting some extra pounds sterling from each P2i out the door.
So it wouldn't be surprising for someone in the mood to enumerate legumes to see a high support burden for P2S, a lower upfront spend and perhaps even smaller pull-through (who knows? That isn't data ONT would make public) and see the P2 Solo as a PITA, a ripe candidate for pruning.
But as the backlash illustrates, the P2 Solo is seen as part of the ONT promise, to make sequencing available anywhere to anyone. MinIONs have great portability but just can't deliver on big projects. Plus, losing the low entry point Solo makes long term nanoporeans nervous if MinION might fall to similar logic - Nava Whiteford a while back suggested that MinION flowcells probably have much lower gross margins than PromethION flowcells and just imagine the simplification of cutting the number of flowcell types in half? A bit after I drafted that, Nava posted a new newsletter article again making the case to phase out MinION.
The P2 Solo faction likely has over-representation of expert nanopore users who have driven new application development and lots of buzzy headlines. ONT is trying to mature out of being addicted to clickbait, but this is still a community subset that has time and time again beta tested ONT's new chemistries, kits, instruments and software.
Is Oxford Nanopore now just lip syncing their "anyone, anywhere" mantra? It was still on their slides, but most of the emphasis (as befits a banking conference) was on where management thinks the money will flow from - and that is clearly high throughput sequencing and niche sequencing to support biopharmaceutical production. Not classrooms or nature centers.
So many of the long term fans who bought into the "we're going to support garage biotech" vision so long propagated by the company find their world greatly jarred by one of the best hacker tools being put to pasture. Perhaps the backlash will cause ONT to reconsider this change and instead continue the P2s but with some. Already outgoing CEO Gordon Sanghera posted to the community site reiterating that the decision was made due to high support burden for P2s
There's other signs that ONT is moving away from their traditional operating approach in practice, even if their JP Morgan presentation still touts the old messages. Kirkegaarde also posted the screenshot below where he was directed to contact a sales representative before an order for flowcells could be placed. That's a very unusual take in this space for routine consumables or fixed services. You can order ONT data generated at Plasmidsaurus without engaging in conversation with a person, why can't you order the flowcells to do it yourself that way? It's not that I'm a complete misanthrope, but don't want to waste my time or theirs with routine ordering - and will ONT be offering 24 hour, 365 day a year phone ordering support?
But the stark reality is the old Oxford Nanopore business model wasn't moving them to profitability and long term survival. ONT was going to morph, but it wasn't clear how. The loyalists are uneasy too about the new CEO, a cipher coming from a large conglomerate famous for vacuuming up other firms. Plus the Sales & Marketing organization has an increasing number of persons with experience at hated Illumina. London Calling will likely be as interesting for the undercurrents of gossip and watching the new management take the reins as it will for the always stellar scientific speaking lineup. See you there! - Unless of course I see you prior at SLAS or AGBT.
No comments:
Post a Comment