Friday, July 21, 2017

A Third GridION X5 Pricing Plan

When Oxford Nanopore announced their GridION X5 instrument in March, I and others attempted to parse the difference between the two pricing plans  -- and I made a bit of a hash of it.  The X5 runs 5 MinION flowcells independently in parallel from a single desktop instrument, which also includes FPGA-based acceleration of basecalling plus a license to perform sequencing-for-hire.  Indeed, Matt Loose tweeted out an image of an "X6" and then mention of an "X7"; the X6 had a MinION plugged into the USB port and apparently the FPGA unit can keep up with seven flowcells all running simultaneously.  Now Oxford has launched an interesting third "Starter Pack" plan that offers an even lower price point for the system.
ONT launched GridION X5 with two plans: one called CapEX (Capital Expenditure) and one called OpEX (Operational Expenditure).  ONT has said that the CapEX plan is really designed to help potential customers out of an accounting trap: some grants apparently have capital budgets that can't easily be converted to non-capital items.  At any company I've been at such irrational budgetary inflexibility would be properly viewed as madness, but I know that isn't always the case.  I grew up with paternal rants against "beancounters", though I've had to adjust my attitude a bit since two in-laws are accountants.  Accounting is useful I know, but turning into a straitjacket is silly.

That anecdote is relevant in the sense that how you treat these things is subject to both your organizational and personal attitudes towards budgeting and spending.  ONT tends to be very decent in their billing, only charging when goods are actually delivered.  But in my mind, once I've committed to a $12K buy those 120 Benjamins are gone.  

Anyway, under the preferred OpEX plan the initial buy is $158,499 (yes, even in high tech toys the silly 9s pricing prevails -- though blessedly it isn't $158,499.99).  That gets you an instrument and 300 flowcells.  Shipping is $1K, and considering you really have no option its a bit silly the price isn't stated as $159,499 -- indeed, I'll just call this $160K.  Then there's a $15K maintenance and service contract that goes to $10K after the first year, so if you just plan to keep the instrument a year that's a $175K commitment.  Also, you get to buy further flowcells at a discounted price of $475.

Now, comparing this to the new option does start running into the hotdogs and buns issue: in the U.S. at least hotdogs are sold in packs of 8 but buns sold in packs of 12.  So getting everything to line up is a bit fun.  That's even leaving out the unpleasant issue of reagent shipping, which is over $100 a pop.  Nobody in their right mind would take those 300 flowcells in one go, so that's an additional charge.

The new "Starter Pack" gets you the instrument, 60 flowcells and 10 sequencing kits  for the tidy price of $49,955.00 (hey, it doesn't end in 99 -- but instead 55!).  Shipping is free. No maintenance and service charge for the first 6 months, but then $10K per year after that.  So one of our first hotdog issues.  Actually, throwing in the sequencing kits (and the new rapid kits require no external reagents) is nice, but really is another set of hotdogs. 

So clearly the new package lowers the cost of entry.  Great way to get a camel's nose under your organization's tents.  Of course, $50K is still a pretty sizable outlay; at many companies you go pretty far up the food chain to get that approved.

Suppose you bought one with this scheme and then started running flat out.  At what point might you regret not getting the OpEx package?

Well, looks like hash time again -- my numbers make the new option look too good, so I must have screwed up something.  But here's my math so you can find the bug(s).  The scenario I used is what if we run 300 flowcells, each using a single rapid kit.  Reagent shipping costs are seen as a wash between the two scenarios and left out.  I also left out taxes; would seem a wash but perhaps not (or could depend on your jurisdiction).  Kits are the Rapid 1D kits, which are $570 for 6 reactions -- so the Starter Pack option needs 40 more of these and the OpEx requires 50 in total. For the Starter Pack option, the remaining 240 flowcells are purchased at the best rate of $500 a piece.

StartPack OpEx
Purchase 49,955 158,499
Shipping 0 1,000
Flowcells 120,000 0
Kits 22,800 28,500
Maintenance 10,000 15,000
202,755 202,999

So you can understand my confusion. The numbers are nearly the same, but the Starter Pack actually comes out better.  Now, after this initial run the OpEx option gives a right to buy flowcells at $475 a piece rather than $500, so some savings will accumulate there and could get substantial if you run many more flowcells.  So if I haven't completely fouled this up, the Starter Pack is a pretty safe bet through 300 flowcells.  If a lab runs ten flowcells a week, that's still over a half a year of running before approaching apparent breakeven.  And if that $25/flowcell extra discount adds up to $7.5K for the next 300 flowcells; not nothing, but not a big payoff for making the bigger commitment up front.

We can also look at the cost from another direction.  Suppose we tried to reconstruct this by buying everything (or best equivalent) separately.  Five MinION starter kits at $1K each is $5K; buying the remaining 55 flowcells would be $27.5K (ignoring the buns issue of that pricing is only available in packs of 48) and the remaining 5 kits would be $2.9K.  That comes out to $35.4K -- so Oxford is pricing the bundling of fee-for-service license, convenience of an integrated solution (vs. a farm of MinIONs hooked to 5 computers and probably basecalling on your cluster) and the FPGAs at about $15K.  Imagine that this price is just for the FPGAs and the other two items are throw-ins -- it suggests that Oxford may price a standalone FPGA box scaled for a MinION at around $3K.

Well, hope I haven't made too bad a mess of this.  The floor is now open to comments, corrections & brickbats. 


James@cancer said...

The most important consideration for labs like mine is the flexibility to offer services. Most of my labs work is internal so I doubt I need to pay for a "license" and could offer services on MinION. However GridION is a nicer package if we want to scale things up in the future. For commercial service providers the GridION probably is must because it comes with that license to offer fee-for-service.

There is a lack of clarity on my part about who ONT wants to prevent from offering services for cash without buying a license. many core labs will take money for doing jobs both internally and externally to their host institution. And commercial fee-for-service labs will also work on collaborations where no fees are charged but consumables are reimbursed. What happens in these situations?

The core/service lab division is a murky one...always has been. I'm surprised ONT make this distinction at all - they want as many users as possible irrespective of whether they own an instrument or not!

brian said...

Hi Keith, my understanding (from talking to Nanopore this week) is that the GridION is currently GPU-powered, not yet FPGA-powered. I think the CapEx vs OpEx model may make more sense in the world of academic grants. I bet it's easier in some cases to request $125k to buy a heavy machine, kind of like how universities are always building new buildings instead of fixing the plumbing...

Shawn said...

Hi, Keith. I don't think there's any reason to think you've gotten your math wrong. The "value" in opting for the upfront costs of the CapEx version is what you've already stated - some budgets are split between CapEx and OpEx. This is pretty common on the academic side - grants are written to buy the big, sexy toys, but they often don't come with ANY money for actually operating the toys. That comes from completely separate sources. It doesn't really make sense, but I've seen this at a LOT of different institutions.

Unknown said...

James: Oxford outlines what they mean by 'commercial service' in their Nanopore Product Terms and Conditions. They use the phrase "intended for or directed towards commercial advantages or monetary compensation". I guess that excludes cost recovery. Actually, the licencing seems quite restrictive, techincally... 3rd party provider customers need to be bound by the same T&C, and need to register with Oxford.

Anonymous said...

Monetization is an enormous challenge, for the year ending Dec 2016, Nanopore reported revenues of 4.5mm. That's what the GridIon and it's convoluted pricing scheme reflects.

It's still a far cry from justifying it's 1.5 bln (?) dollar valuation. It's a good thing they are private, otherwise the stock market's opinion would be far more brutal.