Thursday, October 24, 2019

CMA Drops the (Preliminary) Hammer on ILMN-PacBio

Today the United Kingdom's Competition and Merger Authority (CMA) issued their preliminary report on the proposed acquisition of Pacific Biosciences by Illumina.  The report has no dry British phrasing: they clearly state that the merger is anticompetitive and that the only legal remedy is to block the transaction.  Interestingly, today was also the scheduled Illumina Q3 earnings call, but the subject wasn't even broached there.

The CMA document again has a lot of legalese and plenty of frustrating redactions.  Plus a few interesting reveals.  For example, PacBio portrayed itself to the CMA as a "failing firm", a specific term of art.  Unfortunately, all the arguments provided by the company to the CMA are redacted.  It is noted that many third parties expressed concern that without the merger PacBio might go under. Any arguments that CMA actually considered to vet the claim PacBio was failing are redacted -- there's a continuous set of redactions, other than posed questions and figure titles, from 6.53 through 6.93.  .  That includes, somewhat bizarrely, PacBio's cash reserves, which are public knowledge as of their last quarterly statement.  But in the end, with no unredacted justifications, the CMA concludes that PacBio is not a firm that would "exit the market due to financial failure in the foreseeable future".

One interesting note is that PacBio apparently started searching for a strategic partner as soon as Roche dumped them in late 2016. Alas, other than the fact they were searching for the obvious: better distribution channels and more R&D funding, everything is redacted.  Also redacted are details of PacBio's search for a strategic partner; indeed after saying something in 6.94 we go back to a long run of redactions from 6.95 through 6.117.  Again, a conclusion is made public: the CMA believes that PacBio could shop themselves to someone other than Illumina, and even if that would yield a lower offer than Illumina's it shows there is an alternative.

There's a whole section on PacBio and Illumina's attempts to convince CMA that their systems don't overlap which makes interesting reading, particularly since the redactor went very lightly so there's actually something to read.  The CMA quotes the companies as saying that Illumina and Sequel II are at their narrowest possible gap in price/performance right now, with Illumina expected to have further gains and Sequel II to remain relatively flat.  Interestingly, a reason cited for this is that PacBio has hit the limit on performance for the CMOS sensor electronics used in their instrument and does not expect this to improve significantly in the near term.  Ultimately, in a refrain echoing throughout the document, the CMA rejects PacBio's assertions and states a belief that further improvement in the Sequel platform is possible despite the core technology not advancing.  Obviously PacBio should try to hire away some of the CMA staff, as they see a path forward that PacBio may not.

The redactor gains some steam again in the section on how the companies view other players, but quite a bit remains.  Unsurprisingly, Illumina worries a bit about Thermo Fisher and QIAGEN (the document was clearly written before QIAGEN bailed out and so only has notes stating the new QIAGEN-Illumina relationship without analyzing it in detail) but is most concerned about BGI/MGI.  Also unsurprising is that PacBio views Oxford Nanopore as their biggest competitor.  Indeed, PacBio appears to have told the CMA that ONT is a stronger company (a redaction creates ambiguity on this point).  Both companies expressed the view that numerous potential new entrants exist which could become competitive in the next year and a half or so.

Another interesting technology note: Illumina stated to the CMA that they are working on their own long read technology and would continue to do so even if the merger is approved.  

Another interesting bit is that Illumina and PacBio commissioned consulting house DeciBio to interview customers and build a case for short read and long read markets not intersecting.  CMA quotes from this, but basically complains the study design bordered on push polling and used selective quoting in the report.  CMA also gripes that they weren't consulted in advance on the design of the study and says they "do not place material weight on the survey in our provisional assessment".  

One curious section follows.  PacBio submitted an econometric study suggesting that should PacBio fail, much of their market would move to Oxford Nanopore.  That's a statement that is pretty obvious;  the PacBio continuous long read mode pretty much can only be replaced by nanopore (those wishing a strong challenge can try to use 10X linked reads).  On the other hand, for some circular consensus sequencing (aka HiFi) applications there just isn't any other technology that can readily substitute.  But the CMA argues that price is more important than performance and somehow a disappearing PacBio would have customers chose on cost.  It's an odd take, but has the end result that CMA doesn't give any weight to this submitted argument either. 

The CMA also sought opinions from competitors.  Curiously, while they did solicit and apparently receive documents from BGI, ONT, Thermo-Fisher and QIAGEN, they did not contact 10X Genomics despite the report frequently exploring whether 10X linked reads are part of the competitive landscape.  In any case, a fun party game for sequencing industry followers would be to try to match the semi-redacted statements in this section with who must have said them.  For example, one lists Illumina, Agilent, Roche and two redacted companies as their closest competitors -- is this more likely Thermo Fisher or QIAGEN?  And if one of those, then what is the other redacted company name since whoever didn't write this is probably one of the redactions.  Matching is made harder and more mysterious by there being what appear to be five statements attributed to companies, but the list is of only four.  

On the other hand, a later statement that "long read and short read are substitutable for all applications" (and then expounds on this further for an entire paragraph) reads like several Clive Brown tweets so I'm not sure why they bothered redacting this; it's clearly ONT authored.  I suspect the other two points in this section (8.252) are probably from ONT as well, as they make a strong argument for long reads devouring the short read market.  

There's an interesting section on barriers to entry, including a long but incomplete list of failed sequencing companies.  The CMA homes in on the idea that patents and other IP represent one of the biggest barriers (though they name several more) and give a general assessment that PacBio has an extremely valuable IP portfolio and that after the merger this would be a potent weapon to suppress new entrants.

On bundling and tying the CMA again sees the merged entity as a threat to competition, in particular raising the spectre of the combined company very carefully targeting large customers with an eye potentially wandering to competitors.  They do actually reject the extreme situation of Illumina refusing to support instruments in labs using competitors' technologies (I'm pretty sure that would be outright illegal here in the States under antitrust law). But interestingly, the CMA reports "a number of customers already being offered bundles of Illumina and PacBio instruments".  

An interesting twist is that CMA has thrown Illumina's HiSeq X10 strategy back at the company: because the company forced customers to use the instrument only for certain applications and required specific numbers of sub-instruments to be purchased, they therefore demonstrated outsize market power and control over customers.  

But ultimately the CMA concludes, in one of the few wins for the case that the merger could go through, that bundling is unlikely to be a successful anticompetitive strategy for the merged entity.

An area where I definitely agree with the CMA is the question of switching costs for changing platforms.  The companies tried to downplay this as just a bit of installation and training; the reality for any established shop is that the informatics and lab flow will be significantly different.  That's what the CMA concluded after talking to customers.

On the new entrant front, the CMA contacted twenty four possible new entrants, though many apparently ghosted them.  Others tried to beg off saying they weren't creating systems in this space, though the CMA was apparently frequently skeptical of this.  But in general the CMA found, unsurprisingly, that most of the possible new entrants are very early stage and far from contemplating commercialization.  They also found only cursory competitive threat analyses in the documents provided by PacBio and Illumina, suggesting that neither company is significantly worried about any of the newcomers 

One other angle the CMA is required to consider is what benefits might accrue to customers should the merger proceed.  That could include more investment in developing the platform, more investment in commercial activities,  integration of workflows and informatics, and pushing PacBio into the clinic.  In line with most of the report, the CMA discounts each of these.  Illumina did offer up the example that they acquired and developed Solexa, but the CMA says that this doesn't apply as an analogy -- completely ignoring that Solexa potentially competed with Illumina's existing array business (which the earnings call made clear is truly racing forwards: even in consumer genomics arrays are seeing pressure from sequencing).  Similarly, in an unmerged world PacBio will successfully enter the clinic -- even if it hasn't succeeded so far.  The CMA also offers the possibility that further development spend in a merged entity would be tilted towards "complementing Illumina's portfolio rather than competing with it; such as focusing on longer read lengths".  Actually the construction of that sentence is about as clear as mud -- are they saying that longer read lengths on PacBio are complementary or competing with Illumina? A following sentence suggests they believe an independent PacBio would focus on "reducing the cost or increasing the throughput".  It's an odd argument either way: regular read length improvements would be key to competing with Oxford Nanopore and will drive throughput (and therefore cost per basepair) down.  

The end result of all this (plus plenty I haven't attempted to summarize -- it's a loooooong document even with redactions), is a conclusion that PacBio and Illumina are significant competitors and the merger would reduce consumer choice and innovation.  They unsurprisingly see no way to remedy the issues other than nixing the merger.

Given the UK is a major market, the idea of the companies abandoning it so they can still merge doesn't seem likely (but could we call that Brexit-Seq?).  So this seems to be a complete blockade of the acquisition attempt, regardless of what the still-silent U.S. FTC rules.  Wall Street had a modest but negative reaction; PacBio stock fell another 60 cents or so.  Could five dollars and change represent the Wall Street view of the value of an unmerged PacBio?  Or does that still have a component from bulls who believe that somehow the merger can still go through?  And will Illumina cut off their lifeline to PacBio sooner, walking away  after tossing the breakup fee over to their erstwhile fiancee. 

It is clear from the document that Illumina would be blocked from buying any sequencing company with sales in the U.K. or domiciled in the U.K.  Not that ONT would even contemplate going that way, but it does mean that Illumina needs to pounce on any new entrants prior to them having any U.K. presence (in the section discussing whether the CMA had jurisdiction, Illumina's market share is much more than sufficient to trigger CMA to have authority).  PacBio could probably sell themselves to any of the other players and that would survive this review, just because they still wouldn't have much market share.  I can't help wondering if anyone at Illumina remotely anticipated this regulatory roadblock; if anyone had any inkling this might happen would they have gone forward anyways?


Anonymous said...

Thanks for writing this, its much more fun to read than wading through the CMA documents.

I disagree with your last paragraph. I think its actually now easier for ILMN to buy a UK company, for example ONT. In the process of the investigation the CMA have effectively made the case that ONT is in the clear. It was not rated as a significant competitor to ILMN, with very different products and users. An alternative strategy might be for ILMN to switch to ONT (or another).

That supposes the motivation behind the merger isn't just about capturing and exploiting IP and blocking the high end labs off from competition.

I think ILMN did anticipate some regulatory hurdles. They gave the process a very long time window. In their recent investor calls they say they are working on their own long read tech too (although they've been saying that for 10 years, remember 'Chemistry A and Chemistry B' ?), so maybe it's just bluster.

PacB should get $100M+ from a failed merger, sequel II sales look promising, they can certainly stand alone and compete on merit. If only they'd stop the patent trolling.

Liang Zong said...

All right, it seems in the end illumina gets a failed merger, PacB becomes a failed company, and leaves sequelII users failed customers.