Illumina Inc (ILMN) 2017 Q2 法說會逐字稿

Welcome to the Q2 2017 Illumina, Inc.

Earnings Conference Call.

My name is Adrienne, and I'll be your operator for today's call.

(Operator Instructions) Please note, this conference is being recorded.

I'll now turn the call over to Rebecca Chambers.

Rebecca Chambers, you may begin.

Thank you, Adrienne.

Good afternoon, everyone, and welcome to our earnings call for the second quarter of fiscal year 2017.

During the call today, we will review the financial results released after the close of the market and offer commentary on our commercial activity, after which we will host a question-and-answer session.

If you have not had a chance to review the earnings release and earnings presentation, both can be found in the Investor Relations section of our website at illumina.com.

Participating for Illumina today will be Francis deSouza, President and Chief Executive Officer; Marc Stapley, Executive Vice President and Chief Administrative Officer; and Sam Samad, Chief Financial Officer.

Francis will provide a brief update on the state of our business, and Sam will review our financial results.

This call is being recorded, and the audio portion will be archived in the Investors section of our website.

It is our intent that all forward-looking statements regarding our financial results and commercial activity made during today's call will be protected under the Private Securities Litigation Reform Act of 1995.

Forward-looking statements are subject to risks and uncertainties.

Actual events or results may differ materially from those projected or discussed.

All forward-looking statements are based upon current available information, and Illumina assumes no obligation to update these statements.

To better understand the risks and uncertainties that could cause actual results to differ, we refer you to the documents that Illumina files with the Securities and Exchange Commission, including Illumina's most recent Form 10-Q and 10-K.

With that, I will now turn the call over to Francis.

Thank you, Rebecca, and good afternoon, everyone.

I'm pleased to report that Q2 was another strong quarter for the company.

Revenue grew 10% year-over-year to $662 million.

The demand for NovaSeq again surpassed our expectations, with orders beating forecast by over 30%.

More than 230 NovaSeq instruments have been ordered since its launch in January.

As expected, we're seeing the beginning of the HiSeq replacement cycle, with 2/3 of NovaSeq orders coming from HiSeq and HiSeq X labs accessing NovaSeq's improved sequencing costs and capabilities.

With 2 quarters behind us, we're still at the very beginning of the HiSeq replacement cycle and our pipeline of NovaSeq opportunities continues to grow.

Equally importantly, we're also seeing the early signs of market elasticity as new-to-sequencing and benchtop-only customers accounted for more than 1/3 of the NovaSeq orders received.

Commercial customers are fueling the early NovaSeq adoption given the relative flexibility in their capital budgets.

Interest from academic labs continues to build as customers work through the grant application process, and this is expected to generate additional demand in the second half of this year and beyond.

I am pleased with the great work by our development and operations organizations in the second quarter as we successfully tripled our NovaSeq manufacturing capacity compared to Q1, enabling us to ship and install approximately 80 instruments.

But we're not quite at full capacity yet.

With our success to date, I am confident that the team will make the progress necessary to manufacture at capacity in the third quarter.

After that, it will take a few quarters to work down our healthy backlog of more than 100 instruments, as is typical with our product launches.

Looking ahead, our NovaSeq accessory device to enable customers to access each flow cell lane is progressing to our internal timelines and will ship in Q4.

S1 and S4 flow cell development timelines also remain on track.

We expect S4 will ship in late September to early-access customers, with full availability of S4 and S1 to follow in October.

As expected, NovaSeq has impacted HiSeq and HiSeq X shipments in the second quarter.

Going forward, our outlook for this family of instruments remains muted as we expect customers will choose to access our newest high-throughput platform.

Moving to our benchtop portfolio, new to sequencing customers represented about half of NextSeq, MiniSeq and MiSeq shipments, and our win rates remain stable.

Utilization was within each instrument's respective guidance range.

NextSeq pull-through increased compared to the prior year as commercial customers adopted the platform in production settings.

In Q2, microarray revenue, including services, grew 16% year-over-year to approximately $110 million.

Shipments of array products and services to our direct-to-consumer customers more than doubled versus the prior year.

Also in the consumer market, Helix launched their online marketplace a few weeks ago, which includes an initial 20 products and services that span entertainment, family, fitness, health, genealogy and nutrition.

As you know, we formed Helix to accelerate the adoption of consumer genomics.

The strong growth of our direct-to-consumer customers and success of the Helix launch have together further fueled our enthusiasm for the role our technology will play in developing the consumer genomics market.

Moving now to our clinical markets.

Demand from oncology testing customers continues to be robust, with over 20% shipment growth versus the prior year, primarily driven by commercial molecular diagnostic and liquid biopsy customers.

We reached an important milestone this quarter when our Extended RAS companion diagnostic kit received FDA approval as a Class 3 PMA.

This product was developed in partnership with Amgen to identify patients who are eligible for treatment of metastatic colorectal cancer with Vectibix and is the first oncology IVD that we have brought to the market.

And lastly in oncology, we are excited about the recent FDA pan-cancer approval of Keytruda and the implications for genomics in patient care.

With this landmark approval, for the first time, patients can access treatment based on genomic biomarkers rather than the anatomic origin of the cancer.

We believe this is the first of many therapies where the genetic profile of the tumor will be a critical component of ensuring the best patient care.

Another recent product announcement was the launch of VeriSeq NIPT, a CE-IVD-marked NGS solution.

Since launch, adoption is ahead of our expectations and customer feedback has been very positive.

Reimbursement in Europe is progressing as Belgium recently began covering T21 testing for all women, and the French health authority recommended NIPT coverage in pregnancies with greater than 1 to 1,000 risk.

In closing, I am pleased with our second quarter results given the continued progress of our NovaSeq launch, the trends we are witnessing across the rest of our portfolio and market segments and our teams' execution.

I will now turn the call over to Sam for a detailed overview of our second quarter results.

Thanks, Francis.

As Francis mentioned, revenue grew 10% year-over-year to $662 million.

Sequencing consumables and microarrays exceeded forecast, which led to the outperformance versus our expectations.

Geographically, the Americas grew revenue 8% versus the prior year period, while Europe saw a 15% increase, primarily driven by sequencing consumables.

Asia Pacific revenue grew 12%.

Continued weakness in Japan was more than offset by 16% shipment growth in Greater China, which benefited from NovaSeq strength and HiSeq X consumables and continued NextSeq demand from our NIPT customers and partners.

Revenue from sequencing instruments grew 9% year-over-year to $130 million, driven primarily by NovaSeq demand.

Total instrument revenue, which includes microarrays, increased 8% to $136 million.

Consumable revenue represented 61% of total revenue or $402 million, an increase of 6% compared to the prior period.

Sequencing consumable revenue grew 9% year-over-year to $338 million.

As we shared on our Q1 call, we expected customers to continue to ramp down HiSeq and HiSeq X reagent orders during the quarter, instead using inventory on hand as they prepare to integrate NovaSeq into their operations.

As predicted, HiSeq consumables decreased sequentially, while utilization gains in China and translational liquid biopsy studies led to an increase in HiSeq X consumables.

Turning now to gross margin and operating expenses.

I will highlight our non-GAAP results.

I encourage you to review the GAAP reconciliation of non-GAAP measures, which can be found in today's earnings release and presentation.

Please note that all subsequent references to net income and earnings per share refer to the results attributable to Illumina stockholders.

Our Q2 non-GAAP gross margin was 67%, up 60 basis points sequentially, driven by improved overhead absorption associated with higher production and decreased inventory reserves, which was partially offset by higher sequencing instrument mix.

Our Q2 guidance contemplated a slight sequential decline in margin based on lower sequencing consumable mix and absorption.

Year-over-year, gross margin decreased 540 basis points and was impacted by an increase in array services mix, inventory reserves related to product transitions and lower instrument margin from the NovaSeq introduction.

As is typical with a new platform, we are actively working on cost reduction and expect NovaSeq margin to improve over the next 4 to 6 quarters.

Operating expenses equaled $297 million, up $5 million sequentially as headcount adds were partially offset by the deconsolidation of GRAIL's expenses.

We had contemplated certain activities in our Q2 guidance that will now occur in Q3, including Helix's launch.

This timing as well as new investments initiated given our increased confidence in the full year outlook are expected to lead to an increase in second half operating expense compared to the first half, with Q3 higher than Q4.

Non-GAAP operating margins were 22.1% compared to 17.5% in the first quarter and lower than the 27.2% reported in the second quarter of last year due to the previously mentioned drivers impacting gross margin and increased headcount.

Excluding GRAIL and Helix, operating margin was 24.4% compared to 22.9% in the first quarter.

We reported second quarter GAAP net income of $128 million and EPS of $0.87 per diluted share.

Non-GAAP net income was $121 million or $0.82 of EPS, with Helix dilution of $0.05.

Cash flow from operations equaled $178 million, which was impacted by outflows of $13 million related to Helix as well as a significant estimated income tax payment, primarily related to the gain on the partial sale of our GRAIL stake in Q1.

Improved revenue linearity led to Q2 DSO of 51 days compared to 56 days last quarter.

Capital expenditures in Q2 were $69 million, and we reported an additional $32 million increase in property and equipment recorded under build-to-suit lease accounting, where such expenses were paid for by the landlord.

Consequently, Q2 free cash flow was $109 million.

We ended the quarter with approximately $1.9 billion in cash and short-term investments.

Turning to guidance.

Based on our strong first half results, we now project approximately 12% total company revenue growth and full year GAAP earnings per diluted share of $5.36 to $5.46.

We are reiterating our non-GAAP EPS guidance of $3.60 to $3.70 as a result of lower gross margin expectations given updated mix and higher NovaSeq revenue projections as well as increased operating expense.

In closing, I am pleased with our strong second quarter results, the continued success of the NovaSeq launch and the positive outlook for the rest of the year.

Thank you for your time.

We will now move to the Q&A session.

(Operator Instructions) Operator, we will now open the lines.

(Operator Instructions) And our first question comes from Tycho Peterson from JPMorgan.

I want to start with the consumable dynamics.

Certainly, a big delta, I guess, relative to what we've been expecting.

You mentioned HiSeq consumables decreased sequentially.

I guess, some of the increase was on China and liquid biopsy.

Can you maybe just talk on those dynamics and how we should be thinking about the consumable ramp in the back half of the year?

Sure.

There are a number of things that played out from a consumable dynamics perspective in Q2.

In the high-throughput segments, we did continue to see a decline in the HiSeq consumables as we talked about in Q1 and as we expected to continue to play out.

On the X side, we did see strength driven by both our customers in China as well as you mentioned and we talked about on the call, our translational liquid biopsy customers.

And in fact, as we looked at the X instruments that were connected into BaseSpace, we continued to see our customers running Xs as they usually had.

And so we expect really the transition to happen to NovaSeq as S4 comes out later this year.

But we are seeing customers continue to run their -- at their Xs.

And then in terms of consumables, we also saw strength in NextSeq as we continue to see customers looking to standardize on that platform for their production environments.

Okay.

And then just one clarification.

Is the incremental OpEx in the back half of the year all tied to Helix?

I had a few people asking about just you're not raising EPS due to the $0.15 beat.

Is all that incremental spending Helix related?

No.

Helix is one of the factors, but it's certainly not the whole factor.

Yes, Helix was more of a timing [title] from Q2 into Q3 because of the launch in Q3.

But as I mentioned in the prepared remarks, given the increased revenue projections, we have increased some of the investments in the back half.

And so now, we will have a growth in the second half compared to the first half in terms of operating expenses.

And we haven't changed our Helix dilution, Tycho.

And obviously, gross margin and the components of gross margin that Sam mentioned in the prepared remarks are also a driving factor.

And the next question comes from Doug Schenkel from Cowen.

I guess some related questions.

Again, on consumables, you've talked about stalling in consumable spend as customers work through inventory and also pause to consider shifting from HiSeq to HiSeq in the X to NovaSeq.

It seems like you may have seen some signs that we're moving past this quarter.

Is that the case?

And I guess, one way or the other, at what point do you think it's fair for us to at least contemplate in our models a ramp in spend on a per-NovaSeq basis?

Is that the second half of this year?

Is that something we should be thinking about more as we move into 2018?

Sure.

Maybe I'll start with the NovaSeq, sort of when we get to what the pull-through rates and sort of how we should think about the ramp.

So obviously, we started to see a ramp in NovaSeq consumables play out in Q2, but this is sort of the beginning of that ramp.

So if you look at the high-throughput consumable dynamic, as we said, we've already started to see the HiSeq consumables decline quarter-on-quarter.

We -- as S4 comes out, we expect to start to see some X customers draw down inventories as they start to move their payloads to NovaSeq as they get them.

That hasn't yet started to play out.

Again, we expect that to really start to play out when we launch S4.

And so over the next couple of quarters, you will see NovaSeq ramping up.

You will see that being offsetting the decline you expect to see in X and in HiSeq.

And over the longer period, obviously, we expect to see high-throughput consumables grow.

Okay.

That is helpful.

And you touched on a few things that I was -- I wanted to follow up on.

One is it sounds like there were a higher portion of orders coming from -- NovaSeq orders coming from folks that are new to sequencing or maybe even moving up from benchtops than certainly I would've anticipated.

One, is that the case?

And then I guess somewhat related to that, how -- what proportion of the market do you think is holding off on placing a formal order in advance of: One, the S4 launch; and two, the launch of the single lane loading solution?

Okay.

So you've covered a lot of ground, not surprising, and great questions.

I'd start by saying that, yes, I think we were actually pleasantly surprised to see about 1/3 of the NovaSeq orders came from customers who were either new to sequencing entirely or who were benchtop customers who didn't have any high-throughput instruments at all.

And I can share with you just anecdotally some of the examples.

One of the example's a customer in EMEA that opened up a new lab funded by government research to perform evolutionary genomics work.

And so they saw the NovaSeq as sort of the right price point for them to launch that lab.

Another customer also in the same region purchased NovaSeq to perform routine reproductive health testing in the country, which has been historically underserved in this regard.

So they planned to start offering clinical exome services to their customers, fundamentally, enabled by the pricing of NovaSeq and its lower cost per sample.

The other set of customers that surprised us were customers who were Illumina customers but were benchtop-only customers.

And one example is a lab that's projecting to sequence 1,000 genomes and 300 to 400 exomes annually for the next few years, and they saw NovaSeq is, again, the right set of price point for them to move up from being a benchtop-only customer.

So again, the fact that, that was a 1/3 of the NovaSeq orders was a really good thing from our perspective and a really encouraging sign for the demand from that segment and exceeded our expectations.

In terms of who's not yet bought and how we're thinking about who's bought.

Again, as I said in the prepared remarks, we are at the very, very beginning of the HiSeq replacement cycle.

Less than 10% of HiSeq customers have ordered the NovaSeq so far, so the bulk of this wave is still in front of us.

And let me give you some profiles of people that will be buying in the future.

For example, if you're an academic customer and you came into this year with a fixed budget, what you're doing now, in some cases, is you are writing your grant applications and using that to get the budget for NovaSeq in the future.

And so there are customers like that, that once NovaSeq expect to get it and are securing the funding now.

As I said in the prepared remarks, a lot of this early interest came from commercial customers.

And as you know, they have more flexibility with their budgets and were able to move more quickly.

And so they would be academic customers.

A certain segment of them, would be example of customers that are planning to get a NovaSeq but haven't pulled the trigger yet.

And then in terms of -- you asked about the accessory device, there are customers, core labs, specifically, that are waiting for the accessory device because of how they run their services.

And they like to do the single investing per lane, whether it's to run a genome per lane.

And so for them, that device in -- will be fundamentally enabling, and that doesn't ship, as you know, until Q4.

There are other customers that are looking at S4 the same way, so other examples of customers that are planning their purchases but haven't pulled the trigger yet.

So fair to say that at this point, you're tracking well ahead of plan in terms of number of orders, with a lot of the upside coming from commercial customers and new-to-sequencing customers with the potential for -- a potential acceleration in growth as you roll out some of the new accessories, the S4 chip and as those who are writing grants are actually at a better position to actually buy instruments.

Would that be a fair characterization?

That's exactly right, Doug.

Doug, I would also just -- and remember, this is a multiyear refresh cycle that we're going to see here.

So with 800 customers to address, this is going to take -- and not necessarily every single one of them are fully addressable.

It's going to take a number of years to play out.

And the next question comes from Derik De Bruin from Bank of America.

So just a couple of questions.

Just to clarify, you did say 80 installs for NovaSeq in Q2, correct?

Yes.

Yes, that's right.

Okay, great.

That's what I thought.

So 30 HiSeqs came out of the installed base.

Can you say which of those -- are those related to sort of like NovaSeq orders?

And how should we think about the HiSeqs coming out over the next couple of quarters?

Is 20 to 30 a reasonable run rate to think about this going forward?

So Derik, on that question, we -- it's not actually an item that we forecast.

We have a pretty stringent hurdle for taking things out of the installed base.

It effectively has to be very effectively disconnected and practically in the closet, if you will, or put away into a different location.

So it's not a number that we forecast, and therefore, it's challenging for us to make that comment.

I think it's also therefore challenging to tie that specifically to NovaSeq orders.

Obviously, many of these customers are transitioning to the NovaSeq platform, and one lends itself to the other, but they aren't necessarily exactly at the same customers and locations.

What we typically see is customers who will buy a new platform, and it will take them a couple of quarters.

They'll keep the platform on hand to use if they need some extra capacity.

And then it'll take them many -- a few years to actually officially turn that instrument off.

Great.

And so about the Helix launch the other day, and it's very interesting on that.

Can you give us a little bit more color on, first, after launch, in terms of how the revenue opportunity for Illumina flows through?

And I guess, as you look at '18, does it begin to start to contribute to the trial -- to the top line of your company?

Derik, it's Marc here.

So it's only been a week or so.

It's a little too early to tell.

We're not changing our expectations for revenue for Helix this year compared to our previous comments around that.

But 2018 will certainly be an interesting year.

We got to prove out the model here with this launch and show that the sequence ones, clearly, many times, model really works.

And we believe it does, and that's why we set up Helix and we're very pleased with the launch.

But too early to see how the traction is impacted yet.

There's another launch in the Bay Area this week, and we'll continue to monitor progress over time, and at the appropriate time, report out on that.

And the next question comes from Ross Muken from Evercore ISI.

So maybe just sticking on some of the anecdotal customer comments because, Francis, that was super helpful.

Can you help us sort of understand the discussion around elasticity, particularly on the high end as we're sort of approaching this S4 launch?

What's the sort of feedback you're getting from some of the larger clinical customers?

How are -- at a country level, some of the PopSeq studies sort of progressing?

How are we just thinking about as that price point's about to inflect again, the sort of preparation or the sort of conversations maybe you're having with some of these potential customers and how has that kind of evolved since maybe the initial launch in the beginning of the year?

Sure, Ross.

So the feedback we're getting from customers, the high-throughput customers, is very positive and generally consistent that they are looking at NovaSeq as fundamentally enabling them to do higher-intensity sequencing.

So whether it's the move to genomes, much deeper exomes; or high-depth oncology, whether it's liquid biopsy or solid tumor screening.

So we're definitely having the conversations with those customers who are looking at NovaSeq as enabling those high-intensity sequencing applications, and we're very encouraged about what that means in terms of the amount of sequencing that we expect to be generated through the elasticity enabled by NovaSeq.

We're also seeing NovaSeq help push forward some of the population sequencing discussions we've been having around the world.

Those are long-term endeavors, and some even working for a little while.

But I think at least in a number of cases, it's pretty fair to say that NovaSeq is catalyzing those conversations to move more quickly than they were before NovaSeq.

Now that's also dovetailing with some good results that are now coming out of GeL and other countries talking to GeL and getting more clarity around what it could do for them.

And so it's enabled by NovaSeq, but we also have the good timing that GeL is now at a place where they can actually show the output of their sequencing work.

And so that's helping, too.

And how would you just characterize kind of how the ecosystem has formed around some of these projects because again, given where the price point's going, the scope of some of these studies could be quite substantial.

Whether it's from an informatics standpoint or from a sample prep or sample acquisition, I mean, you guys have done a lot to start to help bring in others that bring in other capabilities in some of these large-scale projects.

How would you say that's kind of developed over the balance of the year so far?

Yes.

And I'll say that we've started actually last year, and partially in anticipation of NovaSeq, to be more intentional about working with the ecosystem to make sure that the -- our customers had access to the best end-to-end workflows, whether they were Illumina or not.

And so we have consciously been cultivating more partnerships and deeper relationships with the partners we work with.

So that's going to continue to be part of our strategy going forward.

That will help play out in the population sequencing efforts where we can start to talk about -- first of all, we actually do have a reference architecture that is deployed at GeL.

And while a lot of that is Illumina, we certainly are working with partners at GeL, and we now have a working implementation that we can talk to other countries about.

And as you can imagine, from other countries' perspective, it significantly derisks the deal to have a reference architecture that is already fully implemented.

And a number of them are looking to replicate, in a lot of ways, almost exactly what GeL has done.

So that's been very helpful in terms of creating the ecosystem and then having an integrated platform like GeL where we can demonstrate it working.

And the next question comes from Amanda Murphy from William Blair.

So I just had a quick follow-up to some of the commentary about just how things might trend through the year.

So you talked specific to consumables being a bit ahead of your expectations.

But I think the placement number also came in quite high relative to guidance.

And obviously, you talked about the backlog and working that down.

So I'm just trying to figure out, in context of all the commentary around sort of -- I guess, some of the prior commentary, I think you said you expect high-throughput consumables to be up in the back half.

So is that still the case?

And then just help us think about kind of the pace of maybe placements in light of the revised guidance?

So Amanda, just one clarification to your question.

We did not guide to a NovaSeq instrument number for the second quarter.

So obviously, the 80-unit figure we weren't commenting on whether or not that was higher than -- lower than our expectations in the prepared remarks effectively because it was roughly in line with our expectations.

With regard to the ramp from here, I think you can expect to see a slight sequential ramp in instrument placements.

Obviously, the number of units that we placed this quarter from a shipment perspective was quite meaningful.

So I think you should expect very slight increase from these types of levels.

And lastly, with regard to HiSeq X and NovaSeq consumables, the net of those 3, I think we aren't necessarily guiding to what that will be in any given quarter.

But we believe that in aggregate, HiSeq consumable -- I'm sorry, high-throughput consumables will grow.

There may be some timing of when X workdown occurs versus NovaSeq ramp-up occurs, so that may introduce a little bit of lumpiness in a quarter here and there.

But in aggregate, over a multi-quarter period, we would expect growth in that segment of the consumables market.

Got it.

Okay.

And then I just also had one on the S4.

So appreciate all the commentary there.

But it feels like -- like you said, there are some customers waiting to place orders, to see that,that fits.

But also, it feels like some customers that maybe have placed orders are also waiting.

So I just wanted to see kind of the risk around that.

You're pretty comfortable that the timeline -- or it sounds like you are, but just wanted to clarify.

And then do you think the beta launch is kind of enough to get some of the existing customers that ordered ready to receive the rest of the class?

Yes.

And I'll say, certainly, we believe that there are customers who haven't ordered yet that are waiting for S4.

But as you point out, there are also customers that have made some initial purchases that are looking to S4 to be a reason why they do a broader rollout across their fleet.

And so there will be both types of customers that really sort of ramp up as S4 comes out.

I have confidence in the time lines.

We expect to have early-access customers out in September.

I feel good about that, and we expect to have broad availability starting in October.

And our next question comes from Bill Quirk from Piper Jaffray.

Couple of questions.

First off, do you have any (inaudible)

Bill?

Bill, you're breaking up a bit.

Sorry.

Can you hear me now?

Can you hear me now?

No.

No.

I'll get back in the queue, sorry about that.

Our next question comes from Dan Arias from Citi.

Francis, on the pull-through for the Nova, obviously, the revenue potential for a lab that's fully ramped is substantial.

And I'm sure you don't want to give up an average annualized number at this point.

Though I guess I'd be happy to listen if you did.

But just given that there are a bunch of labs that are in production mode, would you be willing to talk about just where the highest throughput commercial machines are tracking towards from a dollar standpoint?

So the -- obviously, we know, I can tell you today if it was useful, but it isn't.

I can tell you today, we know what our NovaSeq -- our early NovaSeq customers are running today in terms of utilization, but we also know from launching many instruments that the numbers today are not going to be representative of the ultimate pull-through range that you will see.

The first set of customers, you may in fact -- you probably will, in fact, see it go up initially before it settles into a more normal pull-through number.

And that will take 4 to 6 quarters from launch before we get to a range that's actually useful from a modeling perspective.

And so we'll continue to watch for that.

And as we feel like it's settling into a range that's more normal, that's the stage we will come out and share what numbers you should be using from a modeling perspective.

But it's definitely too early for us to call that number yet.

Okay.

And then maybe just following up on Derik's replacement question, I'm curious how you're thinking about decommissions or trade-ins of the X systems.

Could this be a formal trading program?

Or is it -- are you thinking it could be more ad hoc?

Anything on just how that installed base might track over time would be helpful.

Yes.

So initially, when we first launched the NovaSeq, as we typically do, we did have a trade-in program.

And we had -- we offered our customers the ability to look at instruments they had ordered that were in our backlog that they haven't taken yet and use those dollars to switch over from an X, for example, to NovaSeq.

And so we did that when we launched.

Going forward, it's going to be less of a trade-in.

It's going to be more of a baby commission and they buy a new instrument.

And so that was very much a launch effect rather than a sort of a permanent offer.

And so that's how we expected to play out.

And then from a customer's perspective, in a lot of cases, the math is -- it's pretty compelling.

They know what their run rate is.

In terms of cost per GI, they're on one of their HiSeq platforms or in an X, and they know their sample volumes.

And so they can figure out the breakeven point pretty quickly.

And from a logistical perspective, the way we go about removing HiSeqs from the installed base will be the exact same way we go about removing HiSeq X. So that high barrier will be there as well.

So I think what you'll see is, effectively, the HiSeq X utilization decrease in favor of NovaSeq, and it'll take a while before those instruments start coming out of the installed base.

And our next question comes from Puneet Souda from Leerink Partners.

So on applications, Francis, if you could help me understand, any sense of applications that are gaining prominence and leading customers of NovaSeq so far.

I know S2 is the only one that's out there.

So any sense of what applications are getting ground here.

Yes.

So I'll tell you what customers are talking to us about, and I'll share some of what we're already starting to see.

So we are definitely having conversations, like I said earlier, around customers that are looking to do higher-intensity sequencing.

So many conversations are happening around deeper exomes, about moving from exomes to genomes.

We are having conversations about doing much, much deeper sequencing in cancer biopsies, and then obviously, very, very deep sequencing for liquid biopsies.

Those are all very active conversations, enabled again, fundamentally by the price points associated with NovaSeq.

And then we're seeing -- and then across our portfolios, so not just NovaSeq, there are other areas where we're seeing the move to sequencing happen in some areas from arrays, for example.

We see that happening in the PGS space as we move from microarrays to sequencing.

And so those are some of the areas where we're seeing newer conversations about higher-intensity sequencing.

Okay.

Got it.

And a question about the bottom 600 cohort of your 800 customer base on the high-throughput end.

Just trying to understand,-- when speaking to the academic and core lab operations, are you getting any sense that the lane splitting manifold device and S2 might just be good enough for them?

Or are those -- are you seeing those customers still wanting an S4?

There are customers in both camps.

I mean -- there are customers, depending on the scale of your lab and the number of samples that you're running, there are customers where S2 will be a perfectly fine flow cell.

And we expect to have a set of customers like that.

And S4 is, again, very, very well suited if you're doing genomes or deep exomes.

And so for other applications, I fully expect S2 to be an in-demand flow cell.

And our next question comes from Dan Leonard from Deutsche Bank.

I was hoping you could elaborate more on what you're seeing in the consumer market for microarrays and how your expectations change in your 2017 guidance for microarrays in the consumer market.

Yes.

We came into this year with momentum in the consumer market, and that momentum has continued to accelerate.

And so we talked about the fact that our shipments to our direct-to-consumer customers have doubled from last year.

That's driven by the strength in the ancestry market, driven by the strength in the health trait market.

And that looks like it's going to continue, that momentum.

Now over time, it's possible that they continue -- these customers continue to offer array-based services, and we just continue to see that momentum build.

But it's also possible that over time, some of them decide to move to sequencing.

And so that will be good for the industry.

That will be good for Illumina.

But it means that we will see some revenue shift away from arrays into sequencing.

And just one thing to add to that, Dan.

The array services line was not necessarily a meaningful driver of the beat in second quarter.

Okay.

I'm just looking at it versus Street models.

It seems like it was a big delta.

Can you quantify...

Well, I wouldn't say that -- the Street model number isn't necessarily clear because some of the analysts do model array services in the array number, and some do model it excluding.

So outside of -- just in services alone.

So I think it's actually hard to bifurcate that number based on the differences in how people model that.

Okay.

I guess then, my follow-up, could you clarify, of the increase in revenue guidance, how much of that was due to arrays versus sequencing?

A very small part was arrays.

I mean, the majority was not.

The vast majority was NovaSeq.

And the next question comes from Isaac Ro from Goldman Sachs.

Just a couple of follow-up items.

One on the microarray business, it was up pretty strong double digits.

Interested in how much of that you think was share versus market growth, and what's baked into your outlook for the rest of the year?

So we have really good share in those markets.

I'd say the big driver there is just market growth.

I mean, we are seeing really strong growth, again, in the ancestry portion of the market, in the health trait portion of the market, the 23andMe and sort of their ability to offer tests.

A part of the FDA guidance has been really good for them.

And so in general, we're seeing really robust demand in the consumer market, and that's driven the demand from our direct-to-consumer customers.

Great.

And then as a follow-up, I know it's early days, but for liquid biopsy, I'm curious if you could quantify either as a percentage or a diagnostic revenue or as a percentage of new customers or some metric to give us a sense of how significant that part of the market is for your business at this point.

And then maybe as we look towards the end of the year, the ramp.

Because it's a very fluid market.

I'm interested in just a snapshot today and where we're going in the next 6 months.

Yes.

And it's hard for us to give you the size of that for us.

What I'll say is, look, the -- some of the drivers there are just on the large studies that are happening.

So not necessarily the diagnostic portion of the market.

Let's say, there's some of that.

But some of the big drivers have been just some very large studies that are playing out in liquid biopsy.

And that portion of the market is captured in the oncology testing segment that we highlighted in the script there, Isaac, just so you have that context.

And the next question comes from Steve Beuchaw from Morgan Stanley.

It's Steve Beuchaw here.

So a question on the second half outlook.

Sam, there's just a lot of moving parts in the second half outlook with all the variables that people have done a great job of spiking out here in the Q&A.

Could you give us any sense for what seasonality on the top line looks like?

There's clearly acceleration implied in the guidance.

Is it a small step-up to Q3, a big step-up to 4Q?

Any help there would be very much appreciated.

And then I have just one follow-up.

Yes.

I'll start, and then, Sam, maybe you can comment, too.

I'll say on the one hand, there are a lot of moving parts, but I'd say a really big driver here of the year is NovaSeq.

And so that's sort of a very big theme even in the second half.

And I said in Q3, we're going to get expand our manufacturing capacity a little bit ahead of where we were even in Q2.

And so that allows us to ship more NovaSeqs and then start to sort of address some of the backlog as well.

And so I think while there are a lot of moving parts, I think there's sort of a really big theme around NovaSeq that's pretty straightforward.

So Sam...

Yes.

I think that's pretty much it.

We do have a ramp-up in the second half, and I think it's really driven by a couple of factors.

One is the ramp-up in NovaSeq as -- NovaSeq instruments, as Francis mentioned, as well as the ramp-up of NovaSeq consumables.

And partially offsetting that, there's going to be a reduction in the HiSeq and HiSeq X placements and the offsetting declining consumables as well.

So you have these moving parts, but outweighed by the growth related to NovaSeq instruments and consumables.

Okay.

And then just within that, could you give us a sense for how much, if any, your expectations for the impact of currency have changed, top line and bottom line, and that would be it for me.

Yes.

For the full year, Steve, there's really no material impact, both on revenue and EPS, from currency movements compared to our previous assumptions and overall.

And our next question comes from Jack Meehan from Barclays.

I wanted to dig in a little bit more on the service revenue growth this quarter.

Could you walk through some of the buckets and where you're seeing the strength really drive some of the upside there?

Yes.

So there are probably a couple of buckets.

So in services, we already talked about the strength in the microarray services, driven in large part by our direct-to-consumer customers.

So that's sort of one big bucket of strength there.

And then obviously, there's just service contracts associated with more instrument placements.

And that's been another driver there.

And both are around the same rough dollar size in terms of dollar growth year-over-year.

Obviously, arrays are also from a lower base.

Great.

That's helpful.

And clearly, there's a lot going on with the Illumina thesis at this point.

But we're getting closer to when you had originally talked about launching the Firefly sometime in the second half of 2017.

I was wondering if you could give an update on that product launch timeline.

The team's continuing to make progress on the sequencer, and so we are on track for end of the year, sort of early 2018 launch on the sequencer side.

And then after the team gets that out, we'll focus on the library prep instruments.

So good progress.

No real news at this point, though.

And the next question comes from Tim Evans from Wells Fargo.

I wanted to come back to the elasticity question on the NovaSeq.

And I do appreciate the anecdotal information you shared.

If you just drill down into the customers who are specifically new to sequencing, what kind of work have you done there to try to figure out how many of those there might be out in the market?

We have done our own internal modeling around what we expect the addressable -- the overall addressable market is for sequencing and for the different parts of our sequencing portfolio.

Those aren't numbers that we have shared externally, but we certainly use them from a modeling-the-demand perspective.

I'd say though, in Q2, the demand we saw from the new to sequencing and the benchtop-only customers exceeded our internal expectations from that part of the market.

If you look back at HiSeq X, there was an interesting experiment there that showed when you hit certain price points, a lot of new customers that you wouldn't have anticipated come out of woodwork, and X was a great example of that.

NovaSeq, obviously, with the S4, starts to also address new price points, too.

So who knows where it's going to go.

There's always going to be some new customers and new applications we really haven't been able to anticipate.

And your next question comes from Bill Quirk from Piper Jaffray.

So just thinking about the 1/3 of orders for NovaSeq that went to either new to sequencing or some of the benchtop customers, can you just remind us, how many benchtop-only customers you have, and how should we be thinking about them longer term transitioning into the NovaSeq family?

In terms of benchtop-only, Bill, I don't have that number off the top of my fingertips.

Obviously, with more than 2,000 NextSeqs and 5,000 MiSeqs out there, obviously, there's some multiunits in there.

But there are quite a number of benchtop-only customers.

And I don't think it would be appropriate to expect all of them to move up the kind of complexity curve to a high-throughput instrument.

But obviously, when we do see new to sequencing customers come in, we tend to see them come in through the benchtop.

And as they've grown sophistication, a portion of them do go to the high-throughput market.

Obviously, this is a very nice opportunity for us, but it's just another demonstration of the elasticity that NovaSeq can bring to this market by encouraging those customers who may not have today the dollars or the complexity to encourage high-throughput sequencing with this product they know can bring forth projects and incremental dollars to do so.

Yes.

one exercise you can do is, as Rebecca said, we said there are 800 HiSeq customers out there.

There are about 2,000 NextSeq instruments out there and about 5,000 MiSeq instruments.

So you can do the math that says, well, even if there was a full overlap, that means that there are 1,200 NextSeq customers, at least, that are -- that don't have a HiSeq, right?

So you can start to do some sketching around the boundaries of what that number could be at the low end.

That's very helpful.

And then just thinking about, I guess, some of the long-term margin profile for NovaSeq.

As we think about more the HiSeqs coming out of the overall installed base, and obviously, over time, transitioning to more NovaSeqs, can you just comment about things like service expense and should we expect -- service margin rather, and should we expect, again, longer term that we could see some leverage there for Illumina?

I think there could be a marginal impact there if you're servicing 3 different lines, if you're looking at the HiSeqs, the X and the NovaSeq, and eventually, it sort of converges to NovaSeq, it could be some marginal improvements in terms of margin.

But I'd say those are marginal.

And our next question comes from Paul knight from Janney Montgomery.

Rebecca, could you talk to the level of sales of sequencers to GRAIL?

Yes.

It's a fair question, Paul.

Obviously, GRAIL is now a customer of ours, and like any other customer, we're not actually able to provide specific sales figures to them.

Any sales to GRAIL would be included in the commentary around oncology testing.

The level of microarray services sales was high.

What were the exact drivers around the services number in the quarter?

The big drivers were our direct-to-consumer customers.

So they sell into the ancestry space, into the health trait for the consumer market, and they're doing well.

And that's driving the number that you see.

Meaning your shipment of chips or your analysis of the content?

In some cases, we run the service for them, and so we process the arrays for them.

And that shows up in the services line.

(Operator Instructions) I will now turn the call back over to Rebecca Chambers for final remarks.

Thanks, Adrienne.

Thank you, everyone, for joining us.

As a reminder, a replay of this call will be available as a webcast in the Investors section of our website as well as through the dial-in instructions contained in today's earnings release.

Thank you for joining us today.

This concludes our call, and we look forward to our next update following the close of the third fiscal quarter.

Thank you, ladies and gentlemen.

This concludes today's call.

Thank you for participating, and you may now disconnect.