Nutanix Inc (NTNX) 2017 Q3 法說會逐字稿

Good afternoon. My name is Mike and I will be your conference operator today. At this time I would like to welcome everyone to the Nutanix Q3 fiscal 2017 earnings call. (Operator Instructions).

I will now turn the call over to Tonya Chin, Senior Director of Investor Relations. You may begin your conference.

Good afternoon and welcome to today's conference call to discuss the results of our third quarter of fiscal year 2017. This call is also being broadcast live over the Web and can be accessed in the Investor Relations section of the Nutanix website. Joining me today are Dheeraj Pandey, Nutanix's CEO; and Duston Williams, Chief Financial Officer.

After the market close today, Nutanix issued a press release announcing the financial results for its third quarter of fiscal 2017. If you'd like a copy of the release, you can access it in the Investor Relations section of the Company's website.

We would like to remind you that during today's call management may make forward-looking statements within the meaning of the Safe Harbor provision of federal securities laws regarding the Company's anticipated future revenue, gross margin, operating expenses, net loss, loss per share, free cash flow, business plans and objectives, product features, technology that is under development, competitive and industry dynamics, changes in sales productivity, expectations regarding increasing software sales, future pricing of certain components of our solutions, our plans regarding the adoption of new revenue recognition standards, potential market opportunities, and other financial and business related information.

These forward-looking statements involve a number of risks and uncertainties, some of which are beyond our control, which could cause actual results to differ materially and adversely from those anticipated by these statements.

These forward-looking statements apply as of today and you should not rely on them as representing our views in the future. We undertake no obligation to update these statements after this call.

For a more detailed description of these risks and uncertainties, please refer to our quarterly report on Form 10-Q for the second quarter of fiscal 2017 filed with the SEC on March 10, 2017, as well as our earnings release posted a few minutes ago on our website. Copies of these documents may be obtained from the SEC or by visiting the Investor Relations section of our website.

Also, please note that unless otherwise specifically referenced, all financial measures we use on this call are expressed on a non-GAAP basis, and have been adjusted to exclude certain charges. We have provided reconciliations of these non-GAAP financial measures to GAAP financial measures in the Investor Relations section of our website and in our earnings press release.

Now I will turn the call over to Dheeraj Pandey, CEO of Nutanix. Dheeraj?

Thank you, Tonya, and hi, everyone. Thank you for dialing in. I'm very pleased to report that we had a great quarter with record revenues of $192 million that were above Street consensus and our guided range, growing 67% year-over-year. We were also pleased to see our North America business return to strong growth in addition to continued solid performances from EMEA and APAC.

Third-quarter billings were $234 million, up 47% year-over-year. Gross margins came in slightly above expectations, and our EPS performance was better than our guided range, beating consensus by $0.03.

Our focus on driving sustained growth continued during the quarter, driven by strong repeat business, which represented 71% of our third-quarter bookings, also owing to strong momentum in large deals. This demonstrates steady progress on deepening and extending our existing relationships with customers, particularly in the Global 2000.

Our concentration on increasing our number of large deals paid off with two deals greater than $5 million in the quarter, and 34 deals greater than $1 million. Further, we reached a major milestone with our largest customer, crossing $50 million in lifetime sales and still growing strong.

And while we made great progress expanding further into existing accounts in Q3, we also sold into 790 new customers in the quarter, including nearly 50 new Global 2000 customers. This brings our total customer count to over 6,170.

Our OEM partners, Dell and Lenovo, contributed well to our new Global 2000 customer acquisitions, complementing a strong number of additions on the NX platform. HPE hypervisor adoption continued to increase, with 23% of nodes sold using our own built-in hypervisor, up from 21% last quarter and 9% a year ago, based on a rolling four-quarter average.

Our full stack operating system story -- including our AHV hypervisor, our operations management software, Prism Pro, and our highly portable and -- multi-hypervisor, multi-hardware stack -- continues to be a significant differentiator in our enterprise sales motion.

Last quarter, we discussed some changes we're making in our sales organization to sharpen our focus on larger accounts. This process was a necessary part of our sales team evolution, and critical in our next phase of growth as we target our next $1 billion in billings. We're pleased to report that we made very solid progress during the quarter with strong execution in hiring and alignment across our sales team, particularly in North America where most of the changes were made. We have hired several new team members who will focus exclusively in very large enterprise accounts. And we will continue to hire in support of these sales objectives in the coming months.

In addition to executing on our planned adjustments to the sales organization, we further expanded our market opportunity while also extending hardware choice to enterprise cloud builders with several product announcements. Our software was initially sold with Nutanix sourced industry standard x86 servers. Since then, we have steadily increased hardware platform choice for our customers, adding Dell, Lenovo, and Cisco UCS C-Series servers.

During the quarter, we continued to deliver on this commitment to offering customers the broadest choice of hardware for their enterprise cloud deployments and announced three new supported hardware platforms.

First, we announced term-based licenses of our Nutanix Enterprise Cloud Platform software for Hewlett-Packard Enterprise's ProLiant and Cisco UCS B-Series blade servers.

We also signed a multi-year agreement with IBM to combine our enterprise cloud operating system with IBM's Power Systems to deliver the industry's first non-x86 hyperconverged solution targeting cognitive applications and big data workloads in large enterprises. The solution will be sold through IBM and their partners exclusively and feature AHV hypervisor.

These new supported hardware options will broaden the deployment potential of Nutanix beyond our current hardware partners to a large population of HPE ProLiant and Cisco UCS servers already used inside enterprise data centers, as well as to highest-end power-based applications in the large enterprise.

Operating systems are all about ubiquity and APIs. Our software now runs on a large population of Intel x86 servers, IoT-scale Intel servers in palm-sized form factors, ruggedized military equipment like Crystal and Klas, and soon-to-be-released IBM Power servers. We have made great strides with APIs to build an ecosystem of networking, storage, security, and management software partners, many of whom will be part of our .NEXT user conference in Washington, DC, on June 28.

Today, approximately 50% of our Q3 new customer workloads are considered Tier 1 business-critical applications, including Microsoft SQL Server, Oracle, SAP, Microsoft Exchange, Hadoop and Splunk.

Legacy vendors, typically our biggest naysayers, would want to think otherwise. But any new architecture that changes the consumption model has had a rite of passage similar to ours. Our solutions have strong credibility in the enterprise where a growing number of Global 2000 customers, today numbering approximately 520, continue to expand their mission-critical workloads with us.

We frequently see customers start with one workload as a trial or beachhead, and quickly add workloads over time. A great example of workload progression within a customer was a large deal in the quarter, a very large US retailer doing well over $15 billion in annual sales in over 1,000 stores through the United States.

Our relationship with this customer is a perfect illustration of how we expand in an account over time. Two years ago, we began our relationship with a large-scale virtual desktop infrastructure which essentially takes Windows desktop to a cloud environment. After proving the ease of use and performance of our solution, this customer quickly began to standardize on our platform, adding new workloads every few months. Today, Nutanix is deployed across all their major data centers, and the continued purchase pattern culminated in one of the largest commercial deals in the history of our Company in Q3.

Another great example of continued workload progression is with a global advanced wound care company with whom we signed a greater than $1 million expansion deal in the quarter. This customer started using our solution for its remote office applications and then added to their Nutanix cluster for disaster recovery. Over time, this customer has grown their engagement to include our full stack solution, including AHV for their core [sole] virtualization.

Not another very large deal in the quarter was with a Global 2000 customer that is one of the largest cell phone carriers in the US, with over 50,000 employees and over $30 billion in annual revenue. This was an expansion that encompassed a number of workloads, including a large-scale VDI deployment and a large big-data workload, driven by in our Web scale architecture and the cloudlike simplicity of our solution.

We also had great competitive win against legacy converged infrastructure in a multimillion-dollar expansion deal with a Global 2000 customer based in our EMEA region, a multinational asset manager with over 4,000 worldwide employees whose opportunity encompassed both a VDI expansion and a large big-data cluster.

In addition to increasing our footprint within many large customers in the quarter, we also signed a number of large new customers in the quarter, with many new deals greater than $1 million.

A good example of this was our largest new customer in Q3, a Global 2000 customer that is one of the largest global stock exchanges based outside of the US. Our strong history of quickly extending customer engagements beyond the first workload gives us confidence that we'll grow this and many of our new customer relationships over time.

We also continue to see customers recognizing that certain workloads, having been moved to the public cloud, are better managed on-prem. For example, in Q3, the largest e-commerce company in Indonesia that had formerly elected to run their marketplace in the public cloud moved its mission-critical workload back, for cost and performance reasons, on-prem to our enterprise cloud powered by AHV.

The competitive landscape has grown over the past few quarters, but we continue to perform very well against competition, with strong win rates. For example, during the quarter we signed a large deal with a Global 2000 customer that is one of the largest European-based brewing companies and a household name. This customer selected Nutanix for their Oracle workload in a competitive win, with AHV being a very significant factor in the customer's decision. We had a very similar win with an Oracle workload in AHV at one of the largest medical supply companies in the US.

Our OEM partner, Lenovo, contributed to a Global 2000 penetration, closing a new logo together with us, marking their largest-ever deal with Nutanix. The customer, a large US-based brokerage firm with over $3 billion in annual revenue, selected Nutanix in a competitive win against the traditional hypervisor stack.

Scalability remains a significant differentiator for our operating system, which shows in our ability to manage and grow very large customers. We are very pleased with the number of customers that have grown their footprint with us and standardized their data centers around our operating system.

In fact, we have 269 customers that have purchased between $1 million and $3 million, 40 customers that have purchased between $3 million and $5 million, and 32 customers that have purchased over $5 million of lifetime to date.

Now let me share one final example. We have shared that choice of hardware for our customers is a significant part of our growth opportunity. In Q3, we closed a large software deal running on Cisco with a Global 2000 customer offering wireless and ISP services to over 50 million customers in the US. The new solution, which will run Nutanix on UCS blades, will allow the customer to reduce the size of its cluster from 700 to 100 nodes, utilizing our technology, delivering significant ROI for them.

The enterprise cloud market is growing faster than anyone imagined. Gartner estimates in its May report -- Forecast Analysis: worldwide integrated systems, 1Q17 update, hyperconverged integrated systems -- that the hyperconverged market will grow to approximately $8.5 billion plus in 2020.

We believe this predicted growth is reinforced by the fact that Nutanix solutions are powering more and more workloads in our customers' environments, including Oracle, SAP, and Microsoft business applications and databases. This is a very healthy backdrop for us to continue our strong growth by executing on our strategy. While many of our me-too competitors treat the form factor of hyperconvergence as a destination, we are focusing on an operating system which will holistically re-platform the enterprise data center. Top-down consumption of infrastructure is key to our vision, which is why we are so focused on developer interfaces such as self-service portals and application-centric automation.

We continue to build on ubiquity by offering customers choice of hardware, choice of hypervisor, and choice of public cloud providers for secondary storage, all managed by a single controlled plane of Prism.

Building an operating system is a journey; and no more than one or two are successful each decade, which requires an immense focus on applications, interoperability, performance, security, automation, and reliability. And to make it all ubiquitous -- that is, location-agnostic -- is the biggest engineering challenge of them all.

We remain extremely focused in investing in customer service and product engineering to maintain our lead as we re-architect the enterprise. The hybrid enterprise is the final frontier.

With that, I'll turn the call over to Duston for a more detailed review of our Q3 results. Duston?

Thank you, Dheeraj. For the 23rd consecutive quarter, dating back to our first quarter of shipments, we delivered record revenue in Q3 above our target range. This performance, along with gross margins that were slightly better than expected, resulted in non-GAAP EPS that also exceeded our expectations.

Revenue for the third quarter was $192 million, growing 67% from the year ago, and up 5% from the previous quarter. We billed $234 million for the quarter, representing a 47% increase from a year ago, and a 3% increase from the prior quarter.

We also rebuilt a significant amount of backlog during the quarter. Notably, our Dell OEM business accounted for over 20% of our ending backlog. On a sequential basis, our Dell bookings declined in Q3 by roughly the same percentage decline that we experienced in Q3 of 2016 as Dell's fiscal year-end occurs in January. Our Lenovo bookings increased sharply in Q3.

As Dheeraj mentioned, although we have some required execution ahead of us when it comes to sales segmentation, we were very pleased with the rebound we experienced in our large deal activity. Again, just for clarity, we define large deals as customers purchasing more than $500,000 in any given quarter.

As you may recall, during our Q2 earnings call, we mentioned the following three data points. First, we stated that over the last four quarters, deals exceeding $500,000 averaged about 45% of total bookings. Second, we noted that in Q2 we performed somewhat below this level, primarily in North America. And lastly, we predicted that the North America productivity would rebound over the next few quarters as a result of our planned sales adjustments.

I'm pleased to report that not only did big deals as a percent of total bookings rebound back to around historical levels in Q3, but specifically the North American productivity recovered nicely during the quarter. In fact, the North American sales rep productivity for ramped reps was the strongest performance we have seen since fiscal Q4 2016.

As you may recall, we also committed to provide investors with some additional data points to illustrate our progress in growing our large deals as a percent of our total business. The following data points shed some light on our Q3 large deal activity, Global 2000 customer traction, as well as additional insight into a couple of our software-only deals. You should not expect this level of detail on an ongoing basis.

First, let me review some of the details on our large deal activity. In Q3, we executed 13 deals greater than $2 million in bookings for a total of $45 million in bookings compared to only four deals greater than $2 million in the second quarter. Notably, of these 13 deals, 10 of them either included no VDI, or included other workloads in addition to VDI.

Of the 13 deals greater than $2 million in Q3, eight were Global 2000 companies and four were large government agencies. Three of these deals were to new customers, and the two largest deals approximated $7 million each. Comparatively, there was only one Global 2000 customer included in the Q2 deals greater than $2 million in bookings.

Now, a little bit further insight into our Global 2000 activity. We had a record Global 2000 quarter in Q3. Our Global 2000 bookings in Q3 were 50% greater than in any previous quarter. Global 2000 bookings as a percent of total bookings was the highest it has been in over two years.

In Q3, 20 Global 2000 customers made purchases of greater than $1 million, versus 10 in Q2. Of these 20 Global 2000 customers that made purchases of greater than $1 million, 70% of them either included no VDI, or included other workloads in addition to VDI.

Almost 50% of our existing Global 2000 customer base made repeat purchases in Q3, illustrating the continued success of our land-and-expand strategy. And in Q3, the average repeat purchase amount for a Global 2000 customer was more than 3X of that of our non-Global 2000 customers.

We continue to focus on increasing software sales as a percent of our total product sales. In addition to generating software bookings from our two OEM partners and from premium software additions, I'll next provide a few highlights regarding a couple of our other software-only transactions in Q3.

We completed two ELA deals with Fortune 500 companies, each with 100 or more nodes, and both representing repeat ELA purchases. We booked many term license deals for the Nutanix software running on Cisco UCS. And we completed an initial $800,000 software-only deal running on a specialized defense-related hardware platform.

As I mentioned earlier, we are very pleased with the Q3 performance surrounding our large deal activity as well as our Global 2000 business. We realize we have more work ahead of us, but it is very clear that our focused efforts around large deals and driving further Global 2000 penetration is yielding positive results.

As expected in Q3, our percentage of international bookings decreased from 48% in Q2 2017 to 38% in Q3 2017 compared to 33% in Q3 2016. Our bill to revenue ratio was 1.22 times, which is slightly lower than the estimated range of 1.25 to 1.3. Based on our current projections, we expect Q4 to be around 1.25 times.

Our gross margin for the quarter was 58.4%, which was slightly higher than our guidance and compares to 62.5% in the year-ago quarter and 59.8% in the prior quarter.

We were pleased with some limited success with the price increases we enacted in Q3 to counteract DRAM cost increases. As expected, DRAM and NAND-related SSD costs continue to rise, causing our product costs to increase by about 8% during the quarter. Looking to Q4, we expect significant additional DRAM cost increases while SSD cost increases should start to abate.

I'll touch on margins a bit more when I review the guidance for Q4.

Our operating expenses were $171 million and were in line with our guidance. Although the anticipated headcount increases in Q4 and Q1 2018 will be lower than historical rates, we want to be clear that operating expenses are expected to increase by a minimum of $10 million per quarter over the next few quarters.

With an eye focused on continued top-line acceleration, additional spending will be funneled to demand generation investments with a majority targeting large customer opportunities.

We had a non-GAAP net loss of $61 million or $0.42 per share.

I'll next provide a few balance sheet highlights along with a few other key performance metrics. We closed the quarter with cash and cash equivalents of $350 million. That was down from $355 million in the prior quarter. DSOs on a straight average was 79 days, three days higher than the 76 days reported in the prior quarter. The weighted average DSO reflecting the period and average receivables outstanding was 23 days in Q3 versus 24 in the prior quarter.

We used $16 million of cash flow from operations. As you may recall, the Q2 cash flow from operations benefited by $10 million from ESPP contributions. The Q3 cash flow from operations was negatively impacted by the same $10 million, as shares were purchased under the ESPP. Year-to-date fiscal 2017, we have generated a positive $8 million of cash flow from operations.

We used $29 million in free cash flow during the quarter, although once again this was impacted by the $10 million of ESPP funding outflow. Software as a percent of total bookings based on a rolling four-quarter average increased to 16%, up from 15% in Q2.

I'll now touch on our guidance for the fourth quarter: revenue to be between $215 million and $220 million; gross margins of approximately 58%; and a per-share loss of around $0.38, using weighted average shares outstanding of approximately 152 million.

Regarding gross margins, DRAM costs will once again increase substantially in Q4. All things being equal, this cost increase would push margins below their current level. However, we've set a target to hold margins at 58%, and therefore will alter our mix of software as needed to maintain this margin level.

It's important to note that over the 12-month period ending July 31, 2017, our expected DRAM cost will have increased by nearly 100%, resulting in a 6 percentage point impact to gross margins. Once DRAM costs start to decrease, we expect gross margins to rebound back to the 60% level.

And then, lastly, regarding the new revenue recognition standard, again known as ASC 606, it remains our preference to early adopt this new standard starting in August 2018, which is our fiscal Q1 2018, pending the resolution of a few open items. At that time, we expect to be recast our historical financial statements. And in Q3 under the new 606 standards, based on a tops-down estimate, gross margins would have been approximately 200 to 300 basis points higher than we reported under the current existing revenue standard.

Operator, if you'd now open the call up for questions, that'd be great. Thank you.

(Operator Instructions). Kulbinder Garcha, Credit Suisse.

Just a couple of questions, please. First of all on gross margin, Duston, as we go forward I understand the cost increases impacting the fourth quarter. But given the attach of your software and the software rise in the billings, should that start kicking in and becoming meaningful as we get into next fiscal year? Or how should we think about the gross margin trend beyond just the next quarter?

And then the other question is maybe a broader one for Dheeraj. Just in terms of Dell, just remind me how big it is, just in revenues or billings in absolute. And how is that relationship evolving, especially as the hypervisor you guys are selling (technical difficulty) to gain traction? Many thanks.

This is Duston, so I'll take probably both those, I guess. On the Dell piece, we haven't broken that out, typically. We said back in the IPO it had never been bigger than 15%. It's a bit lower than that today as a percent of bookings. And we've got Lenovo that is also starting to accelerate there also, so we've got those two.

And then to your point of all software here, Dheeraj mentioned HP, which we will get up the end of the year here, and then additional server from Cisco that we're now qualified on. So there's lots of opportunities here to continue that software increase.

Also I think just one thing to add on Dell is Dell is an entire group called CPSD, Kulbinder. And our software with their hardware is a significant chunk of their annual sales quota. Dell is morphing its business from traditional three-tier to the new architecture. And the three operating systems -- VMware, Nutanix, and even Microsoft Azure stack -- are all relevant go-to-market motions for them.

At the end of the day, we control our own destiny with our appliance. And we compete and cooperate with Dell on a deal-by-deal basis and ensure that we have multiple routes to market with software ELAs and subscription for SP, Lenovo HX, Cisco UCS; and, going forward, HP and IBM as well.

And just to clarify, Duston, on the point on -- you said it's 15% or lower than 15% of bookings. That's a comment just specific to Dell. And your hope is, as these other OEMs ramp up, they're just -- your OEM revenue is more diversified in 12 months' time, let's say. That's how we should think about it.

Yes. Yes, that's fair. Now, again, the only thing we'd said, in the past it's never been higher than 15%, and it's a bit below that today.

On the gross margin question, Kulbinder, good question. As we go forward, two things happening; obviously sooner or later, DRAM costs will turn back in our favor. So at some point, we'll have that benefit. Today we are still fighting it. And it continues to be a headwind there, although we've taken stance on margins, as I said. And we're just going to basically alter our mix of software to, at a minimum, try to get to the 58% we guided.

But going forward, we should have the DRAM costs coming back in line. Your point about more and more software is valid. And then the model gets pretty interesting as we adopt 606. Because the flexibility, again, from a software and a margin perspective, becomes much greater; and it will allow us even to play off additional margin growth with top-one growth.

Again, today, when we do an ELA, that all gets deferred and gets routable over the support period. Under 606, the license piece of that software will be immediate hit to revenue and gross margins along with the OEM revenues that we have today from Dell and Lenovo.

And just in the near term, have you changed your approach to hardware pricing? My understanding always was that as your software content builds, you would continue to fairly aggressively price your hardware to drive adoption. Just in the near term, have you held back a little bit on that just -- and is that in any way impacting sales or billings (multiple speakers)?

Not significantly, no. We've got a lot of flexibility in the model to do what we need to do there. We may tweak it a little bit this quarter for the 58% target, but it should not be substantial there.

Okay, thank you.

And again, that flexibility just ratchets it up quite a bit once we adopt 606.

Jason Ader, William Blair.

A quick verification and a question. Can you tell us how Dell did sequentially, and was it a slight headwind? That's the clarification. And the question: talking to investors in recent weeks, it seems there is a growing angst on the TAM ceiling for HCI; that HCI is struggling to move upmarket because of performance challenges in mixed workload and larger-scale environments. Where do you think this kind of story is coming from, and can you help address that concern?

Yes, I will take the second question.

But do you want to take the first question on Dell sequential from their Q4 to Q1?

Yes. If you look at the -- and I mentioned this earlier -- if you look at the -- in our Q3 last year, what Dell defined as a percent of our bookings in their bookings there, the decline quarter-over-quarter this year as compared to last year was pretty much exactly the same.

And I think maybe I'll just head this question off because we will have it in the future. And we will open up a little bit here, just because when we file our Q, everybody typically looks to the receivable balances magically under -- it happens to be on how it ended up this way, but customer D. And any time, it's over 10% -- it's shown in the Q there.

Last quarter, I don't know the exact number; I think it was around 11%. This quarter, I'll just tell you what the number is because the questions will come up once we file the Q. It was roughly around 9%, so not too much difference there. It's a down quarter.

A couple things you got to consider there also: when I mentioned backlog, we had built a significant backlog -- again, 20% of that backlog related to Dell. So obviously if it's in backlog it hasn't been billed and it hasn't gotten into receivables. So I think you need to take that into consideration.

And then just quarter-over-quarter of the growth in the receivable balance itself this quarter grew roughly around 12% or 13%. I think if you go back to last quarter, it only grew maybe 3% or 4%. So just some additional insights there to put that in perspective.

Helpful, thanks.

Thanks, Duston. And on the TAM question, Jason, think about Gartner in the last probably year, year and a half, has already morphed its MQ from converged infrastructure our integrated systems to now hyperconverged integrated systems. The TAM has gone from $3 billion to $6 billion to now $8.5 billion.

What's really happening is that the incumbent legacy vendors, who are doing controlled cannibalization of their legacy business -- they are the biggest naysayers of this re-architecture. And they proliferate their naysaying to analysts and media. In my script I've talked about our ability to scale within the enterprise.

Just to reiterate, we have more than 265, almost 269 customers that have purchased [who are] $1 million and $3 million; 40-plus customers that have purchased between $5 million and $5 million; and 32 customers that have purchased over $5 million with us, lifetime.

I talked about the customers that have more than $50 million with us. And then many of them, they are above $10 million and $20 million as well. And most of this has actually come in the last 24 months.

If you take a look at JPMorgan's CIO survey 2017, cloud wars, the name is the enterprise strikes back dated May 15, 2017, you will appreciate our seat at the table in the large enterprise. Because there's so much going on in terms of re-architecture in the large enterprise. And TAM is a function of workloads, hardware, platform, diversity, and form factors.

People said similar things about virtualization in public clouds as well. If you go back to my script, we talked about numerous such workloads, most of which are mixed workload deployments.

So, I think, again, it's a rite of passage. There's a perception that the incumbents would actually build here and we just have to plow through this.

Thank you.

Wamsi Mohan, Bank of America Merrill Lynch.

Dheeraj, on the salesforce reorientation, how are you measuring the progress? And how has this reorientation changed the sales cycle, size of the deals in the last quarter? How are you expecting this to evolve over the next few quarters? And I have a follow-up.

Yes, so we made some really good progress in Q3. A lot of the heavy lifting of the design is behind us. We continue to hire into our global accounts team. And also doing some in-hiring, moving sales team members as needed to have them focus on the large customers and large deals themselves. Duston actually [re-treated] a lot of those and I also talked about many of those.

We are focused on three different productivities. And we will go and look at the top of the pyramid, the middle of the pyramid, and the bottom of the pyramid where there's global accounts and enterprise accounts and commercial accounts; and how they are doing, productivity-wise, on a quarter-to-quarter basis.

Okay, thanks, Dheeraj. And Duston, your new customer adds, at 790, it was still pretty good, but was that the level you had expected to add? Or are the sales execution issues impacting that number? Could it have been larger? It sounds like the pace of new adds and sales hires is at least slowing down intermittently, so if you could just provide some context there.

Q3 is always -- as you come off a strong Q2, usually -- and we did 900-plus new customers in Q2. So the 790 was a pretty good performance actually for a Q3 there. We mentioned last quarter that we had slowed down headcount a little bit. We're starting to accelerate that again selectively in the right areas and things like that. But there will continue to be a fair amount of headcount adds as we go forward.

Simona Jankowski, Goldman Sachs.

Your revenue guidance was quite strong. And it sounds like you are shifting to more software-only deals to manage the growth margin situation. And I believe you lose out something like 40% of the revenue when you take those deals, which implies that the underlying guidance is even stronger than it would appear.

So I just wanted to find out what is driving that. And then secondly, you talked about what the upside to growth margins would have been under the new accounting rule. Can you also quantify what the upside would have been to revenues? Thank you.

Sure. Let's see, on the -- yes, your point on the top line and the impact by getting the 58%. You noted in there that I said, if needed, we'd swap to some more software. So there's not a ton that we expect to have to do, quite honestly.

We will see how the quarter progresses with big deals, and bigger deals typically for the most part have higher margins as these bigger customers understand the value of the full stack and what we're trying to do here. So we will see how that progresses, but I wouldn't read a ton into that, that we're really damaging the top line based on holding to the 58%. There may be some, but it is probably not significant from that perspective.

And then pretty much on your other question, it's -- I'm just thinking through it. It should be roughly a one-to-one ratio. I've got to think through that again -- revenue and gross profit there, because it's all software.

Sure. Thank you.

For the most part.

And then the gross margin difference is something we already highlight, right?

Yes.

Alex Kurtz, Pacific Crest Securities.

Just question and a clarification. You guys had outlined getting to roughly one-third of the business, over the longer term, from software and subscriptions, and I was just wondering if maybe you could give us an update on that. Maybe some timing over the next couple of years when that would be reasonable, given all of these new OEMs that you're working with.

And then a clarification around the Cisco business. It sounds like there's a little bit of a momentum here. And obviously they have global scale, so maybe an update about where the Cisco relationship sits today, and where you think it kind of play out over the next couple quarters.

Thanks, Alex. Yes, on the second question of Cisco, I think it's still early days. We are basically acting like a startup because it's a startup idea. And if you go back 10, 12 years, this is what virtualization was facing in terms of headwind, because none of the server guys really wanted to bless virtualization. And then obviously the market forces actually convinced everybody that this is the right thing to do. And in fact it didn't reduce the server sales; it actually increased, because people could do more in the same amount of time.

So I think a very kind of headwind is upon us from companies like Cisco and HP. But we believe that the market forces will actually speak. And we see some really strong momentum there from the market, from the customers who actually very much like Nutanix software running on their current Cisco hardware.

So how things evolve in the coming years, coming couple of years? We'll come back and report every quarter. But on the long-term one-third, one-third, one-third, we still believe that's actually a very feasible and sustainable model for us. I think you started seeing some ELAs. You started seeing Lenovo actually make good progress. These two are very early starters, both Cisco and HP themselves.

And finally, I think IBM could be the dark horse. We just don't know yet because we don't fully understand the IBM ecosystem. So if you put all this stuff together. And then there is obviously software that we are selling at high levels to the stack, like Prism Pro, which is our operations management software, self-service portal.

And then the acquisition we made about almost a year ago of Calm.io, which is automation and orchestration -- all this stuff actually goes together to build our overall software portfolio.

So we're quite hopeful that we'll get to the number which is one-third of our overall business in pure software.

All right, thanks.

Matt Hedberg, RBC Capital Markets.

Dheeraj, tier 1 workloads seemed particularly strong this quarter. Can you give us a sense for what some of those non-VDI workloads are? And maybe what did that tier 1 workload mix look like on a year-over-year basis?

The last question, I don't have the data; we'll probably have to check before we say anything about year-over-year. But I talked about Microsoft SQL server is a pretty big chunk of our deployments of that. And Oracle, as you saw, quite a few of those. We have started to see a good bit of SAP and Splunk as well.

So I think -- on the whole, I think pretty much anything that can be virtualized and has been virtualized in the past, we are seeing all those big virtual machines come on Nutanix. And we also opened up this form factor, which is sort of a storage-only grid that can even work with physical applications that have not been virtualized.

The beauty of the software is that it doesn't care if it's hyperconverged or not. We can also cater to some of these physical applications, and some of our customers are actually using that, too. So I think, all in all, we have really been tracking to about more than 50% of these tier 1 workloads across Microsoft and Oracle and SAP and Splunk. And VDI is about one-quarter of our business now.

That's great. And then maybe, Duston, the color on the 200 to 300 basis point gross margin benefit, had 606 been adopted, was helpful. I'm curious, if we look at that on an EBIT margin perspective, is it roughly the same benefit? Or should it differ from that 200 to 300 bps?

It's roughly -- you've got commissions that would flow in there, I guess. But beyond that, it's pretty much the same.

Got it. Thanks, guys.

Mark Murphy, JPMorgan.

I'll add my congrats. I joined a few minutes late. Please forgive me if you might have already covered this. But, Duston, so I wanted to ask you, just the very slight reduction in the bill to revenue ratio to 1.25. Could you help us understand the underlying drivers of that? And also is there some permanence attached to that as we roll forward and begin to think beyond the July quarter? That's my first question, and I do have a follow-up.

Yes. So on the ratio, mostly that was -- we always have some different mix in support and things like that; that goes up and down. But mostly that was because of the OEM piece. Again, a lot of that ended up in backlog. A good chunk of it, 20% of our backlog was Dell. Lenovo had a piece of backlog in there too. So again, that would have been billed and not recognized, so that would have pumped the ratio up a bit.

Okay. And then as a follow-up, so you mentioned a measurable improvement in the number of large deals in the quarter, particularly in North America. And I'm trying to understand does that trend feel sustainable? In other words, do you think it happened naturally? Or was it more of an effect of deals that slipped out of Q2 and closed in Q3? Or perhaps did you set any special incentives for sales teams to buckle down and focus on larger deals in the quarter?

I don't think there was anything forced there for Q3 itself. If you think about the architecture of our sales motion and realize that that we actually do not start with these 18-month-long sales cycles because the beauty of the product. It's an elastic fabric. You can start at three nodes and the same system can go to 300 nodes over time. [See] our customer was kind of -- put the foot in the door with a $100,000 system and then they buy another $0.5 million, another $1 million. And finally in a matter of 12 months, they might come back with a $5 million deal that didn't take 18 months to close, which is what typically happens in the enterprise selling that in anything that's high-value takes 12 to 18 months.

And in our case, because we have actually broken them out into smaller chunks, existing customers come back with a larger deals because the trust has actually built over time. And, therefore, some of these large deals don't take as long as it does for the old legacy infrastructure.

So I think we saw a lot of that happen this quarter. The existing customers came back with large deals, and it just happened to be fortuitous for us.

Great. And the Dheeraj, one last quick one. Just given Microsoft's increasing emphasis as they are talking about the merits of this intelligent edge concept and forcing a huge audience really to think more about hybrid architectures, is there any noticeable movement back toward private cloud or hybrid cloud that you can notice? Whether it's relating to that or elsewhere. And just have you seen any incremental signs of customers moving some of the steady-state, expensive types of workloads off of Amazon AWS and onto Nutanix?

Yes, I think we've seen quite a few, and especially these apps are not written with scale-out architecture in mind when you needed a single large VM to perform really well or if you needed [provability] and things like that.

So, we have -- in fact, I talked about one of the in my script, as well, a very large e-commerce player in Asia-Pacific that brought all of it back because of -- not just performance reasons, but also cost reasons. At the end of the day, they went there because it was easy to use. But finally when they found something which is just as easy to use on-prem. And the fact that they could just do it with very few people, because one of the big costs of on-prem infrastructure is people costs. And when we take that away and bring it -- make it to be as elegant, then people start to compare the economics.

And we've actually seen the economics to be quite different. And we've started to show that in a lot of our TCO analysis to customers that renting a hotel with predictable workloads could be quite expensive. And you're really overpaying for the optionality of pulling the plug by the hour, which is what a public cloud charges you by, in many cases, actually.

Thank you.

Katy Huberty, Morgan Stanley.

What percentage of your customers are paying for premium software today? And should we expect that portfolio of offerings to expand over the next year, and the penetration of customers or billings to increase?

Katy, if you look at the number of customers, obviously there's a healthy mix of OEMs there as well, which is something that you probably saw in Q2 as well. There were quite a few smaller retail customers who basically essentially become software-only customers for us because of our OEM partners selling them their appliances.

But the ones that we are doing direct are mostly very large customers. And I think getting to software-only deals with them takes a little bit of time because they need to first gain trust that this is indeed a platform they want to roll out for everything.

And we've started to see some early signs in the last six months, I would say, when some of our very large customers are now saying, let's see if you can actually do ELAS with the Nutanix operating system.

But should we expect that portfolio of value-added services, analytics and security, to increase, and more opportunity for selling premium software into the installed base?

That is our hope, actually. In fact, there's a reason why we have such an effort going on, and really looking at micro segmentation and netbook overlays. And a lot of things around orchestration and automation is because we want to go and sell those things as a full-blown, full stack operating system.

All right. And then as it relates to the Cisco and HPE business, without a full-blown OEM relationship, should we think about selling into that hardware as a single-digit percentage of the business? Or is it possible that the channel could help you take that to a more meaningful level?

If you were to ask me as a CEO, I would say it's -- we should expect the conservative. And then I think there's definitely something you said about the channels, because obviously there's multiple places where our software can meet the hardware. The one on the left of the spectrum is where it meets in our factory. The middle one is where it meets in an OEM's factory.

But we can also meet at the channel, which is a very interesting thing that we have been thinking about, especially because the channel also wants to make some money on this, rather than just re-sharing our profits to the OEM partners.

Aaron Rakers, Stifel.

Curious as you look at the IBM opportunity, how you are thinking about the ramp of that OEM partnership. Should we think about something similar to the Dell and now ramping Lenovo relationships? Or do you think that ramp could come on quicker?

Thanks, Aaron. You know, I think it's early days. It's early days because we have some early inkling of the product and the workloads. We are looking at -- just trying to learn about the IBM ecosystem, the salesforce. Obviously there's a lot of excitement. And then the fact that they have IBM Global Technology Services, GTS, could be a very new thing for us as well. Power has some real high-end customers and very high-end applications, especially in the space of cognitive and big data applications.

And I think what makes this interesting is, notably for the first time, a single control plane, a single data plane, single hypervisor runtime, can now span Intel x86 and POWER microprocessors' hardware. So I think that's the beauty of what we're building, that we can actually merge these silos. We can be this silo-busting software. And no other hypervisor, no other operating system has actually gone across these. So there's something there which makes POWER no more just a silo.

Okay. And then with regard to the Cisco relationship: I know you mentioned it's early days, but as you garner more momentum, how do you see Cisco reacting, given obviously they have their HyperFlex product? I'm curious, as you see that business grow, is there the propensity for Cisco to become a more formalized partner, in your mind, if HyperFlex doesn't necessarily take off to the degree that they would like it to?

Yes, I think it's a great question. Obviously it's perilous to predict in these situations. But one thing I have learned about the art of negotiation, Aaron, is that what is non-negotiable today -- you have to figure out how to build a process so that it becomes negotiable tomorrow. And that's the process that we've been playing out with for the last nine months.

We are hoping to actually have this process play out where Cisco understands what HyperFlex is, and Cisco also understands the value that we can bring to their rack mount servers. Because as you can see from their predictions and everything else, that the blade business is actually weakening and kind of fading away in this new environment, which is more Web scale and cloudlike.

And rack mounts are a very different growth margin profile. So I think there's something there between us. But we continue to work in this art of negotiation where, what was nonnegotiable yesterday could probably become negotiable tomorrow.

Jayson Noland, Robert Baird.

Dheeraj, I wanted to ask on Lenovo, you mentioned a sharp increase there. And any color you can add, and if you could frame the potential for Lenovo relative to Dell.

And then, Duston, I wanted to ask on OpEx. I think you said an incremental $10 million for the next couple, three quarters. I know you've got the conference in FQ4, but how should we think about that for our models? Thanks.

I think on the first question, we did a really good business with them this quarter. Our joint largest deal to date with them was this quarter. And they are doing well with us in the large enterprise: verticals such as high-end financials, heavy industry, retail, [sled], healthcare and technology and things. These are all very important verticals as we actually start to learn more about the partnership itself.

So I would say that there's a lot of transformation that Lenovo itself is going through as a company. They just split their PCG and DCG division, so there's a lot of go-to-market stuff that we are co-designing and building together. But the early signs look good.

And the fact that the Global 2000 is actually very much friendly towards Lenovo is actually a great sign for us.

And for Duston, there was a question for you as well.

I'm sorry, Jayson, can you -- oh, and the expenses, yes. I think we've got a couple things. Obviously we have our customer .NEXT conference here coming up shortly at the end of June. We look at that obviously ultimately as demand generation.

And then in Q1, we've got a lot of stuff going on. We've got our worldwide sales teams with sales enablement that takes place in Q1 to continue to train the sales teams on new offerings and value selling and things like that. So, along with headcount increases and other things as we go forward there.

I think we're going through the Q1 expenses now, but it has a chance of being, I said at a minimum of 10. It has a chance probably being a little higher than 10, maybe sequentially in Q1. But I think once we get stabilized on the growth, I think the big deals -- as long as big deals continue to be reasonably healthy, that takes care of a lot of issues.

Okay. Thanks, guys.

Andrew Nowinski, Piper Jaffray.

Just a question on AHV. I think you had in one of your slides; it was about 23% of nodes. Maybe just two questions along those lines. Are you charging for that? And if so, what percentage of bookings is AHV right now?

Just like the public cloud, Andrew, we are not really charging for the hypervisor. It's part of the overall value selling. But we go and charge for the operations management software. We have this control plane software which we call Prism Central which comes in the form of Prism Pro. So I think those are the kind of things we charge for, but not the hypervisor runtime itself.

Some of the features of the runtime -- like, for example, network security and things like that, like we announced micro segmentation and so on -- they will be part of the Pro and Ultimate editions. If you recall, we have the good, the better, the best versions of our software.

The good goes with the life of the hardware, but the better and best are terms licenses that we actually sell for. So some of these capabilities which are advanced capabilities go with Pro or Ultimate.

And just, Andrew, on that 23%, too, we calculate that on a percent of total nodes. But as you might expect, Dell is tilted to a lower percentage of AHV. So if you look at it on just NX nodes, it's actually quite a bit stronger than that.

Okay. And then are you seeing customers that deploy AHV, are they actually ripping out their current hypervisor vendor and realizing some of the savings that you had talked about during your IPO roadshow that customers can achieve by using AHV?

Well, by definition, if they are using AHV on our nodes, they are not using any other hypervisor. And if they have other brownfield deployments where they are using the other hypervisor, then we probably don't keep a count on those things. But we know that we can definitely keep track of our own.

And now we are finally getting into some ELA discussions where some of our very large customers of the last two, three years are saying they really want to go all-in on AHV, which is when you'll start to see some zero-sum game.

Got it. Thanks, guys.

I will now turn the call back over to the presenters.

Well thanks, everyone, for joining us today. Appreciate all your questions. As I mentioned before, the most important thing for us this quarter is our annual .NEXT Conference, beginning on June 28 in Washington, DC. And we will look forward to seeing many of you there. Thank you.

This concludes today's conference call. You may now disconnect.