Advanced Micro Devices, Inc. (NASDAQ:AMD) Q4 2018 Results Conference Call January 29, 2018 5:30 PM ET
Laura Graves - Investor Relations
Lisa Su - President and Chief Executive Officer
Devinder Kumar - Senior Vice President, Chief Financial Officer, and Treasurer
Conference Call Participants
Toshiya Hari - Goldman Sachs
Matt Ramsay - Cowen & Company
Vivek Arya - Bank of America Merrill Lynch
Mark Lipacis - Jefferies
Stacey Rasgon - Bernstein Research
Ross Seymore - Deutsche Bank
Aaron Rakers - Wells Fargo
John Pitzer - Credit Suisse
Mitch Steves - RBC Capital Markets
Blayne Curtis - Barclays
Joe Moore - Morgan Stanley
Vijay Rakesh - Mizuho
Greetings, and welcome to the Advanced Micro Device's Fourth Quarter and Full Year 2018 Conference Call. At this time, all participants are in a listen-only mode. A question-and-answer session will follow the formal presentation [Operator Instructions]. As a reminder, this conference is being recorded.
It is now my pleasure to turn the call over to Laura Graves. Please go ahead.
Thank you. And welcome to AMD's fourth quarter and fiscal year 2018 conference call. By now, you should have had the opportunity to review a copy of our earnings release and slides. If you have not reviewed these documents, they can be found on the Investor Relations page of AMD's Web site, www.amd.com.
Participants on today's conference call are Dr. Lisa Su, our President and Chief Executive Officer and Devinder Kumar, our Senior Vice President, Chief Financial Officer, and Treasurer. This is a live call, and will be replayed via webcast on our Web site.
I would like to highlight some important dates for you; Dr. Lisa Su will present at the Goldman Sachs Technology and Internet Conference on Tuesday, February 12; also, Ruth Cotter, Senior Vice President of HR, Worldwide Marketing and Investor Relations will present at the Susquehanna Annual Technology Conference on Tuesday March 12; and our 2019 first quarter quiet time will begin at the close of business on Friday March 15, 2019.
Today's discussion contains forward-looking statements based on the environment as we currently see it. Those statements are based on current beliefs, assumptions and expectations, speak only as of the current date and as such, involve risks and uncertainties that could cause actual results to differ materially from our current expectations. We will refer primarily to non-GAAP financial measures during this call except for revenue, gross margin and segment operational results, which are reported on a GAAP basis. The non-GAAP financial measures referenced herein are reconciled to their most directly comparable GAAP financial measure in today's press release, which is posted on our Web site.
Please refer to the cautionary statements in our press release for more information. You will also find detailed discussions about our risk factors in our filings with the SEC and in particular, AMD's Quarterly Report on Form 10-Q for the quarter ended September 29, 2018.
Now, with that, I will hand the call over to Lisa. Lisa?
Thank you, Laura, and good afternoon to all those listening in today. 2018 marked another year of strong financial performance, driven by our expanded high performance product portfolio despite near-term graphics weakness. We grew annual revenue by 23% with Ryzen, EPYC and datacenter GPUs product revenue growing by more than $1.2 billion for the year. Our new products gained share and significantly expanded gross margin, leading to our most profitable year since 2011.
Looking at the fourth quarter, revenue of $1.42 billion increased 6% from a year ago with approximately 65% of sales coming from our new products. We reached an important milestone in our business in the quarter as our high margin datacenter CPUs and GPUs accounted for a mid-teens percentage of overall revenue. While we expect our data center revenues to be lumpy, the ramp of our data center business is beginning to contribute meaningfully to our financial results.
Looking at our computing and graphics segments, we delivered our eighth straight quarter of year-over-year segment revenue growth. Sales of Ryzen desktop and notebook processors and datacenter GPUs offset lower GPU sales as the channel continued working through elevated levels of graphics inventory. Client processor unit shipments grew by more than 50% from the year ago period. We had our highest client computing revenue in more than four years, and we believe we gain client CPU unit share for the fifth straight quarter.
At CES, Acer, Asus, Dell, HP, Huawei, Lenovo and Samsung, all launched notebooks powered by our new second generation Ryzen mobile processor with Radeon data graphics. The second gen Ryzen mobile processor delivers more performance, enhanced features and longer battery life than any mobile processor we have ever built, and is the fastest processor available for ultrathin notebooks.
The industry's first AMD based Chromebooks launched earlier this quarter from Acer and HP. We expect additional AMD based Chromebooks to launch later this year as we expand our participation in this growing portion of the PC market. Based on the competitive positioning of our Ryzen processors, we expect the number of Ryzen systems that will launched in 2019 to increase by more than 30% in 2018 with the number of Ryzen notebook systems planned to launch increasing by more than 50%.
In graphics, GPU revenue decreased year-over-year, driven largely by lower channel GPU and memory sales, partially offset by a significant increase in data center GPU sales. We saw an improvement in channel GPU sell out throughout the quarter as our partners continued to drain their inventories. There is still more work to do, but we remain confident and we're taking the right actions to further reduce channel inventory. We set a record for professional GPU revenue in the quarter, driven by multiple high volume wins for our Vega based data centers GPUs. We started shipping our new 7-nanometer Radeon Instinct accelerators in the quarter and introduced a major set of enhancements to our data center GPU software that make it easier for customers to deploy Radeon GPUs for AI and machine learning workloads.
At CES, we highlighted the significant gain in momentum we're generating for Radeon across consoles, PCs and the cloud. For gamers and creators, we announced our return to the high end GPU market with the new Radeon VII GPU. Powered by our second generation Vega graphics core and featuring 16 gigabytes of second generation high bandwidth memory, our new 7-nanometer Radeon VII GPU delivers leadership performance in content creation and compute workloads and is very competitively positioned when running the most demanding AAA games at 4k resolution.
In cloud gaming, we announced that Google selected Radeon Pro GPUs to power their game streaming initiative, Project Stream. The performance and differentiated virtualization features of our Radeon Pro GPUs enable Google to deliver an uncompromised High Definition gaming experience on virtually any PC.
Turning to our enterprise embedded in semicustom segment, revenue was flat from a year ago as a double-digit percentage decrease in semicustom revenue was offset by strong growth in EPYC processor sales. As expected, semicustom sales were down from a year ago based on where we are in the current console cycle. This console generation remains one of the most successful ever as Microsoft and Sony combined have now shipped well in excess of 120 million AMD powered consoles.
Fourth quarter server unit shipments more than doubled sequentially based on growing demand for our highest end 32 core EPYC processors with cloud, HPC and virtualized enterprise customers. As a result, we believe we achieved our goal of mid-single-digit server unit share exiting 2018. We had another strong quarter of cloud adoption, highlighted by industry leader Amazon announcing new versions of their most popular EC2 computing instances powered by EPYC processors. Businesses can easily migrate their AWS instances to AMD and save 10% or more based on the technology advantages of our platform.
Microsoft Azure also announced general availability of their AMD based storage instance in the quarter, as well as a new HPC instance powered by EPYC processors that is 33% faster than competitive x86 offerings. We secured multiple HPC wins in the quarter, including Procter and Gamble, the U.S. Department of Energy and one of Europe's largest supercomputers at the University of Stuttgart's High-Performance Computing Center. Lawrence Livermore Labs also announced a new supercomputer, featuring both EPYC processors and Radeon Instinct accelerators that will be used for machine learning and big data analytics workloads.
Customer interest in our next generation Rome server processor remains very high. Rome is expected to deliver four times the floating point performance and double the compute performance for socket compared to our current generation EPYC processors. We probably demonstrated a single 64 core next generation EPYC processor outperforming two of our competitors' highest end server processors in multiple workloads. Rome development is proceeding very well and we are on track to start shipments midyear.
I am also pleased to report that we concluded discussions with GlobalFoundries on the seventh amendment to our Wafer Supply Agreement. The amendment affirms the strategic partnership with GF for products built at 12 nanometers and above, and provide AMD with full sourcing flexibility at the 7-nanometer and below nodes. GF continues to be a critical supplier of AMD's current generation products and will play a key role in our next generation Ryzen and EPYC products with our chiplet strategy.
In summary, 2018 was another strong year for AMD. Increased adoption of our high performance products drove a second straight year of double-digit annual revenue growth, expanded gross margins and improved profitability. I would like to thank the more than 10,000 AMD employees whose dedication with building great products have made these results possible. While headwinds remain in the graphics channel and macro uncertainties are causing some caution in the first half of 2019, we believe we are well positioned to gain share throughout the year and accelerate growth as we ramp our next generation 7 nanometer products. As we entered 2019, we are preparing to launch our strongest product portfolio ever. In gaming, we will launch our high end Radeon 7 GPU in February, followed by our next generation Navi GPUs later in the year.
In client computing, we started the year with our second generation Ryzen mobile processors and we’re on track to launch our third generation Ryzen desktop processors in the middle of the year. And in the server market, we expect to deliver a significant step function performance increase with the launch of our next generation Rome processors in the middle of the year. I am very proud of what we accomplished in 2018 and even more excited about how our long-term investments are set to pay off in 2019.
Now, I would like to turn the call over Devinder to provide some additional color on our fourth quarter and full year financial performance. Devinder?
Thank you, Lisa and good afternoon everyone. 2018 was a strong year for AMD. New product introductions drove the highest annual revenue since 2011, and a significant improvement in gross margin year-over-year.
Earnings per share increased from $0.10 per share in 2017 to $0.46 per share in 2018. Full year 2018 revenue was $6.48 billion, up 23% year-on-year, driven by strong performance in the computing and graphics segment with significant growth in Ryzen processor sales. Although, there was weakness in the graphics channel in the second half of the year, we saw strength in data center CPUs and GPUs.
Gross margin was 39%, up 440 basis points from the prior year. Gross margin improvements were primarily driven by our new Ryzen, EPYC and Radeon products. Operating expenses were 29% of revenue, an improvement of 2 percentage points from the prior year. For the full year, operating income was $633 million, up $409 million from $224 million in the prior year. Net income was $514 million, up $411 million compared to net income of $103 million in the prior year. On the balance sheet, we reduced principal debt by $171 million and improved gross leverage significantly from 4.6 times a year ago to 1.9 times at the end of 2018.
Let me turn to the details of the fourth quarter results, fourth quarter revenue, gross margin, operating margin and earnings per share all improved year-over-year. Quarterly revenue of $1.42 billion was up 6% from a year ago. Strong sales of Ryzen and EPYC processors and data center GPUs more than offset lower channel GPU and semicustom sales during the quarter. Fourth quarter 2018 revenue did not include any IP-related revenue and Blockchain-related GPU revenue was negligible.
Gross margin was 41%, up 740 basis points from 34% a year ago. Year-on-year gross margin improvement was driven primarily by the ramp of Ryzen and EPYC processor sales. Gross margin has increased year-over-year for seven consecutive quarters, driven by a higher percentage of revenue from new products. Fourth quarter 2018 gross margin excludes a $45 million write-down of older technology licenses that are no longer being used. Operating expenses grew 9% year-over-year to $474 million and remained approximately flat as a percentage of revenue from the year ago period. We continued to invest in our product roadmap and go to market activities as we gain market share in important markets.
Operating income was $109 million, up $90 million from $19 million a year ago. Operating margin was 8%, up from less than 2% last year. Net income was $87 million compared to $8 million a year ago. Q4 2018 net income excludes of withholding tax refund plus interest of $43 million related to an IP litigation settlement from 2010. Diluted earnings per share using a diluted share count of $1.180 billion was $0.08 per share compared to $0.01 per share a year ago.
Now turning to the business segment results. Computing and Graphics segment revenue was $986 million, up 9% year-over-year. Revenue growth was driven primarily by continued strong Ryzen desktop product sales and the adoption of second generation Ryzen mobile processors, largely offset by lower channel GPU and memory sales compared to the prior year. Ryzen products continued to ramp and was greater than 80% of total client revenue, driven by OEM and channel momentum. In Graphics, sales were down year-over-year due to negligible Blockchain related revenue in the fourth quarter of 2018, as well as elevated levels of Graphics inventory in the channel. Total Graphics revenue grew sequentially, driven by strong Radeon datacenter GPU sales.
Computing and Graphics' operating segment income was $115 million compared to $33 million a year ago. The increase was driven primarily by strength in Ryzen product sales and significantly higher ASPs in both desktop and mobile compared to a year ago. Enterprise embedded and semi-custom revenue was $433 million, flat from the prior year. Server revenue growth was offset by lower semi-custom revenue. EPYC cost of units more than doubled sequentially, driving significant growth in datacenter revenue in the quarter. EESC segment operating loss was $6 million compared to a loss of $13 million a year ago. The improvement due to higher EPYC cross-sell revenue was partially offset by lower semi-custom revenue and continued engineering and go-to-market investments in the server business.
Turning to the balance sheet, our cash, cash equivalents and marketable securities totaled $1.16 billion at the end of the quarter. We generated free cash flow of $79 million in the quarter. Free cash flow was a negative $129 million for the full year, primarily due to growth in inventory related to new products and the timing of collections. Inventory was $845 million, up $107 million sequentially, primarily due to the ramp of new products. Total principle debt was $1.5 billion as we further reduce term debt by $60 million during the quarter, resulting in total debt reduction of $171 million during 2018.
We expect to pay off the remaining $66 million of 2019 term debt in March 2019, and beyond that there are no long-term debt maturities until 2022. Adjusted EBITDA was $152 million compared to $58 million years ago. And on a trailing 12 month basis, adjusted EBITDA was $803 million, more than doubling year-over-year. Gross leverage was 1.9 times as we ended 2018 and we are pleased to have achieved our long-term gross leverage target of less than 2 times.
Before turning to our financial outlook, let me discuss our Wafer Supply Agreement with GlobalFoundaries. Today, seven amendment of the WSA spans on January 2019 through March 2024. It establishes purchase commitments and pricing at 12-nanometer and above for the years 2019 through 2021. The amendment also provides AMD full sourcing flexibility at 7-nanometer and beyond without any one-time payments or royalties for products, purchase from other foundries.
Turning to the outlook for the first quarter of 2019, we expect revenue to be approximately $1.25 billion plus or minus $50 million, a decrease of approximately 12% sequentially, and 24% year-over-year. Sequentially, the decrease is expected to be primarily driven by continued softness in the graphics channel and seasonality across the business. The year-over-year decrease is expected to be primarily driven by lower graphic sales due to excess channel inventory, the absence of Blockchain later GPU revenue and lower memory sales.
In addition, semi custom revenue is expected to be lower year-over-year, while Ryzen, EPYC and Radeon datacenter GPU product sales are expected to increase. In addition for Q1, 2019, we expect non-GAAP gross margin to be approximately 41%, non-GAAP operating expenses to be approximately $480 million, non-GAAP interest expense taxes and other to be approximately $25 million and we also expect to record $60 million IP licensing gain associated with the static JV in the first quarter of 2019, which will be a benefit to operating income and recorded on the licensing gain line of the P&L.
For the full year 2019 despite near-term business in the graphics channel and a cautious macro environment, we expect high-single-digit percentage revenue growth, driven by Ryzen, EPYC and Radeon and datacenter GPU product sales and as we ramp our 7-nanometer products throughout the year. We expect non-GAAP gross margin to be greater than 41%, non-GAAP operating expenses to be approximately 29% of revenue and a non-GAAP tax rate of approximately 4% of pre-tax income.
In closing, we made excellent progress in 2018. We grew the top-line by more than $1.2 billion, expanded gross margin and significantly improved profitability. We continue to execute our long-term financial model, driven by our new high performance computing products that gained solid momentum last year. We are pleased to enter 2019 with a strengthened balance sheet and a strong portfolio of next generation products capable of driving continued financial growth.
With that, I'll turn it back to Laura for the question and answer session. Laura.
Thank you, Devinder. Operator, we're ready for our first question, please.
We'll now be conducting a question-and-answer session [Operator Instructions]. Our first question today is coming from Toshiya Hari from Goldman Sachs. You're line is now live.
Lisa, I had a question on your full year 2019 outlook. You guys are guiding revenue growth in the high-single-digit range. Embedded in that outlook, what growth rates are you assuming for your core businesses computing, graphics, semicustom and server CPUs? And then on the gross margin side, 41% or greater. Is the improvement year-over-year primarily a function of revenue growth in mix or is the amendment in the WSA playing a role there as well? Thank you.
So relative to the full year guidance of up high-single-digits, the primary growth drivers are Ryzen and EPYC or the two largest growth drivers with datacenter GPU. Also, we're expecting that to be up. And then we would expect that semicustom will be down. If you look at life cycle of semicustom will be in the seventh year of the console cycle. And so we expect semicustom would be down approximately 20% and then we would expect consumer graphics to also be down let's call it double-digits as we really burn off some of the channel inventory that we see in the first half of the year. So Ryzen and EPYC are the largest drivers of the growth, and that’s as we launch our 7-nanometer products throughout the year. And then your second question, Tosh, was…
Yes, gross margins 41% or greater that's a 2 percentage point or greater improvement year-over-year. Is that primarily a function of revenue growth and improvement in mix given Ryzen and EPYC, or is the WSA amendment providing a tailwind as well?
Yes, it's primarily due to the mix. So the mix of the higher margin new products Ryzen, EPYC and the datacenter GPUs.
Our next question is coming from Matt Ramsay from Cowen & Company. Your line is now live.
Lisa, I guess I wanted to attack Tosh's question on the full year a little bit differently. And maybe just talk about a little bit what share gain assumptions you guys are embedding for your 7 nanometer roadmap versus maybe some pragmatism or conservatism on the market. Just given the macro? I mean Intel talked about a flat PC TAM in terms of units and obviously is only I guess guided us to mid single-digit growth for their expansive server business for the year. And you guys are starting off down maybe 25% year-over-year in revenue for the first quarter given what's going on in the graphics business. So maybe you could just walk us through a little bit what you're thinking about the TAM growth of your two main growth markets for the year versus share gains you're embedding? Thank you.
I think as it relates to the market, I don’t think we would have a very different view of the market as others may have stated. Our story really is a share gain story. And when you take a look at the progress that we've made in 2018 in the design wins that we have in 2018. As we go over the 2019 when we look at both Rome as well as Ryzen, we look at the breath of OEM platforms that we have, the customer engagements by workloads and how we see that progressing. We feel very good about the opportunity to gain share as we go through the year, particularly given how competitive the product set is. Similarly on the PC side, we've made nice progress over the last four or five quarters on Ryzen. We started strong on desktop. And then the last couple of quarters, we've made good progress on notebook. We see a broader portfolio with our OEMs as we go into 2019 and we also see a more competitive desktop product line as well as notebook product line. So that’s how we’re thinking about the year.
To your point about starting a little bit with lower guidance in Q1, I think that’s true. When we look at Q1, particularly on a year-over-year basis, there are lot of moving pieces but it's primarily due to the graphics channel. And that’s both the reduced demand, as well as the absence in Blockchain revenue on a year-over-year basis, as well as the semi custom business. But as we move forward, as we go into Q2 and beyond, we see a significant opportunity with the ramp of our new products and that’s how we see the year.
Devinder, maybe just a couple of little things in terms of licensing gains and IP revenue, I got asked a few times and my associates want to clarify here that the gross margin outlook for the first quarter does not include any IP or licensing gains. And then for the full year, are you including any IP revenue in the revenue outlook? And how should we think about the rest of the Thatic licensing gains apply to timing for when those might get recognized? I know it’s a lot there, but lots of question on just the moving pieces there. Thank you.
License gain essentially does not impact gross margin, that's on a separate line all together on the license and gain line of the P&L. So the gross margin is not impacted by that. As far as IP revenues is concerned in Q1 there is no contemplated IP revenue, so the gross margin at 41% guide is based on the products that we have. And for the year at this point outside of the Thatic JV IP licensing gain that we are expecting to record in Q1, while there might be opportunities from an IP standpoint, but nothing substantial contemplated at this point.
Thank you. Our next question is coming from Vivek Arya from Bank of America Merrill Lynch. Your line is live.
Lisa, I'm curious as you look back at 2019 and the success you had with EPYC, the initial success with EPYC, what was the mix of cloud versus enterprise? And then how do you think it trends in 2019, because of all the concerns around slowdown in cloud CapEx and so forth. And as part of that, if you could share with us what your market share assumptions are as we exit the year on EPYC?
So as we look through 2018, we were pretty pleased with our progress on EPYC. And coming off of the fourth quarter, actually it was a fairly strong fourth quarter for us and in fact that we doubled the number of units for our server business. And when you look at that mix, it is more cloud weighted. So we had some large deployments that went online here in the fourth quarter and that was positive for us. That being said that we’re making nice progress in the enterprise and HPC side of the business too. We've had a number of wins in the quarter, as well as going into 2019.
So as we look into 2019, I would expect that the early Rome deployments will also be cloud based. It will be the first ones, but we have a strong set of enterprise platforms. And as I mentioned earlier it’s the breadth of the OEM platforms that gives us good confidence that we were going to a broader set of workloads and having broader coverage in the market. In terms of share assumptions we will have to see how the markets and year play out. But I think what we've said before is that after reaching the mid-single digit market share in the fourth quarter of 2018, we would expect it would take another four to six quarters to reach 10% market share. And I think we're still in that range.
And for my follow up, there were two things that came out on your competitors last few public discussions, which is one this concept of CPU shortages. And I’m wondering if that has had any positive or negative effect on your strategy or positioning? And then Intel has also mentioned being a little more tactical when it comes to pricing. And I’m wondering how that figures into your full year outlook. So both the impact of CPU shortages and pricing, any color would be very helpful. Thank you.
On the CPU shortages, my comment is that there are some pockets of shortage, particularly at low end. However, our focus has really been on ramping Ryzen. And if you look at the Ryzen percentage of our overall business, Ryzen Devinder mentioned was 80% of our client business. So we are really actually improving our mix in the client business. I think that, again from my view, the shortages are temporary but we look at it as really getting consistent share gain. And as we've gone through each quarter in 2018, we've seen consistent share gain and we believe we gain share in the fourth quarter as well. So I think we are well positioned with the portfolio and we will certainly drive that into 2019.
And then relative to pricing, when we think about pricing and when we put together our long-term model, we have factored in that pricing and competition will be aggressive. I think we know that to be the case. What we've seen in the market is consistent with that. But we also believe that particularly as you go into the datacenter, the single largest factor from a buying decision standpoint is the performance and the total cost of ownership of the product. And we believe that Rome will be very, very well positioned in 2019. And so we are cognizant of the competitive environment but feel good that we have the right set of assumptions.
Thank you. Our next question is coming from Mark Lipacis from Jefferies. Your line is now live.
Thanks for taking my question and great to hit the mid-single digit milestone exiting the year. Lisa, as you start -- I just wanted to confirm, you expect to start shipping Rome mid-year. And as you do, what's the biggest risk for Rome? Is it execution on delivering the product, or is it the sales process convincing customers to take EPYC to in volume?
So I think with Rome, our biggest opportunity/risk is adoption rate. I think from a competitive standpoint, the product is very solid, everything going through our development labs looks very good, as well as our customer engagements. And so this is just about getting customers to production as fast as possible. We do expect though that the adoption rate for Rome will be faster than the adoption rate for Naples. And the reason for that is we are in a socket compatible infrastructure. And so customers who don't necessarily need the newest features of the platform can actually use the same motherboard and system that they currently have with Naples and drop in Rome. And so I think that will help us accelerate some of the adoption. But right now, it's about helping customers in their environment. We are widely sampling Rome and there's a lot of work to be done but we feel good about the trajectory.
And a follow up if I may for Devinder. So inventories, looks like they were up about $100 million going into a pretty healthy seasonal quarter that's declining. How should we read that Devinder? And was there just some opportunity to get some wafers more cheaply? Or are you trying to stockpile product in front of launches? And when can we expect that inventory to be a source of cash on the cash flow statement? Thank you.
So the inventory growth is primarily driven by our new products and in preparation for the 7 nanometer next generation product launches later this year as you just mentioned. We overall expect to manage our environment revenue, but also in support of the new products that will be launching ramping throughout the year. Lisa mentioned a few of the launches at CES and also in the script. And we prepared to support that from an inventory standpoint and we'll see how the year progresses and then manage it from there.
Our next question is coming from Stacey Rasgon from Bernstein Research. Your line is now live.
Around the Q1 guide, obviously, you mentioned the GPU weakness but you also mentioned seasonality across the businesses. Is my assumption correct that that means everything is basically down sequentially CPUs, GPUs, Servers? And can tell us is that true? And if so can you give us an idea of the magnitude like what is down most versus least sequentially, Ryzen versus GPU versus EPYC?
So the largest driver as I said is the GPU channel situation on a year-over-year basis. On a sequential basis, typical seasonality for let's call it PCs and GPUs and datacenter is probably around 10% or so. We are forecasting or we believe that that will be a little bit -- down a little bit more than that. And in terms of the ranking, I would say GPUs are down a bit more than the other two segments that's the way we currently see it.
For my follow-up, I had a question just in general EESC. So its revenues are flat year-over-year. The server revenues are up a bunch and still losing money. And I get it there's some incremental investments and everything. But I guess how do we think about the profitability of that segment? How big does EPYC or Rome in the server business maybe to get before it sustainably profitable, especially semi custom has going to continue to decline 20% in 2019?
I think the key, Stacy, is going to be ramping the server business in 2019. You are right that year-over-year we look at it, it's lost little bit of money. But fundamentally, it's investing in the go-to-market programs and launching the new products, investing in the ongoing engineering work that's needed to make sure that customers bring up the product on time. And semi custom, given the fact that is in the seventh year of the cycle, we expect it to be down from 2018 to 2019. But we do expect to go ahead and ramp of our server business and that will definitely benefit the EESC profitability in 2019.
Do you think it's profitable for the full year?
Too early to tell, I don't predict profitability at the segment level. But we'll see how it evolves over the year.
Our next question today is coming from Ross Seymore from Deutsche Bank. Your line is now live.
Lisa, a potentially different way to ask the same question on the full year 2019 guide. Late last year at a few different events, you talked about the server business being second half weighted. If I go through the remainder of this year, it seems like you have to grow better part of 20% plus sequentially in each quarter to get there. I know it's not going to be perfectly linear like that. But I guess my overall question is if the server side of the business is back half weighted. What sorts of businesses are helping the middle portion of the year? Is there some snapback on the GPU side? Is there something going on on the Ryzen side? Just trying to get somewhat of the shape of the year off of a base that's admittedly pretty low as your first quarter starting point?
So I think it is fair to say that our quarterly progression will have a lot to do with our product launches. So you would expect that server will be more second half weighted than first half. In general, our business is more second half weighted though if you think about the consumer portions of our business. But think about it as we would expect sequential growth in PCs, we would certainly expect sequential growth in datacenter, although stronger in the second half. We're also as we see the GPU business right now, we see the first quarter as the low point in the business with the channel getting improving as we go into the second quarter. And we have additional product launches there as well. So that's the way we would see the portfolio. And semi custom, although it's lower on a year-over-year basis, we would expect it to also increase as we go from second quarter into third quarter as well.
And then a question on the GPU side, your big competitor in that market obviously is having some current issues and mentioned that the new product at the high-end wasn’t selling through. Are you at all concerned that the competitive landscape in that market, whether it's because China demand being weaker as they cited or them having to be a little bit more aggressive on pricing would do anything to your ability to penetrate the market with either existing or new GPU products as the year progresses?
So when we look at the GPU market and let's separate gaming and data center. I think on the gaming side from what we are seeing, we did see sell out increase in Q4 versus Q3. So gamers are still buying GPUs. They may be more discerning about price points. And so I can imagine that there might be a bit more softness at the high end versus in the mid range. But we believe that we have a good understanding of what's happening in the gaming side of the business, and it will be driven. Our gaming growth will be driven by new products. We would see that as we go through this year and with our Radeon 7 launch, as well as our Navi launches on the gaming side.
On the data center side, we're making good progress in GPU data center, obviously, from a low base. But the GPUs and the data center are very workload dependent. And there are some workloads that actually we do very well in things like cloud gaming and virtualization and we have some early HPC wins. And so we see this as customer acquisition need deployments. And so those are the sources of growth on the GPU side for us.
Thank you. Our next question today is coming from Aaron Rakers from Wells Fargo. Your line is now live.
Just two questions if I can, building on that last one. In your prepared remarks you noted that you've reached a milestone of which GPUs as well as data center CPUs reach a mid-teens percentage of your total revenue. I'm curious as you contemplated your full year guidance. Where do you think that mix of business can go? And is there a point in time where we actually, you foresee us giving granularity on how big the data center GPU business is trending?
So, I think the mid-teens percentage revenue for us in Q4 was a good milestone. I mean I think that's a meaningful percentage of our revenue. And it was a contributor to our gross margin as we exited 2018. I think as we go into 2019, again, we will continue to give you visibility into where the data center growth is. To be fair of my expectation is that the CPU side of the data center business will grow faster than the GPU side just given what we see in terms of overall deployments. But I think both businesses will be meaningful for us and are a key part of the growth story for us in 2019. And we will continue to give color on how they develop as the year develops.
And then just real quickly on the semicustom side, as you contemplated the guidance as well through the courses here in that 20% decline for the full year. Is your assumption that Q1 marks the bottom and we can grow seasonally off of that level? Or should we think about something differently from a progression through the quarters in '19?
Aaron, that's the correct way to think about it. So, Q1 will be a low, Q2 will be higher, Q3 will be higher, and then Q4 will come down again. That should be the seasonality of semicustom.
Thank you. Our next question is coming from John Pitzer from Credit Suisse. Your line is now live.
So notwithstanding that the macro is extremely uncertain making any predictions difficult. I’m just curious going back to Ross' question. How do you see the cadence of growth sequentially throughout the year? Because as he pointed out if you divide it up equally by quarter, you’re growing north of 20% in Q2, Q3, Q4 half on half growth would be somewhere around 50%. If you assume more seasonal and seasonal is a little bit difficult to get to, because of change in accounting, I think you can get half on half growth to 75%. So as you think about the full year growth, is this something that’s more like 50% half on half or 75% half on half?
It might be a little bit early for us to go there, but what I would say is the following. When we look at the year-on-year compares from 2019 to 2018, Q2 2018 was still a very strong quarter for us, because we still had a significant amount of Blockchain sales in 2018. So I expect that we will grow sequentially into Q2, but I think that will still be, let's call it, a tougher year-on-year compare. But then when we get into the second half of the year, we expect to be fully ramped on the entire 7 nanometer portfolio. And so we would see a heavier weighting in the second half. So we see sequential growth into second quarter, but more heavily weighted into the second half.
And then, Devinder, on the profitability front if you run the numbers from Q1 guidance to what you need to be exiting Q4 at to hit the full year number, you could see revenue up well over $1 billion Q1 to Q4. I’m a little bit surprised that despite that, the gross margin guidance isn’t a little bit better. I know you guided the full year to greater than 41 with Q1 at 41. But I’m curious if you look at Q4 run rate that’s north of a billion higher than the Q1 run rate. How should we think about gross margin and operating margin existing the year?
If you look back the last many quarters I think seven quarters year-over-year, we have improved gross margin on the strength of the newer products. We ended the year quite well in 2018 and achieved gross margin in the fourth quarter that was nice, and now we start the year with 41%. And as we start the year, it's only we're guiding greater than 41%. I think you should expect that as the products ramp in particular with the 7 nanometer product that Lisa referenced, our margins should continue to improve and we will see where we end up by the end of the year. But we do expect that year-on-year margin goes up from where we ended in 2018 at 39% and greater than 41% for 2019.
Our next question is coming from Mitch Steves from RBC Capital Markets. Your line is now live.
First on the C&G side, I realized that there's going to be a lot of sequential difficulties there. But when I think about the long-term growth rate, I know you guys have talked to high singles to double-digit growth. Is that still essential where you think that business can grow at long-term? And then secondly in terms of the next EPYC 2. Is there any reason why it would underperform Intel 10 nanometer in a testing environment? And if so, why would that be?
Let me take the second question and then maybe Devinder you can take the first question. So as it relates to our ROM product, when ROM was originally planned, we had planned that it would be competing against the 10 nanometer product. So that was our expectation when we started. I think we've shown some of the generational performance advantages in terms of double digit performance on a socket basis, 4 times performance on a floating point basis. And the other thing I would say is that with our second generation Rome, the customer set is now more used to our architecture and so some of the architectural and software improvements that we've also spent quite a bit of time on over the last four to six quarters are coming into play. So I think we feel very good about the competitive positioning of Rome. And the other thing to keep in mind is we are deep in development of our next generation Beyond Rome. So the Zen 3 product portfolio is deep in development as well. And so that our goal is to ensure that we have a consistent cadence of very competitive products.
And then as far as your question on the segment growth is concerned relative to high single digits growth for the company level. In both of the segments, there are some factors that come into play. In Computing and Graphics, obviously you feel good about the datacenter side of GPU. But on a compare basis when you look at 2018 to 2019, we do have the impact of the Blockchain and some memory sales from 2018 that don't happen in 2019 from a standpoint of where the market is. And then on the EESC semi-custom in down obviously is a factor when you compare year-on-year for segment. But good growth on the datacenter piece of it in particular with the 7 nanometer Rome product launching in the back half, I mean ramping in the back half of 2019 and that's the way you look at it. It is still early in the year from an overall standpoint. And where we ended up in 2018, we're projecting high single digit growth at the company level for 2019 over 2018.
Thank you. Our next question is coming from Blayne Curtis from Barclays. Your line is now live.
Lisa, I just want to ask, you laid it out a little bit. But in terms of EPYC to thin socket but you have sampled it early in terms of getting over the finish line and getting the shift. I guess the customer still has the wafer production of silicon. And then in particular people talk about six to 12 months for those to test and make sure it does everything with the production of silicon. Can you walk us through that timeline, because obviously you're baking in a pretty steep ramp in the second half? Thanks.
So we started sampling EPYC actually last year so the second half of last year to some of the top cloud vendors, as well as our OEM partners. We've had a significant amount of development with our ODM partners as well. So I think there is a good amount of overlap between the customer development cycle and our development cycle. Now as you said, production of silicon is very key and ensuring that the final certifications are locked in is work that we still have to do here in the first half of this year. But I think we feel good about it. We feel good about the platforms, the customer engagements, the current progress of developments. And so that is very much our plan from an execution standpoint.
And then I just want to ask on the GPU side. It looks like your ASP was up substantially with the data center just kind of curious as we think about the mix of data center. How do you think about the trajectory of that business? I mean is that forced in lumpy as well or is this the go forward level after that pretty sharp initial shipments in December?
So on the datacenter side we do expect that datacenter GPU business will be a growth driver for us in 2019. It will be a little bit lumpy, so I wouldn't say it's a straight line but I would say on a year-over-year basis, it's a important growth driver for us. And as it relates to the overall datacenter CPU ASPs affecting overall ASPs, I think that's true because the gaming ASPs are or the gaming business was lower than expected given the channel inventory situation. So I think it's also just weighted in terms of the units.
Thank you. Our next question is coming from Joe Moore from Morgan Stanley. Your line is now live.
It seems like the GPU datacenter business is doing pretty well. Can you give us some color if you look at mid-teens exposure overall between the two datacenter businesses, CPU and GPU? Can you give us some qualitative sense for how much of that is GPU at this point? And what are the main applications, is it mostly virtualization? What other applications are you seeing there?
So for the mid-teens percentage of revenue in Q4 between CPU and GPU, it's actually roughly similar. And then in terms of the workloads on the GPU data enter side. We do very well in cloud gaming. We do very well with our virtualization solutions and we have some early traction in HPC.
And then I guess just in terms of the business segmentation, I think you alluded to this earlier. But it seems like the conversation typically is around microprocessor graphics and semi custom as three different opportunities. Have you thought about moving the segmentation to more in line with that? And I realize it's easier to talk about and hard to implement. But what's your thought in terms of overtime migrating that more in line with the way you guys talk about the business?
So we look at the business in all different cuts. As you can imagine, we look at it by segment our CG and EESG. CG is very much a PC driven segment and EESG with enterprise and semi custom. And we are, also based on some of the feedback, trying to be a little bit more granular in terms of how the various businesses fall underneath that. So I think as the datacenter business becomes a larger piece of our business, we'll continue to look for how do we give you guys more color and more transparency in how that is going.
Our next question is coming from Vijay Rakesh from Mizuho. Your line is now live.
Just in terms of the GPU gaming side, you mentioned higher inventory levels. I was wondering if you can give us some number on where normal inventory GPU levels are and where are they running currently? Thanks.
So when we look at the GPU channel inventory, we do believe our channel partners reduce inventory in Q4 and they will reduce inventories in Q1. We still think the Q1 levels are elevated and there will be some spill over into Q2, and we will see some elevated inventory levels. We'll have to see how the sell-through really, really plays out. But my expectation is that in Q2 will have improved channel inventory levels and we will return to sequential growth in the gaming side of our business.
And on the 7-nanometer side, is the 7-nanometer GPU and CPU as those ramp in the back half. Are they accretive to this 41% gross margin levels? Thanks.
So the 7-nanometer GPUs on the data center side started ramping in Q4. So that was part of the data center GPU revenue that we've talked about. The 7-nanometer CPU will ramp middle of the year. And the 7-nanometer CPU and GPU are both above corporate average margins. So you would expect that they're above the 41 points and would be accretive to margin as that mix improves.
Thank you. We reached end of our question-and-answer session. I'd like to turn the floor back over to management for any further or closing comments.
Thank you very much operator. We appreciate everyone joining us here today. We're quite proud of the accomplishments in 2018, and look forward to seeing you all on the growth in 2019. Have a great afternoon.
Thank you. That does conclude today's teleconference. You may disconnect your line at this time. And have a wonderful day. We thank you for your participation today.