Oct 22, 2020
Operator
Good afternoon. My name is Rob, and I will be your conference operator.
I’d like to welcome everyone to the Xilinx Fiscal Second Quarter 2021 Earnings Release Conference Call. All lines have been placed on mute to prevent any background noise.
After the speakers’ remarks, there will be a question-and-answer session. [Operator Instructions] Please limit your questions to one, to ensure that management has adequate time to speak to everyone.
I would now like to turn the call over to Matt Poirier. Thank you.
Mr. Poirier, you may begin your conference.
Matt Poirier
Thank you, and good afternoon. With me are Victor Peng, CEO; and Brice Hill, CFO.
We recognize there have been a number of recent reports regarding a potential M&A transaction with Xilinx. Our policy on M&A rumors is to neither comment nor answer questions about them, and we will abide by that policy with respect to these reports.
The purpose of today's call is to discuss our most recent quarterly results and outlook, and we would ask that questions be limited to these topics. Let me remind everyone that during our conference call today, we may make projections or other forward-looking statements regarding future events or the future financial performance of the company.
We wish to caution you that such statements are predictions based on information that is currently available and that actual results may differ materially. We refer you to the documents the company files with the SEC, including our 10-Ks, 10-Qs and 8-Ks.
These documents contain and identify important risk factors that could cause the actual results to differ materially from those contained in our projections or forward-looking statements. In addition to GAAP financial measures, we will be disclosing certain supplemental non-GAAP financial measures used by management to evaluate the company's financial results.
We provide these measures to facilitate period-to-period comparability for purposes of evaluating continuing business operations, by excluding the effects of non-recurring and unusual items, such as amortization of intangibles and certain one-time items related to acquisitions. We believe that sharing these non-GAAP measures will be helpful for analysts and investors in analyzing the company’s ongoing core business.
A reconciliation of non-GAAP financial information to the closest GAAP measure is included in our earnings release and has been posted on our Investor Relations website. This conference call is open to all and is being webcast live.
It can be accessed from our Xilinx Investor Relations website. Let me now turn the call over to Victor.
Victor Peng
Thanks, Matt, and thanks to everyone for joining today's call. I hope you and your families are healthy and well.
Let me now briefly touch on how we're operating with the ongoing COVID-19 challenges before I discuss the business. Our supply chain and development activities have continued without much disruption.
Most of our employees continue to work from home, except in China where employees have been allowed to return to work. And we've continued to progress on our strategy, despite approaching eight months of largely working from home.
Overall, our teams are doing an outstanding job executing and delivering for our customers. Now on to the business highlights.
Fiscal Q2 revenues grew 5% sequentially and were $767 million, higher than the midpoint of our guidance. DCG performed better-than-expected and had a record quarter.
Wired and wireless revenues were slightly better than expectations. In the core businesses, AIT revenue grew sequentially.
Our A&D business was strong as expected, but offset somewhat from lower-than-expected TME revenue, due to some emulation revenue shifting into Q3. ADC performed significantly better than expected, with the auto business showing signs of recovery.
The advanced products category constituted 70% of total revenues. Zynq-based revenues increased 28% compared to the prior quarter and were 22% of the company revenues.
Our Zynq SoC design momentum continues to strengthen across our target market. And we're confident that SoC revenue will be a much larger portion of our overall business in the future.
Now I'll move on to business specific highlights starting with DCG. We saw strong growth in our DCG business during the quarter with revenues crossing the $100 million mark for the first time.
Cloud service provided deployment of Alveo-based compute AI clusters, and Solarflare NIC adapters contributed significantly to this growth. DCG customer traction continued to grow in Q2, including a marquee SmartNIC design win with a Tier 1 U.S.
hyperscaler. This win is expected to realize well over $100 million in annual revenues by FY 2024.
Notably, we won against several of the other top market players, mainly due to the unique value we deliver with our adaptive SmartNIC hardware and our stronger software capabilities from our Solarflare acquisition. Turning to other customer engagements and design win activity.
We see growing 100-gig SmartNIC engagements with hyperscalers and proof-of-concept activities for 200-gig solutions. In compute acceleration, it's great interest in an RT video server, and we have multiple fintech customers engagements in banking and electronic trading exchanges.
Finally, in storage, we saw the next-generation and storage platform commitments from new customers. With respect to our progress in our platform software and ecosystem development, we've had close to 43,000 downloads of Vitis since announcing it late last year.
To date, we've trained over 15,000 developers and have over 1,000 software partners, releasing a growing list of applications. We've also made progress with the AI developer community and have 50 AI application examples, open source on the Xilinx Github account.
Moving to WWG. As expected, we saw a significant revenue contribution from one of our Tier 1 OEM customers that is ramping RFSoC production this quarter.
These RFSoC deployments offered sub-six gigahertz of massive MIMO rated deployments in North America. Our design win pipeline for RFSoC continues to expand globally.
We're also beginning to see revenues from a Tier 1 customer who is using our 7-nanometer Versal ACAP, and we expect further revenue growth in deployment start in 2021. As 5G deployment begins to ramp in more geographies, we are very well-positioned to benefit from the significant increase in deployed radio units, especially in massive MIMO configurations.
Our product leadership with RFSoC and Versal ACAP and the value that these adaptive SoCs provide to our customers are unique in the industry. We're also offering customization of our products to aptly meet our customers' cost, power and for factor requirements by hardening selected IP while maintaining our unique adaptive capability.
We'll share more details about this in upcoming announcements. We also made great progress in the Open-RAN space, where we see a big opportunity over the next several years.
We're working with key stakeholders to drive O-RAN initiatives to ensure 5G and future networks be openly developed since our operating and adaptable. As I mentioned last quarter, we are a member of both the open-RAN Policy coalition and over analyzed, and it contributed to the 3GPP specifications for 5G mobile networks.
Our products empower our customers to innovate and get to market with a differentiated and custom solution faster than any other option. This unique capability will accelerate realizing the promise and performance of massive MIMO and O-RAN.
Last quarter, we announced the T1 telco accelerated card for O-RAN distributed units and virtual baseband units in 5G networks. Customer interest continues to be solid in this area with positive feedback for market leaders like Nokia and Mavenir.
Vodafone recently identified Zynq as a technology front-runner in the massive MIMO category for open RAN enabled radio unit hardware products. In addition, KMW recently announced that it is selective Xilinx as a strategic partner and silicon supplier for the company's base station radio equipment business.
Moving now to the core markets. I'm delighted to say that we're very close to being at to our pre COVID-19 business level, employed to resume growth.
As you know, our core markets are diversified and provide a highly resilient foundation for our overall business. Our core markets provide significant and consistent cash generation to support reinvestment in these markets as well as in our strategic initiatives.
Aerospace & Defense business grew in the quarter and set a new record. We expect this business to be a secular growth driver in the long term with some lumpiness from time to time.
Global defense budgets are generally trending up, and we see emerging technologies like hypersonics and increasing adoption of AI. During the quarter, our 20-nanometer radiation collar Kintex UltraScale FPGA received a Platinum Honor from the 2020 Military and Aerospace Electronics Innovators award in the Interconnect Technology category.
The industrial business is recovering, which is consistent with the recent manufacturing PMI data, and our leadership in E&P remains very strong. We recently announced the world's largest FPGA the VU19P, which is in production.
With 35 billion transistors, the VU19P provides the highest logic density and IO count on a single device ever built. The VU19P sustains our product leadership in the emulation and prototyping market that was established with our 28-nanometer generation of products.
We saw a recovery in the auto market over the last quarter, though it's not quite back to our pre-COVID run rate. We expect this trend to continue into the second half of the fiscal year.
We had several design-ins in ADAS as well as in DMS, where we are a leader. Subaru Selects Xilinx to Power its New-Generation EyeSight System, debuting with the Subaru Levorg in Japan.
Our Zynq MPSoC will be powering their next-generation ADAS systems to offer features like automatic emergency braking, adaptive cruise control and lane keeping assist. We also announced with Continental, the world leader in automotive radar systems that our Zynq products will be powering their advanced radar systems.
Continental has the AMCs first production-ready 4D image radar that can support Level 2 to Level 5 autonomous driving. We expect the auto market to resume robust long-term growth as the industry continues to recover and adoption of ADAS growth.
Now let me turn it over to Brice.
Brice Hill
Thank you, Victor. As Victor mentioned, we had a solid Q2 with broad strength across many of our end markets.
This strength drove total revenue of $767 million, above the midpoint of guidance we provided on our Q1 earnings call and 5% higher than our previous quarter. By end market, the Data Center group revenue grew 22% quarter-over-quarter and 30% year-over-year, driven by continued build-out of an AI compute cluster at a cloud service provider and ongoing strength with our Solarflare products.
Note, we prioritized our orders to comply with the most recent trade restriction rules, which resulted in a significant outperformance. Without the impact of the additional trade restrictions, DCG revenue would have come in approximately flat to Q1 as originally expected.
Wired and wireless group revenue decreased 13% quarter-over-quarter and 36% year-over-year. The expected sequential decline was due largely to the CIV-related order acceleration seen in Q1.
Wire performed ahead of expectations while wireless business was largely in line with expectations. Wired outperformance was driven by ongoing access network build-outs as well as some benefit from the recent trade restrictions.
In wireless, we saw a healthy ramp of our RFSoC product with a Tier 1 OEM for 5G deployment at a North American operator. Across our core markets, encompassing our AIT and ABC groups, revenue grew 10% quarter-over-quarter and 5% year-over-year, driven by improving business conditions across multiple end markets.
More specifically, ABC or automotive, broadcast, and consumer revenue increased 36% quarter-over-quarter and declined 8% year-over-year, with a strong rebound during Q2 in the automotive end market with meaningful improvement across multiple Tier 1 OEMs. Solid broadcast end market performance during Q2 came in line with expectations.
AIT or aerospace and defense, industrial, and test and measurement revenue increased 3% quarter-over-quarter and 11% year-over-year with strong performance in aerospace and defense, which delivered a record quarter. ISM performed as expected, while TME results were less than expected due to an emulation customer program that started in Q2 and is now expected to extend into Q3.
Now, some other financial highlights and metrics. Company-level gross margin was toward the high end of guidance with GAAP gross margin of 70.7%.
The performance was primarily driven by end market mix and lower costs. GAAP operating expenses of $336 million or 44% of revenue were within our guidance range.
Higher sequential operating expenses were driven by higher bonuses due to our first half profitability and our salary increases in July. GAAP operating income was $205 million or 26.8% operating margin.
Our GAAP tax rate was 0.4%, in line with guidance. Note, our lower fiscal Q2 tax rate is driven by a tax benefit associated with the vesting of appreciating stock awards.
GAAP net income was $194 million and diluted earnings per share was $0.79, a 108% quarter-over-quarter increase and 11% year-over-year decrease. Diluted share count increased quarter-over-quarter to 246.8 million shares.
On a non-GAAP basis, gross margin was 71.5%, operating expenses were $332 million, operating income was $216 million, tax rate was approximately 1%, net income was $203 million, and non-GAAP diluted EPS was $0.82, a 26% increase from Q1 and a 13% decrease year-over-year. Note, the difference between our GAAP and non-GAAP is due to M&A-related expenses and amortization and related income tax effect of non-GAAP adjustments.
On to balance sheet and cash flows, total cash and short-term investments increased $100 million to $3.1 billion in the quarter and our total debt remains $2 billion. Accounts receivable increased to $362 million in 43 days compared to 38 days last quarter.
The days sales out increase was driven primarily by linearity of shipments. Inventory decreased to $282 million and days of inventory stood at 114 days, same as the prior quarter.
We generated $248 million in operating cash flow or 32% of revenue and $232 million in free cash flow or 30% of revenue. During the quarter, we paid dividends of $93 million.
Through the first half of fiscal 2021, we have returned a total of $239 million or 52% of free cash flow through both dividends and share repurchases. Turning now to the outlook for fiscal third quarter 2021.
We expect third quarter revenue to be between $750 million and $800 million, which at the midpoint is approximately up 1% quarter-over-quarter and 7% year-over-year. This reflects continued strength in our core markets, led by TME, auto and broadcast end markets.
DCG is expected to be lower after a record Q2 and WWG is expected to increase as 5G deployments and ramps continue. Some additional color into our outlook by end markets.
Within AIT, PME sales are expected to increase meaningfully due to strong emulation and prototyping program revenues. Aerospace and defense sales are expected to moderate from a record quarter, but should still be in line with historical levels.
Industrial science and medical is expected to decline modestly as fiscal Q3 is generally a seasonally lower quarter. We continue to see general recovery in manufacturing activity in the U.S., Europe and Asia.
ABC markets are expected to continue recovery, driven by strength in auto, where we are seeing increased demand from our ADAS platforms at our Tier 1 customers. Our broadcast end market is also expected to strengthen, as live sports and other live events, like the U.S.
election coverage, increase. DCG sales are expected to decline from a record quarter in Q2.
As mentioned previously, the DCG business saw some order acceleration during Q2 related to trade restrictions. WWG is expected to be up modestly, with a strong increase in wireless, as 5G ramps continue across multiple OEMs in multiple regions, offset by a decline in wire due to trade restrictions, COVID-related slowdown and seasonality.
Please note, Huawei has been removed completely from our outlook across the business. Fiscal Q3 non-GAAP gross margin is expected to be between 68.5% and 71.5%.
Non-GAAP operating expense is expected to be between $333 million and $347 million. Non-GAAP other expense is expected to be between $12 million and $16 million.
Non-GAAP tax rate is expected to be between 6% and 9%. In closing, we are pleased with our performance in the first half of fiscal 2021, reflecting the strength of our business across diverse end markets and continued transformation to a platform company.
Our adaptive SoCs, including Zynq, MPSoC, RFSoC and our upcoming Versal, are both broadening and deepening the market and customer set for Xilinx, allowing us to compete more effectively in areas traditionally served by ASSPs and ASICs. We remain as confident as ever in the opportunities ahead of us and furthering our technical and market leadership.
Thank you. And let me now turn the call to the operator for Q&A.
Operator
The floor is now open for questions. [Operator Instructions] Please limit your questions to one, to ensure that management has adequate time to speak to everyone.
If there is extra time at the end of the Q&A session, you are welcome to ask a follow-up question. And your first question comes from the line of Ross Seymore from Deutsche Bank.
Your line is open.
Ross Seymore
Hi, guys. Thanks for letting me ask a question.
I just wanted to ask about some of the surprises in the quarter and what it means going forward. On one side, it looked like the AIT side didn't grow nearly as strongly, but it looks like it's going to rebound in the out quarter in December because of the emulation side of things.
So any color on why that keeps getting pushed out? And then similarly, on the data center side, it was a great quarter by any measure in September, but your guidance for December sounds like some of that goodness was a pull-in due to the trade restrictions.
So any sort of color about that would be helpful.
Victor Peng
Yes, Ross, let me take that. So look, on the emulation and prototyping, as you said, some revenue just moved over, but there's nothing fundamentally different about what we're doing there.
As I said in the prepared remarks, we really have maintained leadership with the view ITP, and we've established that for several years now. It's just a situation with a particular customer.
I wouldn't read anything into it beyond that. Regarding the data center, yes, we are saying that we saw some pull-in because of the most recent restrictions.
But you can think about that really as an order of one quarter pull-in. This is not a big thing.
And again, I guess what I would say is that I think that it's still an indication that we're getting really good traction in data center. Of course, kind of, we said even without perturbations like trade -- new trade restrictions.
Things are really lumpy right now in the data center as we're still scaling the business. And in general, also, it's a pretty consolidated business.
So it would have some natural lumpiness. And it's just causing some more perturbations on a quarter-to-quarter basis.
But overall, we're still tracking a good growth. And so we're still very confident about what's going on in data center.
Ross Seymore
Thank you.
Operator
Your next question comes from the line of Aaron Rakers from Wells Fargo. Your line is open.
Aaron Rakers
Yes. Thanks for taking the question.
Just on the topic of 5G, we saw Verizon announced this morning that I think they deployed more base stations on 5G in the last two months than they did of all of 2019. So in that context as we see the pace of this start to accelerate, can you just revisit how you guys see the content expansion in 5G?
And just in general, where or how we should think about the progression of that cycle and what it means for Xilinx? Thank you.
Victor Peng
Yes, it's a good question, Aaron. Yes.
As you said, one of the things we're really pleased about is we're starting to see our RFSoC being deployed in sub-6 and in North America. And we also see more deployments around the world.
So we're very, very excited that we are seeing that ramp. I think just like the previous question, with trade and a few other things, in particular, in 5G, that's caused some perturbations.
But if you step back in the big picture, right, we still are absolutely confident this is going to be a very significant opportunity for us. We have the strongest lineup we ever had.
We have Versal win for 5G. We have the RFSoC deployed in areas.
And you'll be hearing more about what we're doing even with the RFSoC family in terms of further the announcements. So, yes, it's still the first wave.
You've heard me talk about three generations of equipment. What's being deployed right now is still the first wave.
And we certainly see that it's good to see North America starting to pick up. China continues, and it's had some choppiness because of the various trade issues.
But we are seeing other geographies starting to deploy, and that is going to drive the big opportunity that we talked about. But still first innings, right?
So we got engagements with the second-generation equipment, but that's yet to deploy. And then there'll be at least another third generation, if not, perhaps, some after that as well.
Operator
Your next question comes from the line of Ambrish Srivastava from BMO. Your line is open.
Ambrish Srivastava
Hi, thank you. Victor, I just had a follow-up on the data center side.
So good to see it hit $100 million run rate. A couple of questions from that.
I just want to make sure I understood the upside that came, largely came in the compute side. And given now that it's $100 million business, could you please help us understand what are the relative sizes of the various components within that business or how big storage versus compute versus others?
Thank you.
Victor Peng
Yes. I would say in the last quarter, it was pretty much compute led the way, followed by networking and then storage.
And I would say that it still continues to look like you know, there’s a good strength in that order, although between compute and SmartNIC, that’s probably going to in any given time vary right, because again this perkiness and just big wins and adoptions and then there's some digestion. But you know, I would say, certainly, we're seeing a lot of strong interest in compute.
And again, to refresh your memory on this, some of the areas in terms of the applications, we're very strong in video, database. You could in fintech in SmartNIC because of the low latency aspect.
You could also think of as computing as well. But certainly, in general, separate apart from fintech, I think SmartNIC, in general, we're seeing a lot of pull, right?
So very pleased about that big win that went that I referenced. But we see continued other, really good momentum and good pipeline development there.
So I would say compute and network are both strong, and it varies from quarter-to-quarter, which is really higher. And by the way, I don't want to say that nothing is happening in storage.
We've seen good wins in storage as well. It's just that I think we've consistently always felt like our opportunity there is probably not quite as large as the other two.
And that's sort of what we're seeing right now.
Ambrish Srivastava
Right. But the relative sizes are networking and storage at then compute today, right?
Victor Peng
I'm sorry, say that again?
Ambrish Srivastava
The relative sizes of the various segments, that was my other question. How big is the --
Victor Peng
I think in the long run, compute is still the largest. And as I said in the last quarter, compute was a very big contributor and then networking next.
And but I guess what I'm just saying is that at any given quarter, it could move between compute and networking and occasionally storage perhaps. But I'd say, as a more, what would I expect integrating over time in the long run, now compute is probably the biggest.
Some of that will take a while to flow through. Networking will be very strong, and we're seeing a lot of strength there.
And then storage.
Ambrish Srivastava
Got it. Makes sense.
Thank you.
Operator
Your next question comes from the line of Tristan Gerra from Baird. Your line is open.
Tristan Gerra
Hi, good afternoon. Just as a follow-up, any type of revenue update that you think you could guide for data center in fiscal 2022, now that you have more visibility with your SmartNIC hyperscaler win?
How should we try to quantify the data center revenue opportunity next year for your -- for Xilinx?
Victor Peng
Tristan, I would say that, well, we do see strength improving in multiple markets. I think we still want to take pretty much a cautious approach here because there's still certain uncertainty in terms of the pandemic, how the Congress response, of course, you have the presidential election.
So I think I'd defer to give any guidance right now in FY 2022. Certainly, when we have our Analyst Day in December, we'll go through all of that.
But what I would -- let me put it this way, though, I would say despite a number of different things happening in this FY 2021, we do still feel like we will hit double-digit growth. And over the long run, as I said in my prepared remarks, we're confident that data center will be our greatest growing market of all the markets we serve in terms of growth rate and that we will have sustained solid double-digit growth over the next several years.
Tristan Gerra
Okay, that's great. And then just a quick follow-up.
It looks like you're not holding XDF this year. I wanted to understand how important XDF is in terms of building developer community around the use of FPGAs in data center?
Is that something that you plan on hosting in the future? Has anything changed in terms of your vision on how to build FPGA traction and virus for data center or are there other means to develop that whole ecosystem?
Victor Peng
Well, I would say the XDF, like for everybody, that was for us a very big event and we literally would bring 1,000 or 1,000-plus in certain geography of people together, altogether. So, obviously, because of the pandemic, we wouldn't be able to do that.
Then we looked at thinking about doing things virtually. And in the end, what we decided to do is actually modify that a little bit, and we have a sequence of different events called Adapt, right?
This is new to continue to essentially convey the same information, get the same kind of users in various people together in a virtual setting. So, instead of having one big bang kind of event, if you will, we're spreading that over a series of events, virtual events that we call Adapt.
So, I wouldn't say that we've changed our view of how important it is to engage with customers, partners, users, and so forth and of course, as well. I just think the mechanism, which we're going to deliver this year in particular because of the pandemic and so forth, we're changing it up a little bit.
That's all.
Tristan Gerra
Great. Thanks again.
Operator
Your next question comes from the line of Blayne Curtis from Barclays. Your line is open.
Tom O'Malley
Hey guys, this is Tom O'Malley on for Blayne Curtis and thanks for taking my question. I just wanted to triangulate a bit more into the data center business.
There was clearly a portion in September that was but the business has grown quite nicely from December to March to June. When you ex out the Poland and you look into December, you're obviously guiding that segment lower.
But should we think about that business growing off of that June base or were the Polands enough where you're resetting to a lower base in December or should you see it higher than that $86 million or so that you printed in June?
Brice Hill
I think when you started out talking about how looking at the run rate overall for the year, I think that's how to really think about how we're growing here. Because, again, the quarter-to-quarter, we've always said typically is bursty and then you add on top of that, very short fuse kinds of government actions that's also caused some perturbations.
But the important thing to look to see is, I think, in this stage of where we are with that segment is the year-to-year progress. And that's why I said, even though we're not generally providing guidance, I just want to say, in general, we still feel confident about, even in this year, growing double-digits, right?
So, I think that's the key thing. I mean, in any given quarter, we are seeing volatility some that -- to some extent, I think, is natural, some that are a little bit unusual just because of the -- I guess, what I'd say is the trade environment today.
And again, I think that we're seeing lots -- the things that we watch for, too, is the fact that we have lots of customer engagements, the fact that we still see our pipeline growing and that we see again, early discussion about not just on the compute side, but on the SmartNIC engagements as well as in storage and memory. So we’re pretty -- we feel pretty good in terms of year-to-year progress, and we're laser-focused on keeping that consistently growing double digits.
Tom O'Malley
Great. And if you could sneak me in for a follow-up.
I just wanted to isolate the medical business. Clearly, there's probably some increased revenue due to what's going on in the world right now.
And you guided aerospace and defense for a record quarter. Can you talk about what percentage of your business is medical?
I know that on prior calls, you indicated that, that percentage was maybe increasing a bit. Can you talk about what you expect for that business, given maybe some pull-ins given the state of the world?
Brice Hill
This is Brice. We don't break out exactly what is medical, but we have seen accelerations in the area for the COVID environment.
And we do think it's a driver for our business going forward long term. We had an announcement this quarter about an MPSoC product that will be deployed for AI X-ray capability with one of our customers.
And we think that's a good example of Victor's strategy of the adaptable SoCs at the edge, bringing more capability in the environment and helping Xilinx address more workloads. So you're right, it's a good driver for us, and it's those types of applications at the edge that will be a driver for that particular end market.
Victor Peng
And by the way, I would also add that aside from direct medical equipment, which we're designed into a lot of places, we also are seeing medical applications and modeling and simulation in the cloud, right? So we've talked about a number of ISVs that are focused on medical and they're doing cloud acceleration.
So we see that also within the DCG business, if you will.
Tom O'Malley
Great. Thanks.
Operator
Your next question comes from the line of William Stein from Truist Securities. Your line is open.
William Stein
Hey. Thanks for taking my question.
Two questions about data center. First, Victor, earlier, you talked about double-digit growth in this market.
And one of the things that I've tried to pay attention to is what you said at Analyst Days. And the last one, I think the company presented this 36% Sam bogey from 20 to 24.
I'm wondering if that's still a realistic expectation? It seems like you're beating it in some quarters, not in others.
Is the long-term view still that you can hit that 36% growth number or perhaps better or worse?
Victor Peng
Yes. I mean, I think what I would say is that, we don't think things have changed that much other than, yes, 2020 was a challenging year for a variety of things.
So there'll be some impact of that. But I'd say, overall, we're still in that ballpark of growth, right?
And again, we'll talk about that at Analyst Day. But we still feel very confident, and we do really think that this is a long-haul kind of strategy, right?
We're not as focused on what happens quarter-to-quarter. We want to make sure that we're progressing year-on-year.
And not just on revenue, but again, the pipeline, our ecosystem development, the Alveo boards business, which, again, I want to bring everybody back to. We had no business there two years ago.
And we're tracking to that being a good portion of our revenue for FY 2021. And so I think that new revenue stream, which is new, we have to stand up an entire different distribution channel, bars and distributors and system integrators, and we've done that now.
And I think -- yes, I think we're -- overall, I would say, you're not going to hear big changes, but stay tuned for the Analyst Day in terms of more granularity on that.
William Stein
That's helpful. If I can have one follow-up?
You've spoken a bit today and in the past about SmartNIC, and I think that's a very good business for you. There's a competitor that's highlighting this product, they call a DPU, although I'd say they're not the only one, there's at least two other companies that have highlighted similar products.
I'm wondering if as Xilinx sees it, is this a marketing distinction, or is this a more meaningful different product? And if it's more -- if the difference is more meaningful, would Xilinx potentially have a play in that market?
Victor Peng
Yes. Let me first sort of say what I think is a good validation, which is that more and more people are saying the SmartNIC, they have to do a lot more customization in the data path, and they can't get away with just mainly a fixed function with a little bit of flexibility that the customers are demanding the ability to do more specialization for their unique needs, whether that's for security or just how to optimize and tune, right, the overall economics of their data center, right.
And so I'd say that's a good trend, and people are now taking different approaches. That particular architecture is using a large array of fairly typical embedded processor cores, if you will, which really isn't the most effective, efficient way to do that.
We approach it, obviously, with really very, very flexible and adaptive infrastructure that we have on our products. And we still have ARM SoCs.
So from a software program portion of it, that's pretty typical. But how we actually do the acceleration and how we allow people to customize in the data path is, I think, much more powerful to deliver high-performance as well as low latency, and really, again, customized to exactly what they need.
Plus, I mean, you could make very significant changes on the same piece of silicon after things are So just in terms of being able to make new features, capabilities or fixed new software homes -- I mean, sorry, security homes afterwards, we provide the most feature proofing, if you will. So I think we have that strong capability.
I think I would also add one final thing is that our software and systems knowledge was significantly increased when we completed the Solarflare acquisition. So not only do we have this really uniquely powerful hardware, adoptable hardware, but now we also have really good experts in terms of production and quality drivers and overall system expertise.
So I think that certainly served us well in competing against all the top players, including the one that you mentioned.
William Stein
Thanks Victor.
Operator
Your next question comes from the line of Christopher Rolland from SIG. Your line is open.
Christopher Rolland
Thanks for the question guys, and solid quarter. My questions are around WWG, and even in the late stages of 4G, you guys were doing more than $200 million a quarter likely in WWG.
And I know we're on the early side here, but I thought we were expecting a content bump and then also historically, FPGA has been on the early side of the deployment. So I guess, why aren't we seeing more versus our expectations some ways back?
And then also perhaps you had a competitor, Qualcomm. It sounds like it's getting into the macro business as well.
Do you have any views on their parts and whether they are a competitive threat for you guys in WWG? Thanks.
Victor Peng
Yes. So let me address a number of points there.
So maybe on the -- we typically enable people to get there really quickly. I think we did see that, right?
I mean, I think what happened with Samsung in deploying in South Korea ahead of all geographies. Frankly, even ahead of China, which I think everybody largely expected would go first and we were there, right?
And we're there. In fact, we were there in such a big way.
We were totally candid in saying that we don't expect to hold on to some of that because we enabled exactly what you said, getting to market very rapidly, and then over time, in the area in terms of the baseband, where we're traditionally not quite as strong, we got displaced. But we continue to grow where we have been traditionally been strong.
And in fact, we have a strong hand in the radio. And again, it's going to be more radios, more different configurations with massive MIMO many different bands and so on.
So I think we still absolutely believe that we will beat the 4G kind of run rate, but again, things are kind of in the early stages. And I guess the other thing is, I don't want to keep getting back to this trade thing, but it is quite significant that at this point one of our top customers and communications is zero, right?
So I just want to say that, that is a headwind that doesn't just disappear overnight. In fact, I would say, considering we're holding, losing one of our top customers in any segment, but in communications and especially 5G wireless, that is a significant challenge.
But despite that, we still see that we're going to exceed that. And what wasn't around in the 4G, again, not only because there's so much more complexity and challenges in 5G, but the whole evident that we talked about O-RAN, I think that's a really potentially very big opportunity for us, and we're the ideal solution for that.
So stay tuned. It's still -- I know it's -- everybody is looking for the quick inflection, but this is a long journey here, and I think that we're still in very good shape.
Christopher Rolland
Yes, I think those are fair points. Do you have anything to say on the Qualcomm announcement and...
Victor Peng
I'm sorry, during the -- yes. Look, that just broke now, and look, I'm not going to pretend to be fully up to speed on that.
I guess it doesn't surprise me again because 5G is such a big opportunity, and it's just showing folks who adjacencies are now wanting to move in. And look, I look at it more competition keeps you on your game, right?
And that's why I feel like we're in a very strong position because we certainly upped our game with integrating things like analog capability and RFSoC, we've done with Versal, and you'll hear more announcements soon about other technologies that we're bringing to bear in that market.
Christopher Rolland
Thanks, again, Victor.
Victor Peng
Yes, thank you.
Operator
Your next question comes from the line of C.J. Muse from Evercore ISI.
Your line is open.
Kevin Feeney
This is Kevin Feeney on for C.J. So just want to talk about wireless, and I was just kind of curious on what are a couple of, I guess, the critical inflections you see over the next couple of years?
Like is that 5G and C-Band coming out or some of the other things you see ramping up? And then I guess, can you talk a bit more about design win activity you've seen maybe in second-generation equipment versus what you were seeing in first-gen equipment?
Victor Peng
Yes. I think in terms of the second generation, and again, people are in different stages.
Some people are still feel early, others are pretty far along. I think one of the things that we think that just from a timing perspective, right, Versal was really not going to be in time for the first wave.
But Versal is there for the second and again, the public, the one that's publicly announced is the work we're doing at Samsung, and we feel very good about that. Again, we have more coming down the pipe in terms of the RFSoC family.
So, that's been really successful, and we've seen that deployed. I think we've talked about how a lot of the regions have not really deployed very significantly.
And so I do think the fact that North America is starting to do their deployments and drive is a good sign. I think we -- I think we've consistently felt that most of the action is going to be in sub-6, but millimeter wave will be there.
And our SoC is being deployed in some millimeter wave systems as well. So, yes, I think it's still pretty early because we haven't seen those second-generation equipment being deployed, but we're going to be in a lot of those systems, both with Versal and RFSoC.
And again, stay tuned for that.
Operator
Your next question comes from the line of Matt Ramsay from Cowen. Your line is open.
Matt Ramsay
Thank you very much. Good afternoon.
Victor, a lot have been focused on the call here on WWG and the data center businesses where there's potential inflections, and you guys are battling it out for design wins and that all makes sense. I wanted to actually focus on the primary AIT and ABC businesses.
I think yourselves and your customer base have probably had some time to digest and another hiccup in the silicon road map at your primary competitor, and you guys have done some great work on those businesses with consistent road maps for a while. So, I just wonder if you could update us on sort of the broad-based design win environment that you're seeing versus Intel, Altera.
And what that might mean for the next several years? And then just a quick one for Brice.
I noticed it was a 14-week quarter in the guidance, and that's probably a holiday quarter, but if that's significant or not, that would be helpful. Thank you.
Brice Hill
Yes, Matt, this is Brice. Thanks for the question.
I'll start and then Victor, I think, will add a couple of points on. First, on the 14 week, we do think it will be relatively small.
Those are holiday days, but we do have approximately $10 million in our forecast for that 14th week. So, that's about the size of it.
When we think about the core markets that you talked about, really, I think the first thing I'd say is our strategy for the adaptable SoCs, you could really see in some of the announcements during the quarter of getting designs in various end markets that are going to help propel the business faster than just the traditional FPGA market, but also gaining new sockets. Examples were in automotive, where we had two announcements on MPSoCs for automated driving systems and 4G radar.
I mentioned the example of the health care, AI, x-ray design win that we had. So, we look across those end markets, and we see that in the short term automotive, and broadcast are going to grow quickly as the economy recovers, but there's also drivers as those devices get put in play in the edge and in autos.
And we think that will be significant for the company. And then on the larger section of end markets, aerospace and defense, Victor mentioned the radiation tolerant device that we got an award for.
We also think that radar innovation will help drive utilization for our devices going forward. And then when we think about industrial, the recovery is strong in industrial.
We think that's tracking the economy. And again, we just focus on the edge opportunities we have for machine learning and AI and high performance, low power device operations, which is where our products specialize.
And then finally, we talked about test measurement and emulation, and that's a big business for us, and we expect to have a strong quarter going forward. And we talked about our $35 billion transistor device that we have for that market that is optimized, and we think that, that market grows over time with product complexity.
So really good drivers across all those businesses and we're confident that we'll have growth in the coming years that will be beyond what's normal for the traditional FPGA market.
Brice Hill
Yes. The only thing I'll add with regard to that competitive angle that you asked about is that, yes, we continue to and we're confident we'll continue to get our share relative to them overall.
In fact, I think we've said it before in many instances now, we don't -- if they do compete, they're not necessarily in the short list of, I'd say, the most challenging competitors when we get down to design wins. Very often, it's really some other kind of product or certainly from a vendor perspective.
I'd like to feel that most of that is really attributed to the fact that we're innovating and executing so well, as you said, the fact that they may have some other challenges in addition to that. Yes, that could have some effect.
But I think, by and large, we're making our own opportunity there, and we think that will continue.
Matt Ramsay
Thanks. Really appreciate the color there guys.
Operator
Your next question comes from the line of Michelle Waller from Needham & Company. Your line is open.
Michelle Waller
Hi, guys. Thanks for taking the question.
I'm on for Quinn. So just one quick one for me.
It seems like Nvidia has been gaining strong momentum in inferencing, based on some of their recent benchmark results. Just wondering if you guys can update us on your progress in inferencing, and how that opportunity is shaping for you?
And actually -- and then, a quick one for Brice, just on the tax rate beyond the December quarter. It looks like it's been fluctuating over the past few quarters.
And just kind of wondering what we should assume beyond the December quarter?
Brice Hill
Okay. I'll tackle the tax rate first, and then Victor will comment on the other.
So for the tax rate, yes, you probably noticed in Q2, we had an ultra-low tax rate, 1% on the non GAAP. And what really happens in Q2 is our share-based awards vest.
And if they've appreciated, we get a tax benefit for that. So that's what's driving the lower tax rate for the quarter.
When we look forward, you see a more normal rate. So, Michelle, the Q3 guide is probably more normal.
And then when we think about longer term, we'll probably increase 1 or 2 points over time as our Singapore rate goes from 0% to 5%. So, hopefully, that answers the question.
Michelle Waller
Yes. Thanks
Victor Peng
Yes. And then, in terms of inference.
First of all, I want to make sure everybody really does understand an inference goes well beyond the data center, right? So, for instance, all those applications you kind of referred to, recent ones in automotive wins, were in wins and other kinds of edge applications that I don't mean just edge computing, but just edge devices as well as endpoint devices.
So we're doing very well, I think, in the broader inference. Within data center, I think we've always said is that most applications are not completely dominated by the AI portion, the neuro network processing.
And we're able to accelerate other portions of the application. That's why we kind of coined for whole application acceleration, right.
And so I think the areas where we tend to shine is where we're actually accelerating other parts of the application as well as including AI. And yes, we continue to be focused on inference.
We're not really driving towards training. So the competitor you mentioned what they're doing, that's fairly new.
We'll see how that plays out. And -- but again, I think that we're accelerating not only AI inference in the cloud, but also just applications that don't necessarily have AI at all.
And I think that's what to think about it is that we are not building hardware that has to be dedicated to only do one thing. Again, we have this great dynamic range, if you will, of all these things that we can accelerate and support, and enable people to really differentiate.
Michelle Waller
Okay. Thanks.
That’s helpful.
Operator
Your next question comes from the line of Srini Pajjuri from SMBC. Your line is open.
Srini Pajjuri
Thank you. Hi, guys.
A couple of questions, Victor. First on the SmartNIC market.
It looks like a lot of you guys are talking about DU and Marvell and NVIDIA, et cetera, it looks like there's an inflection in the market itself. I know SmartNIC has been around for a while.
I'm just curious as to what's driving that inflection? And then I think you did mention about $100 million potential from one of the design wins.
If you could give us some idea how big that business is for you? And when do you see that hitting that $100 million run rate?
And then I have a follow-up.
Victor Peng
Okay. Yes.
Look, I think the reason why there's so much momentum towards SmartNIC is, one, is data centers are means to upgrade just the bandwidth that they have. But I think the clear recognition that having driver code to support more networking is really causing the economics of the data center to be somewhat bottlenecked by that.
So by offloading, running that code, which you don't get paid for right, you don't get monetize, monetize those codes by doing that. If you could shift that off to the -- to a SmartNIC rather than just a dumb NIC, then that's a big win in and of itself.
But then there's -- it goes beyond that because there's lots of things that people want to do in terms of security. And then indeed, you could actually do computation, right, well, data is in motion, right.
And I think this is all part of the bigger picture that the data center architecture is being disrupted, right? I think many people really -- people smarter than I in this, coin the phrase that the entire data center is really the computer, that people want to compose resources of how many CPU cores they need, the amount of network down, service level that they need there and then storage, and there’s really going to be computing distributor throughout all of that.
And we play in all those areas, right, with one scalable architecture and that is unique. Other people have to do that with multiple different types of architectures and devices though.
So the SmartNIC is just one part of that whole, I guess what I would say is disruption and revolution in the data center architecture. And again, we -- the flexibility, the ability to customize exactly what people need and also to change things that are pretty significant way after they've been deployed, I think those are the unique value that we enable our customers with.
So in terms of the $100 million that we talked about, I think we referenced that we see that run rate hitting in the FY 2024 time frame. Obviously, the revenue will start somewhat before that, but it will be a fairly quick ramp.
And that is just to be clear, one customer. We've got multiple that we've already won, and we've seen business from.
But obviously, there's different things in the pipeline that we haven't yet converted to wins in revenue. But yes, a lot of good activity in SmartNIC.
Srini Pajjuri
Got it. And then more of a strategic question, Victor.
Given all the M&A that we are seeing, in particular, NVIDIA looking to acquire ARM. It looks like, certainly, there is a case to be made for having both the accelerator and the CPU together or at least as part of the same package or module.
You guys have been -- have done extremely well on the acceleration side. And you also have ARMSoC capability, which you've been investing in the last few years.
So my question is, does it make sense? Or why wouldn't it make sense for you guys to be a bit more aggressive on the CPU front, even kind of targeting things like ARM server market, because that seems to be finally taken -- I mean, I don't want to use word the word taken off, but at least there seems to be a lot of interest in that market, and you already have the expertise in house.
And I'm just curious, given the large TAM opportunity out there, why wouldn't it make sense for you guys to target the market?
Victor Peng
So first I would say is that this group has been very disciplined and respectful about what method top of us not commenting or answering any questions regarding the rumors. So, since you were asking me about how I think about CPU market, I'll focus it that way as opposed to any other noise around M&A.
I believe that Xilinx has got this unique technology and capability. We invented the FPGA in this whole mission of a silicon architecture that after manufacturing, you can modify it to a very large extent through software.
And that being a very scalable architecture, and over time, adding more and more capability to that, to the extent, as you point out, a number of years ago, now we've had multi-core ARM SoCs in there. So we do have SoC capability.
My own background as I was back when I was doing real engineering, a microprocessor designer. So we certainly have many other people with that expertise, but we really feel like we want to stick to the core strength of this company, right, and the core strength of the company is us understanding how to create that hardware architectures that are very capable, including this adaptability, flexibility, together with the software that you need to sort of make that happen, right?
And I think it serves us well to stay to our really core strength, but we do have expertise in those areas. But I think from a strategy perspective, that is not something we're thinking about is going out to server class kinds of CPUs.
I think we have a lot of capability in SoCs and we'll exercise that. But really, it's the combination of all those elements, right, the SoC, together with the adaptable engines together with now our AI engine, a multicore tiled kind of many people consider the spatial processor.
I mean, I think we have pretty much all the skill sets of the future, which is heterogeneous computing, high-performance heterogeneous computing. And that's what we'll focus on, right?
That view of the world as opposed to say, hey, you're not going to take on companies that have historically been all about CPUs and computing. So...
Srini Pajjuri
Got it. Thanks for the time.
Operator
And your final question comes from the line of Chris Caso from Raymond James. Your line is open.
Chris Caso
Yes, thank you. Question is on O-RAN.
And if you could expand upon some of the comments there about where you expect to play on that? And I guess specifically, where would you expect content to go on Xilinx -- content to go in the O-RAN architecture?
And what's the value proposition for Xilinx against competing solutions? Where you guys win on that architecture?
Victor Peng
Well, I think with what we've done with our RFSoC and then future things that we're doing again, you can think of it in one sense is it's consistent with our story of how we're transforming to more of a platform company as opposed to a component or device company. And I think, as I've said, not only is RFSoC a unique silicon architecture, but we have our own IP and we've worked with customers in terms of -- we talked about being formula, but there's other functions, primarily on the radio side and some of the other things in digital front end that we've really gotten a lot stronger on terms of just our own state of the IP and the system what we say.
Now, for O-RAN, we're still partnering, too. So, I'm not around professing to state that we have all the elements.
And from a software perspective, we're partnering both with AltioStar as well as Naviner and we've worked with other players as well. So, I think that a lot of it is not just the silicon, but we've come a long ways in terms of understanding the system solution, right?
And again, you'll hear more about that. And I think in O-RAN, in particular, because O-RAN is going to have to be -- when that's deployed, it has to interoperate with a lot of different things and these deployments are unique and all these different carriers have unique things.
And I think this flexibility that we just talked about that permeates all of our products is going to be very important in the O-RAN situation, right, to be able to not have to do custom silicon for all those things that have one base of hardware, a platform that you could then customize for all those different deployments and different configurations. So, that's why I think we think we're really well suited for O-RAN, and we have both some of the knowledge and we'll partner with others as well.
Chris Caso
Thanks.
Operator
And we have reached the allotted different questions. Mr.
Victor Peng, I turn the call back over to you for some closing remarks.
Victor Peng
Thank you. So, look, I just want to say in closing, I'm extremely proud of my team's excellent execution, delivering these very solid Q2 results despite COVID-19 challenges and even the most recent trade restrictions.
We're making great progress on our data center customary wins and engagements that you've heard, we're seeing good adoption of Alveo, and we continue to grow our software and ISE as well as the broader ecosystem. And we've had a lot of discussion on -- while in the near-term, quarter-to-quarter, revenue is still lumpy, we're very confident in our data center strategy and we're very confident about being able to sustain double-digit growth over the long run.
We expect WWG business to resume growth. Again, as RFSoC volume continues to ramp and then once Versal also begins production deployment, and of course, as 5G, in general, becomes more broadly deployed around the world in all the different geographies.
While O-RAN is still in its early days, it's gaining momentum, and we believe this is a very big opportunity for us in the future. Lastly, our core business is showing strength in several markets, including very encouraging sign for recovery in the automotive business, which was impacted a little bit more than others.
So, we're extremely focused on our platform transformation, growing our ecosystem and deploying a very robust and complete software development environment. We're also enhancing our capabilities to deliver customized solutions for our customers to meet their needs in the most optimal way, balancing performance, power costs, again, enabling very fast time to market and also a degree of future proofing.
So while 2020 has been -- presented several challenges, we remain confident in our strategy and sustained solid growth over the long run. So thank you for joining us this afternoon on the call.
And I'll turn it back to Matt for a few housekeeping comments.
Matt Poirier
Great. Thanks, Victor, and thanks, everyone, for joining us today.
We'll have a playback of this call beginning at 7:00 PM Pacific, 10:00 PM Eastern later today. For a copy of our earnings release, please visit our Investor Relations website.
Our next earnings release date for the third quarter of fiscal year 2021 will be on Wednesday, 25th of January, after the market closed. This completes our call, and thank you very much for your participation.
Operator
Ladies and gentlemen, this does conclude today's conference call. Thank you for participating.
You may now disconnect.