EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
We call it trivial only in the context of our overall short thesis, in that an NP-5 design loss is only one of many ways that EZCH would fail to hit the Street's 30% growth target. It's hard to argue that the entry of a lower-cost, horizontally-integrated competitor into this niche field won't at all challenge pricing and market share. Perhaps EZCH does have an NP-5 design win in one class of routers, but maybe Cisco and ZTE also source BRCM silicon for another 100Gb device. Or maybe the vendors leverage BRCM's presence to negotiate EZCH down on pricing. And Huawei's order delay might be because they are using BRCM silicon (EZCH only speculates that it's in-house). At a price of 32x P/E, the valuation does not incorporate the potential for these suboptimal scenarios to play out.
As for the specific issue of ZTE and Cisco being customers of EZchip's NP-5 product, I recognize the disclosure you point out. Yes, "Cisco has selected a customized version of the NP-3, NP-4 and NP-5 for its principal CESR platforms" and "ZTE has also selected the NP-4 and NP-5 for several of its CESR platforms." These statements don't elaborate on the extent of the Cisco / ZTE relationships / contracts with respect to the NP-5.
Management had plenty of opportunity to clarify the extent of the NP-5 contracts in its conference call. Here's the call transcript: http://seekingalpha.co...
Search for the term NP-5. And see for yourself whether the company uses language that confirms that customers have already selected the NP-5, or that management "believes" and "expects" NP-4 customers to transition to the NP-5.
EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
You laud EZchip's customer base. OK, let's talk about that customer base. EZchip earns 40-50% of its revenue from a customer that could build an in-house chip within a year or two, should it decide to dedicate resources to the task. EZCH longs argue that Cisco won't bring designs in-house because it would distract focus from other areas. OK, this seems reasonable when Cisco's costs for the EZChip designs are $20m to $30m (we're not even including the costs that go to MRVL for the manufacturing). But the EZchip long thesis is based on revenues doubling over the next 3 years. Would Cisco still be willing to use EZchip designs if it were costing it $70m to $100m? And frankly, why would Cisco even be willing to pay that much? Why not just tell EZchip that they don't want to. What is EZCH going to do -- stand up to Cisco and lose 50% of their revenue base?
Another 20% of revenue is transitory, as Juniper decided to move to in-house silicon in 2009. ZTE is a lumpy, unpredictable 10% contribution and EZCH has little visibility on their near-term spending habits. The final two Top 5 router vendors are Alcatel (not a customer) and Huawei (a partial customer in the best-case scenario). Examples abound in the tech hardware sector of niche merchant vendors with extreme valuations whose shares fell overnight on the loss of a key customer account.
I'm not sure whether anyone knows with certainty whether BRCM is selling NPUs to Cisco. Broadcom earns $8bn a year, making it unlikely that they would call out a design win that stole a measly $25m of revenue from EZchip. But BRCM is just one of the many risks present in this name. What you've neglected to focus on are EZCH's production delays, constant shareholder dilution, the low likelihood that management can live up to the market's expectation of 30%/year revenue growth, etc. That 30% growth hurdle is the consensus view. Anything less and investors/analysts will be disappointed and the stock likely falls.
EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
Thanks. Management's language on this issue has been vague. But it's really not an important part of our argument for why we think EZCH is overvalued. So rather than get into a back-and-forth debate about vague management commentary around a trivial matter, we just asked Seeking Alpha to remove that argument. The article has since been edited.
EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
Mullen,
Thanks for the comment. I'll respond from two angles.
First, our central thesis is that EZCH shares are overvalued, in that the current market cap is far too high relative to a reasonable expectation of what this company's discounted future cash flows are going to be. In other words, even if EZCH does retain its NPU customers, the presence of BRCM will prevent EZCH from being able to command high prices from its customers for it next-generation NPU products. If EZCH gets too aggressive about pricing, Cisco and others would be better off re-focusing efforts to bring designs in-house, or would go with BRCM. So the entry of BRCM is yet another reason why EZCH is unlikely to meet its lofty revenue targets.
Second, you're right that Broadcom's track record is not beyond critique and it could fail to gain customers for its edge router products. But it's nevertheless dangerous to completely dismiss the entry of a $20 billion company with 8,700 R&D employees into EZchip's core end-market. The CEO of a $20 billion company does not spend 5 minutes at his annual investor day to expound on a revenue opportunity that doesn't exist. Here's a quote from the Broadcom CEO:
"I want to take one area in particular, which is the network processor space. We've got some tremendous technology from NetLogic. We looked at all the different companies and we evaluated their technology. We're looking to buy one. And we decided NetLogic was by far the best, so we bought them. And how has that done? It's gone very well. We're shipping our 40-nanometer product in volume today. We're sampling our 28-nanometer product, which we believe is best-in-class in the space. We're ramping a lot of new customers, seeing good traction. We've grown that group every quarter since we acquired it. We're gaining share and we're expanding from the data plane into the control plane. So an opportunity to grow into a new space, okay, with those processors. So we're very excited about this technology. We think it's doing very well. We have very high hopes for it going forward." - Broadcom CEO, December 2012 Analyst Day
EZCH now has direct competition in high-end NPUs where it historically had little or none. Whether this competitive dynamic impacts EZCH on price, market share, or growth opportunities is hard to speculate, but we think that the stock isn't fully discounted for this piece of information.
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
Duke,
The technological edge that you base your thesis upon (integrated vs non-integrated) assumes that router vendors are completely price insensitive. BRCM has scale advantages over EZCH that allow it to undercut EZCH while providing similar functionality. I can also pull a Linley quote to support this case, "The BCM88030 has a design win at NEC, and we expect Broadcom to increasingly challenge EZchip for new opportunities" (see the Executive Summary tab). http://bit.ly/11kgnGV#
Moreover, the merchant silicon vs. in-house design is also a price choice. Cisco has the engineers on staff to build an industry leading NPU should it decide. Remember QuantumFlow? Huawei clearly decided that it could do just as well with its in-house engineers at a lower price, hence their lack of EZCH orders in the past quarter. http://bit.ly/11nyPCx
EZCH is priced at such a high valuation (30x+ GAAP P/E), that if Broadcom steals a single EZCH customer, the stock will experience a permanent value impairment. I doubt EZCH would willingly divulge this competitive tension on its quarterly calls, so we'd recommend you also read the Broadcom transcripts to understand what is happening here. Why put your capital at risk in such a precarious scenario?
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
Traffic management, broadly defined, existed even in the NP-2, a device goes back to 2005. This isn't very cutting edge technology and in fact Broadcom already sells integrated traffic managers alongside its 200Gb packet processors. http://bit.ly/XnOXx0
A more credible piece of the bull thesis is that NP-5's integrated TCAM gives them some sort of edge versus Broadcom. But as you point out, BRCM is the leader in TCAM and can easily sell TCAM as an external expansion alongside its BCM 88030. The spec sheet explicitly states this. And as we pointed out in our initial report, Slide 35 of Broadcom's Investor Presentation shows that the BCM 88030 is targeting the Edge Router and not just the Switch market. http://bit.ly/XnOYRM
As for Marvell, we don't think they spent $75m on Xelerated unless they had a specific customer in mind. If this customer were one of EZchip's, management's 3x revenue target by 2016 becomes much more difficult to hit.
ServiceNow: Beating Revenues By $4m Does Not Negate Overvaluation Concerns [View article]
Pone, that's a great question. It's revealing in and of itself that investors/analysts are beginning to contemplate a time when growth slows, the clock strikes midnight, and NOW's loses its Cinderella valuation.
Wall Street expects NOW to grow revenue from $244m in 2012 to $390m in 2013, suggesting $145m of incremental revenue growth this year. Let's assume that the Help Desk market is $2bn in size. Contracts are up for renewal every 3 years and Gartner data shows that 50% of the market want to move to cloud. That implies that $2bn * 33% * 50% = $330m of incremental revenue is available each year. BMC has claims it has a 60% win-rate vs. its largest SaaS competitor (i.e. NOW). If we also assume that Cherwell, SAManage, HP, and CA each have just a 33% win-rate vs NOW, that leaves NOW with 40% * 66% = 26% of the market. This implies that NOW can add 26% * $330m = $86m of revenue each year. Compare that to the $145m of growth expected by that market and you have a substantial gap that can only be filled by ancillary products, each of which are priced at about 1/10 the cost of the Help desk. Unlike the market, we don't believe this is a large enough revenue opportunity for NOW to grow into its valuation. The growth hurdle becomes even higher in 2014, when Wall Street predicts $174m of incremental growth ($390m to $564m).
All of this also assumes that NOW can roll-over existing contracts that come up for renewal, which we estimate is ~$65m this year (est. 2010 revenue). This is almost twice that of the ~$35m of contracts (est. 2009 revenue) that were up for renewal in 2012. As an aside, it is much easier to post a 90% renewal rate when only $35m of NOW's $240m was up for renewal in 2012. As more and more contracts come due relative to incremental revenue, NOW's renewal rate should correspondingly fall. We believe this might take investors by surprise.
This is just one way to think about. We hope it helps.
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
We've read those other research articles ourselves and would have to disagree with the consensus sentiment.
1) Because EZchip has not paid a dividend, does not repurchase shares, and runs an emerging technology business, the likelihood that the cash on hand will be returned to shareholders in the event of a downside scenario is relatively low. So we do not give them full credit for the cash on the balance sheet in the company's valuation.
2) EZchip's NPU is susceptible to replacement at any time. Beyond the immediate competitive threat from Broadcom, EZCH's 'top-tier' customers like Cisco and ZTE have the capability to design a replacement chip within 2 years, should they choose. Cisco, for one, has 750 chip designers on staff. This risk emerged on EZCH's Q4 2012 conference call, where management's comments led many to believe that Huawei chose in-house silicon rather than EZCH. http://bit.ly/13WRMfT
3) While you are correct that EZchip has products in the pipeline, its ability to win customers on the new NP-5 and NPS platforms remains uncertain.
EZchip: Immediate Competitive Threats, Customer Defections And An Unsustainable Valuation [View article]
If an investor published a long thesis on EZCH and the stock went down on the publish date, nobody would claim that it 'backfired.' We remain confident that our investment case will play out over time.
EZchip: Immediate Competitive Threats, Customer Defections And An Unsustainable Valuation [View article]
Thanks for the question mullenji. See slide 8, "EZchip's ASP" in the Q4 2012 presentation linked below. EZCH can only achieve its 3x revenue growth by 2016 if the ASP (average selling price) jumps from a factor of 1 to 1.4x. http://bit.ly/XE7qsL
Mellanox: Growth Has Vanished, Trust Is Lost, And Premium Multiple Should Soon Evaporate [View article]
Intel has already chosen its allegiances with the acquisitions of QLogic's, Fulcrum's, and Cray's interconnect products. Intel has been working with these engineering teams for over a year now with the goal of co-locating the interconnect with the processor in its next-generation HPC server chip.
MLNX's conference calls repeatedly indicate that Mellanox is pursuing an independent solution from Intel and that both striving towards a 100Gb platform by late 2014 or early 2015. But if Intel's next-generation processor only functions with the co-located Intel interconnect, it matters not who get there first. In general, customers upgrade their HPC clusters primarily for the Intel chip and not for the interconnect.
EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
As for the specific issue of ZTE and Cisco being customers of EZchip's NP-5 product, I recognize the disclosure you point out.
Yes, "Cisco has selected a customized version of the NP-3, NP-4 and NP-5 for its principal CESR platforms" and "ZTE has also selected the NP-4 and NP-5 for several of its CESR platforms." These statements don't elaborate on the extent of the Cisco / ZTE relationships / contracts with respect to the NP-5.
Management had plenty of opportunity to clarify the extent of the NP-5 contracts in its conference call. Here's the call transcript: http://seekingalpha.co...
Search for the term NP-5. And see for yourself whether the company uses language that confirms that customers have already selected the NP-5, or that management "believes" and "expects" NP-4 customers to transition to the NP-5.
EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
Another 20% of revenue is transitory, as Juniper decided to move to in-house silicon in 2009. ZTE is a lumpy, unpredictable 10% contribution and EZCH has little visibility on their near-term spending habits. The final two Top 5 router vendors are Alcatel (not a customer) and Huawei (a partial customer in the best-case scenario). Examples abound in the tech hardware sector of niche merchant vendors with extreme valuations whose shares fell overnight on the loss of a key customer account.
I'm not sure whether anyone knows with certainty whether BRCM is selling NPUs to Cisco. Broadcom earns $8bn a year, making it unlikely that they would call out a design win that stole a measly $25m of revenue from EZchip. But BRCM is just one of the many risks present in this name. What you've neglected to focus on are EZCH's production delays, constant shareholder dilution, the low likelihood that management can live up to the market's expectation of 30%/year revenue growth, etc. That 30% growth hurdle is the consensus view. Anything less and investors/analysts will be disappointed and the stock likely falls.
EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
EZchip: Bullish Case Continues To Myopically Overlook Competitor Comments, Production Delays, And Stock-Based Compensation Expense [View article]
Thanks for the comment. I'll respond from two angles.
First, our central thesis is that EZCH shares are overvalued, in that the current market cap is far too high relative to a reasonable expectation of what this company's discounted future cash flows are going to be. In other words, even if EZCH does retain its NPU customers, the presence of BRCM will prevent EZCH from being able to command high prices from its customers for it next-generation NPU products. If EZCH gets too aggressive about pricing, Cisco and others would be better off re-focusing efforts to bring designs in-house, or would go with BRCM. So the entry of BRCM is yet another reason why EZCH is unlikely to meet its lofty revenue targets.
Second, you're right that Broadcom's track record is not beyond critique and it could fail to gain customers for its edge router products. But it's nevertheless dangerous to completely dismiss the entry of a $20 billion company with 8,700 R&D employees into EZchip's core end-market. The CEO of a $20 billion company does not spend 5 minutes at his annual investor day to expound on a revenue opportunity that doesn't exist. Here's a quote from the Broadcom CEO:
"I want to take one area in particular, which is the network processor space. We've got some tremendous technology from NetLogic. We looked at all the different companies and we evaluated their technology. We're looking to buy one. And we decided NetLogic was by far the best, so we bought them. And how has that done? It's gone very well. We're shipping our 40-nanometer product in volume today. We're sampling our 28-nanometer product, which we believe is best-in-class in the space. We're ramping a lot of new customers, seeing good traction. We've grown that group every quarter since we acquired it. We're gaining share and we're expanding from the data plane into the control plane. So an opportunity to grow into a new space, okay, with those processors. So we're very excited about this technology. We think it's doing very well. We have very high hopes for it going forward." - Broadcom CEO, December 2012 Analyst Day
EZCH now has direct competition in high-end NPUs where it historically had little or none. Whether this competitive dynamic impacts EZCH on price, market share, or growth opportunities is hard to speculate, but we think that the stock isn't fully discounted for this piece of information.
Amerco: Move And Store Your Cash In This Stock [View article]
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
The technological edge that you base your thesis upon (integrated vs non-integrated) assumes that router vendors are completely price insensitive. BRCM has scale advantages over EZCH that allow it to undercut EZCH while providing similar functionality. I can also pull a Linley quote to support this case, "The BCM88030 has a design win at NEC, and we expect Broadcom to increasingly challenge EZchip for new opportunities" (see the Executive Summary tab).
http://bit.ly/11kgnGV#
Moreover, the merchant silicon vs. in-house design is also a price choice. Cisco has the engineers on staff to build an industry leading NPU should it decide. Remember QuantumFlow? Huawei clearly decided that it could do just as well with its in-house engineers at a lower price, hence their lack of EZCH orders in the past quarter.
http://bit.ly/11nyPCx
EZCH is priced at such a high valuation (30x+ GAAP P/E), that if Broadcom steals a single EZCH customer, the stock will experience a permanent value impairment. I doubt EZCH would willingly divulge this competitive tension on its quarterly calls, so we'd recommend you also read the Broadcom transcripts to understand what is happening here. Why put your capital at risk in such a precarious scenario?
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
http://bit.ly/XnOXx0
A more credible piece of the bull thesis is that NP-5's integrated TCAM gives them some sort of edge versus Broadcom. But as you point out, BRCM is the leader in TCAM and can easily sell TCAM as an external expansion alongside its BCM 88030. The spec sheet explicitly states this. And as we pointed out in our initial report, Slide 35 of Broadcom's Investor Presentation shows that the BCM 88030 is targeting the Edge Router and not just the Switch market.
http://bit.ly/XnOYRM
As for Marvell, we don't think they spent $75m on Xelerated unless they had a specific customer in mind. If this customer were one of EZchip's, management's 3x revenue target by 2016 becomes much more difficult to hit.
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
ServiceNow: Beating Revenues By $4m Does Not Negate Overvaluation Concerns [View article]
Wall Street expects NOW to grow revenue from $244m in 2012 to $390m in 2013, suggesting $145m of incremental revenue growth this year. Let's assume that the Help Desk market is $2bn in size. Contracts are up for renewal every 3 years and Gartner data shows that 50% of the market want to move to cloud. That implies that $2bn * 33% * 50% = $330m of incremental revenue is available each year. BMC has claims it has a 60% win-rate vs. its largest SaaS competitor (i.e. NOW). If we also assume that Cherwell, SAManage, HP, and CA each have just a 33% win-rate vs NOW, that leaves NOW with 40% * 66% = 26% of the market. This implies that NOW can add 26% * $330m = $86m of revenue each year. Compare that to the $145m of growth expected by that market and you have a substantial gap that can only be filled by ancillary products, each of which are priced at about 1/10 the cost of the Help desk. Unlike the market, we don't believe this is a large enough revenue opportunity for NOW to grow into its valuation. The growth hurdle becomes even higher in 2014, when Wall Street predicts $174m of incremental growth ($390m to $564m).
All of this also assumes that NOW can roll-over existing contracts that come up for renewal, which we estimate is ~$65m this year (est. 2010 revenue). This is almost twice that of the ~$35m of contracts (est. 2009 revenue) that were up for renewal in 2012. As an aside, it is much easier to post a 90% renewal rate when only $35m of NOW's $240m was up for renewal in 2012. As more and more contracts come due relative to incremental revenue, NOW's renewal rate should correspondingly fall. We believe this might take investors by surprise.
This is just one way to think about. We hope it helps.
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
EZchip: Customer Concentration And Uneven Revenue Stream Have Forced Risky Shift To The Data Center [View article]
1) Because EZchip has not paid a dividend, does not repurchase shares, and runs an emerging technology business, the likelihood that the cash on hand will be returned to shareholders in the event of a downside scenario is relatively low. So we do not give them full credit for the cash on the balance sheet in the company's valuation.
2) EZchip's NPU is susceptible to replacement at any time. Beyond the immediate competitive threat from Broadcom, EZCH's 'top-tier' customers like Cisco and ZTE have the capability to design a replacement chip within 2 years, should they choose. Cisco, for one, has 750 chip designers on staff. This risk emerged on EZCH's Q4 2012 conference call, where management's comments led many to believe that Huawei chose in-house silicon rather than EZCH.
http://bit.ly/13WRMfT
3) While you are correct that EZchip has products in the pipeline, its ability to win customers on the new NP-5 and NPS platforms remains uncertain.
EZchip: Immediate Competitive Threats, Customer Defections And An Unsustainable Valuation [View article]
EZchip: Immediate Competitive Threats, Customer Defections And An Unsustainable Valuation [View article]
http://bit.ly/XE7qsL
Mellanox: Growth Has Vanished, Trust Is Lost, And Premium Multiple Should Soon Evaporate [View article]
MLNX's conference calls repeatedly indicate that Mellanox is pursuing an independent solution from Intel and that both striving towards a 100Gb platform by late 2014 or early 2015. But if Intel's next-generation processor only functions with the co-located Intel interconnect, it matters not who get there first. In general, customers upgrade their HPC clusters primarily for the Intel chip and not for the interconnect.
Amerco: Move And Store Your Cash In This Stock [View article]