With the soaring demand for bandwidth-hungry services like IPTV, VOD and other forms of digital media, carriers are facing constant pressure to increase the amount of delivered bandwidth. Not only the change is in the amount of traffic, it is in the complexity and variety of the data as well. The days of "one network, one service" networks are long gone, with converged packet-based networks taking the lead. As a result, communication networks are faced with unprecedented challenges which require a higher level of efficiency, reliability and flexibility.
Networking equipment vendors like Cisco have been dependent on in-house application-specific integrated circuits (ASICs) in order to launch improved and advanced products, coping with last years' developments. However, according to certain observers, traffic has reached a threshold, in terms of amount and complexity of data, raising the demand for a more intelligent and flexible solution – high-speed network processors (NPUs).
Network processors combine the speed advantages of silicon chips with the intelligence and flexibility of microprocessors. Offering a combination of performance, speed and flexibility, NPUs allow system vendors to deliver cost effective solutions which can easily adapt to changing market requirements. Despite their advantages, NPUs are relatively expensive, complicated to design and consume a lot of time to integrate.
As there is no argue about their advantages, NPUs’ high price and complexity has always represented a barrier for large-scale adoption. Opinions are split as to the true need for NPUs and their potential market size. Some estimate that NPUs will be the basis of most high-end network boxes while others view them as tiny niche products. On one end of the spectrum we can find George Gilder and the Linley group, who foresee monstrous growth and industry-wide adoption of NPUs. On the other end, technology analysts like Andrew Schmitt believe dedicated hardware can support most of vendors' needs, leaving NPUs as a viable solution only in niche cases "where the flexibility they bring to the table matches the cost and risk premium".
The company most identified with high-speed NPUs, EZchip, has positioned itself as the undisputed leader in the field. Even though current market for NPUs is still small, EZchip has over 100 design wins, including some of the top communication vendors in the world. In its first years, no one was really paying attention to the company, except George Gilder, who has been following and recommending it enthusiastically long before “YouTube” and IPTV became such buzzwords. When asked what made EZchip’s technology so appealing Gilder has a clear answer: Scalability. EZchip’s NPUs are built in such a flexible and sophisticated manner, that it enables them to scale together with market demands and technological advances. In the same manner that Intel’s processors can evolve with the PC market, so can EZchip’s NPUs evolve with the networking equipment market.
EZchip was incorporated as a subsidiary of Lanoptics (LNOP) on December 1999, right before the dot-com bubble burst. By 2002, Lanoptics started shipping the NP-1, its first NPU. Ever since, the company has done nothing but scaling, always one step ahead of market needs. EZchip launched its second generation (the NP-1c) and third generation (NP-2) NPUs on December 2003 and July 2006 respectively. The fourth generation NPU, the NP-3 is already being sampled, while the 5th generation, NP-4, a single-chip 100-Gigabit network processor with integrated traffic manager, is expected to be available for sampling during 2008.
Although EZchip's sales grew on a year-over-year basis for the last 4 years, financial results were unimpressive with most of the design wins not maturing into meaningful sales. Relatively low revenues and rigorous R&D took their toll on the bottom line, with an average annual loss of 10.5$ million for the last 4 years. However, after years of anticipation (and cash burn), it looks like things are finally happening, with 2008 representing the inflection point. According to the company, a substantial growth in the largest NPU market- the CESR (Carrier Ethernet Switches and Routers) market, is expected. This fast growing market, dominated by Cisco, Alcatel-Lucent and Juniper is expected to be worth 6.5$ billion in 2007, growing to more than 9$ billion in 2010. EZchip's management hopes that the migration towards ultra-fast, converged networks will ultimately lead vendors to design and build their platforms based on programmable NPUs such as EZchip’s. In fact, management claims this shift is already occurring.
Where things stand
The CESR market is dominated by three large Vendors (Cisco, Alcatel-Lucent and Juniper), two of which are EZchip's customers.
EZchip is in volume shipments to a tier-1 customer (Rumored to be Juniper) which has based "several" of its new platforms on NP-2. This customer is expected to be active throughout 2007 and 2008, switching to NP-3 some time next year. Juniper’s first Ethernet box, launched October 2006, is thought to be one of several platforms based on EZchip's NP-2.
A second tier-1 customer (Rumored to be Cisco, the market leader) has selected the NP-3 for "several" platforms as well. EZchip has already shipped Cisco a special version of the NP-3, co-developed with Marvell, and expects volume shipments to start in 2008.
The third Tier-1 vendor, Alcatel-Lucent, is sticking with its own ASICs for now, so NPUs are irrelevant for it. Among the rest of the CESR players, EZchip expects to have a 50% market share. Altogether, EZchip's management expects to command at least 60% of the overall CESR NPU market.
In terms of competition from other NPU companies, the two prime competitors are Sandburst (now part of Broadcom) and Xelerated, neither of which has design wins with either Cisco or Juniper. Although the relation between EZchip and its two large customers is not exclusive, EZchip's CEO estimates that chances to be displaced by other NPU companies are low thanks to the "sticky" nature of the business. Each platform containing EZchip's NPUs is a result of a great deal of joint integration and customization efforts, so it is unlikely that Cisco or Juniper will go through such process again with a different vendor. Paradoxically, the complicated design and long customization for each platform become advantageous in fending off competition. EZchip is in bed with Cisco, and plans to stay there.
EZchip's gross margins currently stand at, and are expected to remain around 60% according to management, not bad for a chip company. I wonder whether shipping large quantities to Cisco, who is notoriously known for pressuring its suppliers' margins, won't hurt Lanoptics's profitability. Yet, the fact management explicitly chose to forecast their margins might imply that it already knows how much Cisco is willing to pay.
With OPEX around 3-3.5$ million per quarter, EZchip has a breakeven point of 5.5$ million in revenues per quarter. This level might be achieved quite easily as soon as 2007 with the ramp up of shipments to Juniper combined with several other smaller customers.
Even though EZchip is not a publicly traded company, Lanoptics (LNOP), holds 78% of EZchip's share capital with the other 22% held by Goldman Sachs and JK&B Capital. This makes EZchip an "almost publicly traded" company. Lanoptics has been gradually exchanging its shares for EZchip’s shares and eventually expects to assume full ownership of EZchip. So far, GS and JK&B have been reluctant to exchange their EZchip shares with Lanoptics shares, but this situation won't last for long, as EZchip's CEO estimates a share swap could happen within the next 6-12 months. The current opinion shared by most people who follow this company is that once full ownership of EZchip is achieved, Lanoptics will officially change its name to EZchip, change its Symbol and do a fancy SPO under its new name. The higher Lanoptics' shares climb, the higher the odds are for the share swap to materialize. There are two obvious events that can push Lanoptics shares substantially higher: Reaching profitability and commencement of volume shipments to Cisco. If the CEO believes the conversion will occur in the next 6-12 months, it may imply that either profitability or volume shipments (or both) are 6-12 months ahead.
Current vs. Future Valuations
Lanoptics’ market cap is currently 230$ million, putting a 300$ million price tag to EZchip itself, not cheap for a company who had revs of 4$ million in Q1. However, if the NPU market grows as fast as some forecast and if EZchip is successful in getting large orders from Cisco and Juniper (These are two big “If”s) the price could skyrocket in the coming 12-18 months.
EZchip's visibility is still very low (8-10 weeks) so it does not provide any guidance. However, management does quote future market size and market share estimations that should obviously be taken with a pinch of salt. Nevertheless, the numbers are breathtaking:
The high speed NPU market is forecasted by The Linley Group to be around 225$ million in 2008-9. By that time, EZchip’s two large customers (Cisco and Juniper) are expected to represent more than 50% of the CESR market, and a large portion of the NPU market. Assuming EZchip gets 40% of the market (90$ million), and maintains gross margins of 60% with OPEX of 14$ million annually, leaves them with 40$ million in annual operating profit. Even if in 2008 they earn half of that, a PE of 25 should get their market cap around 500$ million All of the sudden a 300$ million valuation doesn't look overpriced.
When taking a more bottom-up conservative approach, the numbers look saner. EZchip expects to be inside “several” platforms of each of their two large customers. Assuming the minimal number the term "several" can refer to is three, we are looking at a minimal potential of at least six platforms. According to the company, each successful CESR platform from a leading tier-1 vendor can potentially generate sales of tens of thousands chips annually, while each successful platform from a smaller vendor can potentially generate sales of several thousands chips annually. Assuming an average of 25,000 units per 6 tier-1 platforms and 2500 units per 10 small platforms, the numbers of chips sold annually might be around 175,000. Assuming a price per unit of 250$, total sales could reach 43.75$ million. Gross margins of 60% and OPEX of 14$ million, leaves EZchip with 12.25$ million in annual operating profit somewhere in 2008-9. A multiple of 25 assigned to EZchip leads to the current 300$M market cap.
Despite the differences between the two valuations, both represent massive growth for EZchip’s business. Such anticipated growth is certainly not assured, with the actual degree of NPUs adoption remaining unclear. Regardless of EZchip’s future performance, the optimism within the company’s headquarters in northern Israel, is felt all the way to Wall St.
Disclosure: Author has a long position in LNOP
LNOP 1-yr chart