Interview with Robert O'Dell, Co-founder and Senior Vice President of Marketing.

Introduction

Founded in January 2000, Wintegra is a fabless semiconductor company with over 80 employees. Wintegra is structured as a U.S. corporation and has marketing and applications engineering centred in Austin, Texas, and chip design and product engineering in Ra'anana, Israel.

The company's WinPath access packet processor family, which is designed to enable the next generation of access networks, was introduced in the fourth quarter of 2001.

Wintegra has received a total of $39 million in funding in three rounds, and has been generating production revenue since the third quarter of 2002. First-round investors included Concord Ventures, Magnum Communications and Marvell. Strategic investment partner Texas Instruments (TI), Israeli venture capital firm Genesis Partners, and China Development Industrial Bank (CDIB) of Taiwan invested in the second round. The most recent round of $13 million was closed in March 2003.

The President and CEO of Wintegra is Kobi Ben-Zvi, former co-General Manager of the Networking and Communications Systems division of Motorola. Robert O'Dell, who worked in product planning and systems definition roles in the same division, is Vice-President of Marketing.

Wintegra states that it is the first semiconductor company to take a balanced approach to IP and ATM, and believes it is set to become the number one provider of protocol handling components in the access infrastructure market.

Location

While Wintegra's marketing and applications engineering take place in Austin, chip design and product engineering are centred in Ra'anana, Israel. This allocation, says O'Dell, is simply historical. "A lot of us came out of Motorola Semiconductor and the PowerQuicc family of products has historically been designed out of Israel. In the early days there was a lot of movement to Wintegra and a lot of people joined from Motorola."

While Wintegra's CEO, Kobi Ben-Zvi, is based in Israel, O'Dell describes Austin as "a pretty good place to run the U.S. operations of a fabless semiconductor company. The cost of living is pretty good and there's good infrastructure and talent here."

WinPath

Wintegra claims that its WinPath product solutions allow equipment operators to migrate from ATM cells to IP packets without jeopardising existing revenue, investment or technology. WinPath can be employed in such equipment as base stations, media gateways, access multiplexers, service switches and access routers.

Features of WinPath include technologies found in ASICs, network processors, communications processors, and standard peripherals. The product can be 'designed-in' by customers immediately or WinPath can be programmed for customised applications and specific optimisations.

Differentiators

Wintegra believes that one factor differentiating it from the competition is the company's strong focus on packet and protocol handling. Indeed, according to Robert O'Dell, "Wintegra is the only semiconductor company that has as its mission packet handling and protocol handling in the access infrastructure. Within that focus area there is no other vendor that has as focused and as comprehensive an offering."

Also, the company stresses its multi-protocol support ethos and points out that most protocol chips developed specifically for access have been single-protocol.

Added to this, Wintegra believes the fact that the company has a software-compatible architecture enabling packets and throughput rates to be addressed "from 150,000 to well over a million packets-per-second" is another significant advantage over competitors. "That kind of packet-per-second rate is what access needs," says the company. Wintegra stresses that it has a 10 to one range and that, when a protocol is written, it runs on the entire processor family.

Wintegra differentiates its products by frequency and the number of engines on the data path and has products with one, two and four engines and frequency variations from 166 to 233. This, according to Wintegra, means customers can develop one project at one performance range and then move up or down. "This hits the 'sweet spot' in the access range," says O'Dell.

In addition, Wintegra says that - unlike network processor companies - it has itself developed the data path software on which its products are based. Critically, according to O'Dell, "Whereas typically with a network processor 90% of customers will customise it and write their own - they end up having to develop their own - with our customer base it's the opposite in that 90% of the customers don't develop their own."

The fact that customers use Wintegra's data path software and overwhelmingly choose not to make modifications is seen by Wintegra as a definitive testament to the company's strengths.

Competition

Wintegra has a number of competitors in the access space, who, the company believes, can best be outlined in terms of the four broad categories of component solutions used in access networks for protocol handing.

According to the company, "At the low-end you see microprocessors being used for packet handling and interworking functions - from companies like Motorola and the PowerQuicc family and so forth".

"At the high-end you see network processors trying to downgrade themselves - in terms of cost, power and space - to try to get into the access space. In that area you see all the traditional network processor players - Agere, Intel and so on."

"Then you have ATM switching chips, from companies like Agere and PMC-Sierra. There are also a number of chips out there - from companies like MindSpeed Technologies - for ATM SARs and HDLC controllers."

"Finally, the fourth category of competition, is the customer building his own FPGA or ASIC. We compete against a lot of people but what we bring is the performance scalability and full software solution into the access space, which just hasn't been done before."

Architecture

Wintegra stresses that, though it is not a physical layer company, it handles both control path and data path functions on the same chip. As a result, according to the company, "It's more than just a network processing kind of play because we're doing control path."

The complexity surrounding protocol handling previously caused certain vendors to advocate immediate migration to IP packets. However, Wintegra believes the process will be gradual in existing infrastructures and that it will require multi-service support, such as ATM and Frame Relay.

The company states that regional and regulatory concerns limit the migration while the fact that established carriers own over 90% of the local exchange market and have large capital investments that must be depreciated also adds to the need to interwork with branches of the network relying on legacy technology. In addition, Wintegra sees outstanding QoS issues where IP is concerned.

Wintegra believes these IP QoS problems will begin to be solved in commercial IP networks in the next two years, and commonly adopted in four years, and that MPLS will play an important role in this process. In the meantime, IP services can be provided over an ATM link layer, given that ATM allows reliable QoS techniques to mix voice and data properly.

Wintegra stresses in particular that it aims to facilitate the migration to IP from all of the different legacy protocol formats. As such, the company has clearly had to develop an architecture able to encompass non-IP protocols.

In this regard, the company feels it was able to correctly forecast industry trends: "If you go back and look at where we were when we started in January 2000," says O'Dell, "in that timeframe the word legacy was hardly heard. It was IP-everywhere, IP-only. The market research numbers at that time suggested that IP was going to dwarf everything."

And, says the company, "At that time we took a contrarian view and said we don't believe it will be that fast. Based on our experience in this business over the past 15 years - many of us working at Motorola - we didn't believe it would happen so quickly."

As a result, Wintegra developed an architecture competent in both ATM and IP. "And it turns out that we were one of the very few to think that way," says O'Dell. The company states that, by way of contrast, many competing architectures started with IP and subsequently tried to work back, while others focused on what was needed to be successful in ATM and then tried to migrate to IP. "But we started with both," says Wintegra, adding, "Now that vision has expanded to Frame Relay, and point-to-point protocols and covers the whole space - TDM and Ethernet as well".

Product development

The trend towards IP has influenced Wintegra's product development in a number of ways. "What it has meant for us," says O'Dell, "Is that we've had to spend a considerable amount of time getting the legacy done." And, adds the company, "Since we have the IP capability we've been extending the ability to interwork IP to everything else. In order to cover the 'everything else' we've had to spend considerable time on non-IP-related protocols."

Wintegra says it currently offers approximately 40 protocols on the same chip, less than 25% of which are IP-related. "One of the things that we do is solve the problem of how to interwork and migrate to IP, which means conversion."

Wintegra believes that product development at the company is driven by its customers. According to O'Dell, "We're market driven and use our customers to validate market trends. Not everybody is like that. A lot of chip vendors will find one Tier 1 to sponsor a new chip but we tend to look at market trends and then try to validate that with Tier 1s".

Wintegra states that customers have influenced the "depth and nuances" of its product features, largely in terms of the sub-features and the levels of flexibility, in ways that the company might not have considered. "There are a number of cases," says O'Dell, "where customers have suggested capabilities that we went out and implemented after validation in the market". By way of an example, the company points to the inclusion of network address translation within its DSL packages, an area of development described as an emerging trend.

Partnering

Wintegra has a strong relationship with strategic investment partner Texas Instruments (TI), which invested in Wintegra's second round financing in March 2003. "We consider ourselves a partner of TI", says the company, "And we work with different business units depending on the different market areas".

The relationship allows both companies to plug gaps in their product offerings in that, according to O'Dell, "TI is number-one in digital signal processing (DSP) in the access space and we are not a DSP company - we go alongside those DSPs in many applications to aggregate streams of bits from various DSPs and to perform protocol conversion and QoS functions for TI. We do that in several different areas, such as wireless and VoIP and DSL".

Market focus

Wintegra states that, in terms of sales categories, it has certain major focus areas. "Wireless is one area that we tend to track," says O'Dell, "DSL or wireline access in general is another. Also multi-service access and VoIP." In addition, adds the company, "Optical CPE is another focus area for us - SDH to the business or PON kind of techniques and that's something we certainly see as a trend."

Wintegra stresses that in each of its focus areas the company has seen Tier 1 wins. "We have over 90 design wins," states O'Dell. In the wireless sector in particular, Wintegra sees signs of renewed life. "I think that there's certainly a lot more activity going on in wireless - particularly 3G wireless," says O'Dell. "There's a new wave of designs starting." Wintegra sees evidence that customers are starting to revamp their base stations or radio network control devices in this area.

The company also sees renewed life in the voice coding systems area and wireless media gateways. "But," adds O'Dell, "DSL is also hot and has always been hot." Wintegra also sees a trend towards recognising the importance of content within the network. "So regardless of the medium, regardless of the layer 2 protocol, regardless of whether it's layer 3," says the company, "The importance of identifying, prioritising and routing packets based on the content is a key trend."

Wintegra states that it has a "fairly balanced" portfolio of projects spread between the U.S., Europe and Asia. "It's not quite a third, a third and a third but it's close to that," says O'Dell, who adds, "DSL adoption in Asia has been amazing to watch. We're continuing to open new sales offices. We have one in Paris and Seoul, Korea, and have more planned for later this year."

In addition, Wintegra operates a systems-engineering office in Glasgow, Scotland. "We did that six months after we started the company in order to get our European focus," says the company, "Glasgow is a good launching point to the rest of Europe. Scotland seems to have a good ability to work with the various countries and cultures."

In addition, says Wintegra, China is an area of focus and the company has a representative sales office in the country. "Infrastructure in China has not developed as much as elsewhere yet," says the company, "But give it time..."

Future strategy

Wintegra, which has to date taken $39 million in equity investment, says it expects to seek an initial public offering (IPO) with its current level of equity investment. According to O'Dell, "We expect to be profitable this year and IPO in mid-2005 is our expectation."

In terms of a strategy to build the business, the company states that it sees significant opportunities and significant work still to be done in the access infrastructure area. "The best way to grow the company is to fulfil our mission of being the number one supplier of protocol handling components into that space," says O'Dell, who adds, "We still have a lot of work to do. We're on a great growth track but there's still a lot of untapped opportunity for us."

As with any fast growing company operating in uncharted territories, Wintegra's strong momentum brings with it the possibility of derailment. "There are always things that we could write in the risk statement of a 10-Q," says O'Dell. "There's a ton of things that could affect us." However, the company believes that, broadly speaking, the level of threat has receded in recent months.

"The biggest thing I could have written in a 10-Q six months ago," says O'Dell, "would concern the continuing telecom carrier spending. But now that seems to have reversed course. It looks like we've found the trough finally."

This optimism is largely shared by other semiconductor vendors the company is in contact with, says O'Dell, as well as by many customers in the access space. Also, adds the company, "One of the things that we do see is an urgency from the customer base to get products out and that was not the case 18 months ago. It's now opportunity verses budget."

O'Dell stresses that the biggest concern from customers 18 months ago was controlling their budget, "And if that meant slowing down a project or building your own software in the control path rather than buying a control software package from a third party provider then it took precedence."

Now, however, says the company, "The schedule is the focus - it's about getting the product out. And if that means spending some upfront investment and hiring a few people then so be it."