What will be the next hot Wall Street IPO? Some will shake their heads and think it a joke, and there may be something in that. But precisely now, when we are counting the companies that will not accompany us down the road; when good economic news in the world in general and Israel in particular is hard to come by; these are precisely the times when we inveterate optimists would like to mention the companies that will lead Israel's technology industry and reach a Nasdaq IPO.
This week, this exclusive club is joined by Mellanox Technologies. Naturally, we cannot promise that Mellanox will be the next IPO, but we certainly think the company is organizationally mature for an IPO, and could reach the final lap for an IPO in terms of business readiness.
Mellanox has the trinity needed for success: advanced technology for a large market; a strong management team; and leading investors. It seems that this combination, together with other factors, will make this company one of the next winter's surprises. Mellanox recently raised $56 million at a company value of $250 million.
Mellanox was founded by Eyal Waldman, a founder of Galileo. and a group of senior Galileo managers. Galileo was one of the greatest success stories of Israel’s venture capital industry. The company became a leading chip maker for the telecom industry, with sales to industry leaders. Last year, Galileo merged with US chip maker Marvell Technology Group (Nasdaq: MRVL).
Company investors include industry leaders. Two of them can legitimately be crowned “The Most Important Jews in the Venture Capital Industry”: Sequoia Capital partner Pierre Lamond and US Venture Partners general partner Irwin Federman. Sequoia’s investments include Cisco Systems (Nasdaq:CSCO), Redback Networks (Nasdaq:RBAK), NetScreen Technologies (Nasdaq:NSCN) and the recent Israeli success, Verisity (Nasdaq: VRST). US Venture Partners invested in a rather small Israeli company called Check Point (Nasdaq: CHKP), and is a member of Check Point’s board of directors. There are other successful exits as well. Another noteworthy investor in Mellanox is Atiq Raza, chairman and CEO of the eponymous Raza Venture Management, and who earlier served as CEO of chip maker Advanced Micro Devices (Nasdaq:AMD).
“Globes”: Galileo, led by Avigdor Willenz, is considered to be a company whose spirit included idealism. Does Mellanox have a similar spirit?
Waldman: “I think so. One of the most important things for us is to try to do everything possible in Israel. The cards we developed are manufactured by PCB in Migdal Ha’emek, rather than in the US or Far East. Despite the well-known trend, and the fact that all our investors are Americans, Mellanox Technologies’ parent company is registered in Israel, not the US.”
Are you nuts? That’s a recipe for war with the tax authorities. Whoever heard of an entrepreneur founding a company in Israel?
“It was very important for me that the company be Israeli, and that the company operating in the US be a subsidiary of the Israeli company. We plan to move into another building in Israel shortly, and we wanted to buy Made in Israel office furniture. The problem was that it was too expensive, and we didn’t want to compromise on price. So negotiated at great length with a company until we were able to buy Israeli products at the price we were willing to pay. I consider Mellanox to be a Zionist company, despite the cynicism that statement certainly arouses.”
Will you be forced to sell?
‘Possibly. I’ve learned never to say never. The plan is to build an independent company. As far as we’re concerned, the next stage is an IPO.”
When?
“This isn’t the first time I’ve been asked that question, and my fixed answer is that Mellanox has an experienced board of directors. I am certain that the board knows the proper time to float the company.”
What companies are in your market?
“Intel (Nasdaq:INTC) and Mellanox were the first two companies to have silicon components ready in January 2001. All others were had them only after ten more months, in October. Six companies currently offer InfiniBand silicon components, and the number will grow to about ten. They are Intel, Agilent Technologies (NYSE:A), IBM (NYSE:IBM), RedSwitch, Banderacom and Mellanox. 3Com (Nasdaq:COMS) and [Lucent Technologies (Nasdaq:LU) spin-off] Agere Systems (Nasdaq:AGR/A) have the potential to enter the market, and there are others.”
How big is the InfiniBand market?
“The market is estimated at 6 million servers in 2005. Assuming only half of them use InfiniBand, and each server has two channels, that means 6 million channels. Each channel has two switches, which means 12 million InfiniBand components. In addition, there will be components in routers and core servers. If we add 3 million more components, we reach 15 million, at a conservative estimate. On the basis of a price of $50 per channel, the market is worth $750 million. That is only for servers, and excludes developments in the storage market, which could reach a similar size, and the computer embedded systems market, such as video communications, which amounts to billions of dollars.
“You can compare this with the optical communications market, in which each company builds the systems from A to Z. The InfiniBand market is not vertical, but horizontal. Some companies will make the silicon, some the mother boards, some the systems, and other will design the appropriate software for different functions. The broad horizontal break-down of the market is symptomatic of its large scale, which attracted many companies. Another indication is that giant companies like Intel and IBM are also entering. Companies that want to do everything understand that they cannot succeed, because the market is so huge. Companies like Hitachi and EMC (NYSE:EMC) will also enter the market and give it a further push.”
There is an impression that the market, including the silicon sector, is saturated with frightening giants. What is your advantage?
“I think the first measure is whether the solution meets customers’ needs. We have several advantages over our competitors. First, we have the core of a very good silicon company. Let me explain: most young chip companies do not reach the revenue stage because their technology fails or they are bought out along the way. It's very hard for a small company to stay active in this market. Galileo’s success was quite exceptional. It's hard to reach the sales stage, but we have done it, because we built a quality and balanced skeleton of all the company’s divisions. Second, components architecture is vital. If you compare our chips to Intel’s for example, you’ll discover we have an edge in the price-capabilities ratio.”
Are you cheaper?
“Actually, we’re not. I think we're more expensive, but we offer much better performance. We already have 10 Gbps chips, while Intel’s only offer speeds of 2.5 Gbps. That is a huge difference. Commercial manufacturing customers want the fastest chips, because there is a time lag between beginning production and the commercial stages. If they use the slower products, they’ll be left behind. They will therefore choose Mellanox’s 10 Gbps chips.
“Third, we have excellent connections, even though we are a small player, the underdog. Our customers and partners are the world leaders in the sector. I cannot divulge names, but in the storage, server and communications fields, we are linked with the biggest names in the business.”
Which company has the largest InfiniBand market share?
“My feedback tells me that Mellanox still has the largest market share.”
How did you adjust the company to the economic climate? Isn’t a large financing round usually a recipe for waste?
“We are a very modest company in terms of expenditures. We are based in the Yokne’am industrial zone and we have an office in Tel Aviv. We are on the road to profitability, because we already have substantial revenue and we are very scrupulous about expenditures.”
Waldman declined to confirm revenue estimates of $10 million.
Isn’t there a conflict of interest between you and Intel as the leading competitors in the market, at a time when Intel is also an investor in Mellanox.
“We have an interesting relationship. Intel is both an investor and a partner in all kinds of work we do. At the same time, it is a potential customer and a competitor. The relationship is a strange one, but fortunately the teams managing both companies allow co-existence and an excellent relationship. We are able to do this by cooperating on some things and competing on others. It is quite strange, because Intel did not invest only in us, but also in another competitor in our limited market, Banderacom. As I understand it, Intel did so in order to guarantee the market’s existence and competition. Intel is big and strong enough to do this. They do it quite successfully, in my opinion.”
But when the market matures, won’t they want to eliminate competitors, and have the advantage of knowing your intimate secrets?
“I don’t think Intel has an interest in eliminating us. Intel wants as big a market as possible, and it knows it cannot do that alone. It is necessary to watch market developments. We think as many companies and solutions as possible are needed for the market to develop.’
InfiniBand politics
Mellanox develops InfiniBand chips. The word might not resonate much now, but InfiniBand is one of the newest and most interesting computer markets. Three factors limit computer speed and power: CPU power; the memory’s bandwidth; and input/output (I/O) response speed. When the subject is the computers that manage communications networks, which is moving over to ethernet standard-based integrated optical networks, the I/O bottleneck in network data centers is the most critical problem. The InfiniBand standard is designed to be the solution. InfiniBand is considered to be especially promising, since it was defined from the outset by people who understand the industry and nearly all the industry’s companies are behind it. Intel, Compaq (NYSE:CPQ), Sun Microsystems (Nasdaq:SUNW), Microsoft (Nasdaq:MSFT) and Hewlett Packard (NYSE:HWP) are the leading companies in the field and experts in its problems. They created the InfiniBand standard as the most advanced and the most suitable.
It all started when Intel identified the I/O problem and defined the Next Generation I/O Standard. The decision woke up Intel’s competitors, and they set up a different standard, called Future I/O. A bitter war between the standards lasted for a year, but peace talks opened in October 1999, which resulted in the combined InfiniBand standard.
Throughout this period, Mellanox worked on developing silicon components that would solve the I/O problem. The company initially worked according to the Next Generation I/O standard, and later began developing chips according to the Future I/O standard. Waldman reminisces about those days, “The struggle between the standards was a game of politics and prestige. It was fascinating to watch from the sidelines a struggle in which we had interests and relationships on both sides. We had designs for products according to both standards. When unity was declared, we were in an excellent position and in practice had the technological lead in the market.”
Published by Israel's Business Arena on 13 February 2002