One of the great debates taking place right now within the optical components industry is whether to outsource production, and move to a fabless model like most logic chip makers. One one side of the argument, Finisar (FNSR) is staying in-house, while rival JDS Uniphase (JDSU) is preparing to send its manufacturing operations to Fabrinet in China.

Advanced Technology vs. Standard Economics

Just about every aspect of Cisco's (CSCO) business is envied throughout the network equipment industry. Its balance sheet, its market share, and its distribution network have scared away dozens of potential competitors. Yet somehow the executives at Tasman Drive have become dissatisfied with owning 70% of the routing and switching market, and would now rather compete for 5% of the server market.

While there is little reason for Cisco to expand out of networking, there is also little reason to expect that the rumored [[IBM]] OEM deal can help Brocade (BRCD) take much share from Cisco in the nearly $20 billion a year Ethernet switch market, particularly when the Foundry products it acquired last year have done best in segments where customers buy directly from manufacturers. Additionally, Brocade itself is straying from its dominant position in storage networking in order to pick up a sliver of the LAN switch market.

In an industry with so many new technologies, and so many companies with limited internal R&D resources, M&A strategy is central to long-term success. And while many acquisitions become clear winners or losers in the years that follow the deals, there are signs of trouble that can be spotted at the time any transaction is announced, independent of revenue or earnings multiples.

One of the most noticeable signs of trouble is when an acquired business has significantly different gross margins and inventory turns. This typically means the manufacturing processes will produce few cost efficiencies when merged, because they are so different from each other. Additionally, this also signals an expertise in vastly different technologies at the acquired company, which makes retention of key personnel critical to the deal. These risks do not exist when like companies get together, which happens when data networkers stick to buying other data networkers, and optical manufacturers stick to buying other optical manufacturers. Contrast the successes of Ciena (CIEN)-Lightera, Juniper (JNPR)-Unisphere, and Cisco (CSCO)-Kalpana, to the failures of Ciena-World Wide Packets, Cisco-Monterey, and Nortel (NT)-Bay.

Mellanox (MLNX) reported revenue of $25.2 million on Wednesday, in line with most estimates. While its sales were down 13% sequentially, it was able to manage a year-over-year revenue increase of 1.7% for the quarter.

From a product standpoint, one of the most interesting developments was the rapid increase in 40G, or QDR, revenue. Accounting for just 4% of revenue in the second quarter, and 12% in the third quarter, 40G jumped to 34% of Mellanox's top line in the fourth quarter. While there will be framing, media, and transceiver differences between 40G InfiniBand and the four expected variants of 40 Gigabit Ethernet (due for standardization in June 2010), the company believes its early strength in 40G interconnects will cross over to 40G LANs. While traditionally known for InfiniBand, Mellanox also sells 10 Gigabit Ethernet server NICs, including a recently announced 10GBASE-T LAN on Motherboard Controller that uses Teranetics' PHY chips.

There is a well-known tendency in this industry to emphasize technology at the expense of sales and marketing. It’s never been unusual to see companies with more patents than paying customers. But Nortel (NT) has never had this problem. Nortel salespeople have always been the sort to tell you they can meet any of your problems. As a result, they rarely have holes in the product line, whether you’re looking at data center switches or wireless base stations, they always have an answer. The problem with this approach is that it has left them with more product lines than just about any other equipment supplier, but very few category leaders.

Aim for #1 or #2 or Just the Top 10?

In spite of CEO Mike Zafirovski’s too little too late attempt to emphasize market position last year, Nortel has long been the antithesis of the company that wants all of its product lines to rank #1 or #2 in each category. And this is a big problem in networking, where market value has a habit of catching up to market share, regardless of current P/E or price/sales ratios. Even in this terrible market, Juniper (JNPR) has a higher market cap now than me-too data networking vendors like Cabletron, 3Com (COMS), and Newbridge did during the industry’s bubble in 1999.

Since Voltaire (VOLT), Mellanox (MLNX), and QLogic (QLGC) announced QDR InfiniBand products last year, many industry observers have been wondering if 40G InfiniBand will leap ahead of 40 Gigabit Ethernet, especially with the IEEE still figuring out what its standard should look like. Yet OC-768 Packet-over-SONET ports have been available on routers for nearly seven years now, and they are hardly a dominant technology in spite of having been first to market above 10G.

What's been impressive about QDR InfiniBand is not the fact that it's here, but its cost, under $500 a port, or less than a 10GBASE-SR transceiver module. OC-768 data ports still cost the same as they did seven years ago, about $600,000, and will need a lot than the 100+ ports AT&T (T) has purchased for its MPLS core in order to come down in price. Moreover, OC-192 Packet-over-SONET ports still go for well over $100,000, even with intermediate range 1310nm optics.

It is hard to discuss 40 or 100 gigabit without someone asking about video applications. Yet it is even harder to find a planned 40/100 deployment being designed to carry video. In theory, it might make sense that video would push large corporations and carriers to upgrade past 10G, but in practice we're finding that:

• As a multicast service, IPTV is using far less bandwidth than a unicast service like Video-on-Demand would
• At $600,000 a port, a 40G router interface requires 6,000 months of revenue from a typical triple play customer to cover its cost, but just 6 months of revenue from a large corporate account
• Carriers are building 40G MPLS nodes in markets where they offer IP/VPN, but not IPTV
• Moving image applications are being demonstrated for 100 gigabit, but those images come from Computer Assisted Engineering models owned by research labs, not film clips owned by Hollywood studios
• Financial traders want to upgrade past 10G to cut latency even further, but with the advent of FAST (FIX Adapted for Streaming) compression, a typical trade will be about the size of an SMS message, so a leading commercial application for 40 and 100G will sub-kilobyte text files, not multi-gigabyte video files

The Technology of the Future

After years of speculation about a possible IPO or acquisition, Force10 Networks finally made a move this week, announcing a planned merger with Turin Networks. Force10 has historically focused on the 10 Gigabit Ethernet market, and would have a hard time surviving the telecom bust had it chased after the Gigabit market dominated by Cisco (CSCO), Extreme (EXTR), and Foundry (now part of Brocade (BRCD)). But with a new IEEE standard coming out next year promising even higher line rates, many are now wondering if a startup will be able to grab share from the traditional switch vendors by focusing on 40 and 100 Gigabit Ethernet. However, this is unlikely to happen.

New Technologies Used to Create New Vendors