VCSEL start-ups face tough times
More than two-thirds of firms making VCSELs will ultimately be forced out of the market, according to an industry expert. Jim Tatum, a senior manager in Honeywell s VCSEL division, believes that too many start-ups entered the market at the peak of the optical-communications boom and that a period of consolidation is now inevitable.
"At my count, the number of firms that sell VCSELs is between 30 and 35 and the market is just not big enough to support this many," said Tatum. "A lot of these companies are talking about offering long-wavelength components. That s a space that s very full and has an embedded technology that s going to be difficult to displace."
The two strategies that start-ups usually employ to win business are difficult to play in the VCSEL industry, says Tatum. "One is to say that you re a technology leader, which is a difficult argument in the VCSEL market against the likes of Honeywell and Agilent. The other one is to say that you re going to be a cost leader and the cost game is a hard one for many start-ups to handle."
As a result, Tatum forecasts that a wave of forthcoming consolidation and attrition will leave just a handful of VCSEL suppliers. "At a minimum there will be two, because the world essentially requires second sources, and I think 10 is probably too many."
Last in, first out
Such stark predictions cast a shadow of uncertainty over the future of new VCSEL-market entrants in the US, Asia and Europe. In Asia there has been an explosion in the number of companies that offer VCSEL wafers and packaged devices over the past couple of years. The growing list includes PROWTech, Optoway, Vichel and Optowell in Korea and GigaComm, TrueLight and VTERA Technology in Taiwan.
In Europe the latest newcomers include Ulm Photonics, a spin-off from the University of Ulm in Germany and Avalon Photonics, a spin-off from the Swiss Center for Electronics and MicroTechnology in Zurich. While in the US, start-ups making VCSELs and related products include Cielo Communications, PicoLight and Bandwidth9.
All of these start-ups face fierce competition from the big players, such as Honeywell, Agilent, Infineon and Emcore. Their sheer size, reputation and experience in volume manufacturing put the big firms in a strong position to survive any market shake-out and emerge as the dominant suppliers, perhaps acquiring some of the start-ups on the way.
Today the dominant market for VCSELs is short-reach data communications. Devices operating at a wavelength of 850 nm have made inroads into Gigabit Ethernet and Fibre Channel applications, replacing edge-emitting lasers at 780 nm, which have inferior reliability and efficiency.
"Probably 80 to 90% of the VCSELs in use are in these applications," commented Tatum. "Honeywell has shipped upwards of 20 million components into this space, so there s a pretty significant track record there. I struggle to think of anyone making short wavelength links that is using conventional edge-emitting lasers."
Another emerging market is short-reach parallel optical interconnects, where an array of VCSELs - typically 1X4 and 1X12 - is connected to an optical fiber ribbon to create a high-speed link that can transfer tens of gigabits per second between routers and optical networking hardware (see FiberSystems International, February p29).
In contrast, the market for devices operating at 1310 and 1550 nm is likely to prove a much harder one for VCSELs to win over. Although long-wavelength VCSELs are just starting to arrive on the scene, the technology is seen as immature and there are concerns over its long-term reliability and limited output power. One of the first products on the market is Cielo Communication s 1310 nm parallel array module that transmits up to eight 2.5 Gbit/s channels.
"I think that there s still some uncertainty in the industry. Clearly there are VCSELs that have demonstrated reasonable performance, but I would argue that today they are inferior to edge-emitting technology," said Tatum. "In the long-wavelength region the reliability [of edge-emitters] is pretty good and VCSELs have a lot to live up to before they get any penetration into that market."
"Adopters are going to be hesitant to accept the technology until there are massive amounts of reliability and field data," Tatum concluded.
