In 1995, ADVA expanded into optical multiplexing with a 4-channel WDM for enterprise mainframes, and Inrange, one of the largest of ADVA's current partnerships, became one of its first customers.

In 1997, ADVA introduced an 8-channel WDM for service providers and recognised that the enterprise market, though a lot smaller than long-haul, required the same type of bandwidth drivers. 1998 saw the launch of a 16-channel DWDM, and key milestones were reached with ADVA's one hundredth customer and 11.4 million Euros in sales.

In 1999, the company launched an IPO on the Neuer Markt (the German equivalent of the Nasdaq). To date, no plans to list on the Nasdaq itself have been announced.

Also in that year, ADVA introduced the FSP-I and FSP-II generations of its Fibre Service Platform. These two products remain the key elements of ADVA's product strategy, and accounted for over 80% of the 60 million Euros of sales in year 2000. Europe deployed 55% of FSP sales, the US 40% and Asia-Pacific 5%.

In 2000, ADVA acquired First Fibre, whose CityLight product has been re-branded by ADVA as the FSP-500. This is a lower cost product than the FSP-I and is used by BT, for example, as a low cost non-DWDM platform.

Other acquisitions have been SCSI, Fibre Channel and storage over IP company, SAN Ltd, of Cambridge, UK; Cellware Broadband, a video and ATM technology company based in Berlin, Germany; and a Siemens ATM and software development team in Oslo, Norway. In addition, ADVA has invested $1 million seed funding in OptXCon, an optical switch company of RTP, North Carolina.

At the end of 1999, ADVA had a headcount of 132. This has grown to 450 today, of which 150 are in engineering and 100 are in procurement or production. ADVA has two production facilities, one in Germany employing about 150 staff, and the former First Fibre facility in York, UK, with about 50 people.

Brian McCann pointed out that in the year 2000, ADVA shipped over 60 million Euros (approximately $52 million) worth of kit to partners and end customers, which he feels ranks ADVA as "being a major player in the optical networking space". Over 10,000 WDM channels have been installed, and the company, he said, has successfully "leveraged success in the enterprise and moved into the metro access, and now the metro core markets."

In the past 12 months, ADVA has evolved from being a one or two product company to providing a "total portfolio of solutions to address all metro and enterprise service applications." Despite the lack of visibility of the ADVA brand, particularly in the US, Mr McCann asserted that ADVA is a significant player. Europe is a strong market, with ADVA having a dominant share in the German-speaking and UK and Nordic regions.

In fact, ADVA employs a three-prong distribution channel of OEMs, direct sales, and VARs and distributors. Until 2000, ADVA was exclusively OEM, all solutions carrying another badge. In 2000, ADVA altered its strategy by additionally using VARs like Inrange, Hitachi Data Systems and StorageTek, with whom it has co-label agreements. As a result, the direct presence is becoming stronger. In 2001, more than 50% of business will be with ADVA-label or co-label products.

Nevertheless, ADVA regards the OEM strategy as more than worthwhile to both itself and partners as it "plugs a hole in our OEMs optical networking product line", and the OEMs, in turn, "drive great revenue".

In terms of direct sales, ADVA cites BT, purchaser of the FSP-500 and the FSP-II, though service and support are outsourced. BT is described as not just a customer but also a "true partner…we will actually bring enterprise customers to them."

ADVA's product portfolio is designed for optical transport and service provisioning in enterprise and metropolitan area applications. Since ADVA's initial success was within the enterprise, its products have been designed from the enterprise upward, unlike other players in the metro DWDM market.

ADVA's products are therefore smaller, work on AC power, and support GE (850/1310nm), Fibre Channel, ESCON, ATM and other technologies. ADVA products support applications from T1 (1.5 Mbit/s) up to 10 Gbit/s, and from non-DWDM, to coarse DWDM and DWDM.

ADVA does not provide an all-embracing product. "We focus on transport but at relatively high bit rates. Our sweetspot is basically 100 Mbit/s and up, though we have products addressing the lower speeds (FSP500) that will also do 155 Mbit/s, 622 Mbit/s, GE and Fibre Channel. Our other platforms continue to move up." The FSP-1000, for example, is "all about optical feeders". Signals at 155/200 Mbit/s or even gigabit signals can be taken and aggregated up to a 2.5 Gbit/s wavelength to "ride" the optical network.

ADVA has developed its platforms to interoperate. The FSP-3000/2000/1000/500 are the four main platforms, each with its own specific advantages across a common network management platform, and thus enabling the entire optical network to be managed from one central station.

FSP-I
The FSP-I and FSP-II products are enterprise-located solutions. FSP-I is a fibre-conversion platform, taking 850nm up to 1310nm, 8-channels point-to-point in the chassis, offering four coarse DWDM channels and is service transparent between 10 Mbit/s and 1.25 Gbit/s. It also has protection with a switch over of 50ms and is described by ADVA as "very easy to use - just a plug 'n play."

One of ADVA's first customers for FSP-I, and still one of the biggest, is COLT Telecom. Mr. McCann points out that "COLT is not an incumbent, so when they install fibre they install plenty of it. They don't need DWDM up front." With the FSP-I, McCann explained, COLT customers are able to start with 100 Mbit/s and can be remotely upgraded to Gigabit Ethernet with no change in hardware.

"That is a compelling business case. COLT can charge $2000 per month for a 100 Mbit/s service and $4000 per month for Gigabit Ethernet."

The customer can therefore obtain four times the bandwidth for just twice the price, and COLT's margin on the upgrade is almost 100%. The system cost is $7000 per channel, both sides.

FSP-II
The FSP-II targets applications where there is less fibre available and can be deployed in access rings as well as point-to-point. FSP-II is a 32-channel system with full line protection, service transparency from 10 Mbit/s up to 2.5 Gbit/s, has OADM capability and 4:1 TDM built in.

"I can actually for example get 128 ESCON channels from a fibre pair", McCann explained, "which was unprecedented for a 1999-deliverable product."

Verizon and BT, with its MetroWave service, are two customers employing the FSP-II. The product is sold by Cisco as the Metro 1500.

"FSP-II has been a strong driver for us and really placed us as a leading provider in the DWDM metro access space. A large company could, for example, buy this from one of our partners and we would provide 24x7 maintenance and support on it, or, if you wanted to outsource it, you could go to one of our carrier partners and they would turn it up, like a service".

FSP-3000
The FSP-3000 was designed to advance ADVA from being a two-product company to that of being a total solutions supplier. The FSP-3000 is an optical ring platform for both access and core metro deployment.

According to McCann, ADVA's goal is to convert T1s and E1s to optical wavelengths. ADVA is certainly able to achieve this from a price point of view, and, ultimately, aims to enable carriers to "deploy wavelengths as easily as they are deploying T1s today."

Vendors such as Lucent and CIENA are not perceived as competitors by ADVA because its own products are much more scalable, much lower cost, smaller footprint and more flexible than for example CIENA's offering in the same space. ONI is seen as much less of a threat in Europe. However, McCann explained that technology, for example SONET or SDH, represented the real competition.

Capital expenditure cutbacks, ADVA maintains, are not affecting optical networking products nearly as much as SONET/SDH or IP products, simply because the former can save money.

"By deploying an optical platform instead of a SONET or SDH platform you have a lot more flexibility. It can be any speed or feed."

McCann noted that optical networking is scalable, whereas non-optical upgrades from OC-48 to OC-192 are "extraordinarily expensive" and take a long time. Optical networking has a much faster impact and is much quicker to implement. CIENA and ONI are successful because optical networking has greater fundamental value.

The new platform does support 10 Gbit/s and optical amplifiers, as well as matching the same optical grid as the FSP-3000, enabling the latter's EDFA's and network management to be utilised. The FSP-2000 has both point-to-point and ring functionality.

ADVA points out that Cisco will not be a partner on this product as it competes with Cisco's own 1540, which is different architecturally.

Cost per service on the new platform, using the TDM/WDM combination, could be as low as $3000 per application port or $6000 both sides. Similar functionality in the FSP-II is about double that figure. This driving down of costs, ADVA believes, will enable the vision of T1s becoming wavelength services a reality.

ADVA concedes that still to come are optical cross-connects, greater distances and greater numbers of channels, but nevertheless argues that ADVA is not late with these features because, at present, the cost tradeoff means that the metro market is not yet able to afford them. "Most of the metro market is not yet filling 32 wavelengths; we don't need 128 channels, so we feel no need to fight the channel wars."

In fact, the cost tradeoff is mainly in the optics. McCann explained that, at the moment ADVA uses 200 GHz spacing. To go to a higher number of channels, 100 GHz spacing would need to be designed into the system, even if the extra channels were not lit, and that would require more expensive parts. "The lasers have to be more controlled, the WDM filters have to be tighter and flatter, and they carry a premium that has to be implemented now."

The task of both maintaining coverage of what could over the next few years be a flashfire ignition of global enterprise demand that will be 100 times larger than it is today, whilst also expanding its strategic evolution vertically (in a move which will also multiply ADVA's market size by another order of magnitude at least) may impose demands for cash, management and technical resources which may be impossible to satisfy. This would present the company with a rapid set of difficult sacrificial options, and there is a risk of getting the mix of such decisions wrong. That is particularly likely to be true if, as seems likely, the growth of this demand is not at all smooth.

Given the present lull in worldwide demand, ADVA management may justifiably regard the possibility of being overwhelmed by demand as pretty fanciful, but good times will come again, with the biggest impact likely to be in the sector they serve. When it does come it is likely to be all the more explosive for the intervening lull, since design cycles and technical maturity tend to develop independently of commercial volumes.

A more obvious problem is that whilst the company's existing customer base appears to be well looked after and loyal, and in that sense the company's customer service may be excellent, its brand recognition in the general market outside that customer base remains not only weak but also indistinct, precisely because it is such a differentiated company, particularly in the US.

Of course, in step-by-step sales evolution across its potential enterprise customer base, this may not matter since the company presumably has large numbers of reference customers who can affirm its competence and quality at each step. Also of course it is pointless and potentially wasteful to have a brand image, which is far above a company or product's needs.

Nonetheless this could be a problem in terms of larger moves such as lateral marketing or vertical product acquisitions, and the wholesale marketing of a larger product range the company might want to make.

Yet another problem is that the company's current business is, by its nature, very labour intensive, with quite a low average-value per sale. ADVA's 350-customer base (not all of whom of course might have been customers in that year) gave it sales of $52 million in 2000. This contrasts with a few tens of customers served by CIENA, whose sales level is 20 times larger, and similarly for ONI Systems, whose sales are rather larger than ADVA's but which has relatively few customers, and whose orders are normally quoted as multi-million dollar ones.

There are at least three obvious problems about a labour intensive business: firstly, particularly when the labour needed is high quality engineering, the business may have difficulty in growing above a certain size in any particular location. In any case this is bound to be true where the engineers are required for long periods on the enterprise campus. Secondly the business cannot achieve the same economies of scale with volume that is achievable by more capital-intensive businesses, though of course this can also offer a small advantage in a downturn, where the reverse tends to be true. Lastly it is hard for a labour intensive business to expand rapidly in the face of fast growth in demand.

The final problem that ADVA has, and which it has itself identified, is that, in principle, it is not a particularly technologically oriented company; ADVA is generally a user of technology, not a generator of it, and for several reasons is relatively conservative technically. The company's real forte is its ability to generate customer-specific solutions and to back that up with good service. Such a company could have a problem if technology evolved too fast or if shifts in technology threatened its customer service capability.