Project Network: JDSU (cont.) The last!
JDSU’s Business Strategy
JDSU has laid out a simple four-part strategy with the fourth element left tantalizingly blank:
 assemble the most complete product portfolio
 move rapidly into subsystems and modules
 increase manufacturing capacity
 ?
Before the merger that created JDSU, JDS Fitel was a leading manufacturer of passive components, while Uniphase was a powerhouse in active components. The two had significant customer overlap with little product overlap—a characteristic that would shape JDSU’s future acquisitions as it built its product portfolio.
#1. Broadening the portfolio
A rundown of major products and expertise added in JDSU’s shopping spree:
SDLI—arrayed waveguides (AWGs) and Raman pump lasers. 200 domestic and foreign patents and 170 pending. 80 engineering PhDs.
Etek—2nd largest passive components manufacturer in the merchant market. Low- cost, highly flexible manufacturing process. Modules include optical amplifiers, configurable add/drops, transmitters and receivers for long haul and metro/catv markets; China market and factory
Ocli—thin film filters, critical in moving to higher channel counts. Ocli had 80% of market
Cronos—MEMs, critical enabling technology for NG core switching devices
Sifam—fused couplers, DWDM multiplexers and filters. Throughput on couplers highest in the market
Epitaxx—active components, optical traffic monitors; highly sensitive Avalanche Photo Diode (APD) receivers
#2. Move rapidly into subassemblies and modules
JDSU’s increasingly broad component portfolio has enabled it to develop modules and subassemblies with high degrees of functionality and to accommodate its customers severe time to market constraints. Modules or subassemblies are now offered in the four key network functions: transmitters, multiplexers, amplifiers, and switches.
#3. Increase manufacturing capacity
Making components is only just evolving from a cottage industry. Many parts are still made or assembled by hand. JDSU filled a third of the positions at the Zurich factory with Swiss watchmakers. Yields in active components average less than 20% for some products compared with the 90%+ yields in the traditional semiconductor market. And yields are not much better on the passive side which requires at present a people-intensive, time consuming process. JDSU has turned to veterans in the semiconductor business to boost yields—currently as low as 1% in some products. If glass were not so brittle, JDSU would use robots.
The need for productivity gains has caused JDSU to adjust its business model. It cannot quadruple employment; employment costs must grow less than sales. Production capacity is being realized through a three-pronged initiative: plant expansion, automation, and outsourcing. JDSU now has 23,000 employees and 26 factories. Floor space has gone from 1.2 million square feet at the end of 1999 to 3.6 million in June and should reach 5 million by the end of 2001 (excluding the SDLI factor). Capacity increases realized in the first quarter: wideband filters, 33%; lithium niobate modulators, 40%; isolators, 100%; 10Gbps Avalanche photodiobe receivers, 300%. Yet some components are still capacity constrained.
JDSU has some forty automation projects in development (fiber processing, component handling, optical assembly, measurement and testing, inspection, wafer handling) with actual gains from automation: work cell and WDWM, 60%; optical amplification and splicing, 134%; testing of WDM multiplexers and amplifiers, 1600% There is no equivalent of AMD in the optical market [not sure about this; Straus is difficult to understand], and JDSU is, as Jozef Straus stresses, “taking matters in its own hands (CSFB conference, 11/29/00). JDSU outsources to Celestica and is considering other non-mission-critical purchases but, unlike the semiconductor industry, there are few suppliers in the optics space (Tony Muller, DeutscheBanc conference, 11/13-16/00).
#4. ?
Although just speculation, this objective could build on the other three and target the development/production of OICs in concert with JDSU’s NG partners.
JDSU’s Business Strategy and Network Effects
II. Knowledge/intangible assets
JDSU’s ratio of intangible assets to fixed and financial assets is on the order of 9:1. Return on these assets produces an ROIC of about 60%. Acquisitions and productivity efforts have, until this quarter, been funded out of cash flow from operations and with JDSU’s appreciated stock.
 30 new products at the OFC show in March 2000
 numerous design awards, notably those for AWG DWDM modules
 library of test data from long history of product life-test and quality control certification, enabling it to qualify new products more quickly than its competitors
 focused, forward-looking, and forthright management with depth and breadth of experience
 scale—revenues about double those of all nearest competitors in merchant market combined
 broad patent portfolio—including industry’s only patent for protection against catastrophic failure (COMD)
 broadest product portfolio, positioning it to meet all the needs of its customers
III. Investment in intangible assets
 invests in people and has single-digit attrition rate, a central metric in an industry where top talent is scarce and coveted
 9% of revenues spent on internal r&d
 reorganized sales/marketing by customer instead of product
 automation/productivity initiatives
IV. Networks in which JDSU participates
Sycamore leads the Optical Domain Service Interconnect (ODSI), a coalition of 130 NG systems firms. The Optical Internetworking Forum (OIF) is the big guys’ vehicle. JDSU participates in both.
Functionally, JDSU is active in all the NG cluster markets—core, submarine, edge, and metro, which positions it to work with its systems customers on total solutions—where interoperability and total network management—are key. JDSU is not a niche player.
V. Interactivity of the networks
Increasing product complexity brings the need for greater integration and interoperability. Juniper’s Sindhu points out that “what we are selling is not a box—we are selling a machine that is an integral part of a network that is living and breathing and growing—and this makes the relationship [in the value chain] a long-term one….What makes relationships strong is whether or not a company is helping the end customers to actually solve their problems”—scalability, stability, increased revenues. jump123.com.
Systems manufacturers, particularly emerging NG systems vendors, generally want to purchase solutions, not parts. Modules and subsystems are self-standing packages designed to perform functions important to these vendors. JDSU, by moving up the food chain, not only sells modules and subassemblies at higher prices, capturing value added to customers, it meets the needs of those customers more comprehensively and further distances itself from competitors who cannot integrate discrete components (Epoch, 7/27/00).
VI. Compatibility within the network
Networks are transaction-based.
The optical components/modules business is characterized by coopetition. Companies are both competitors and customers. GLW, for example, competes with JDSU and SDLI in the market for lasers and filters, packaging their parts in amplifiers it sells to customers like Nortel. Conversely, JDSU’s customers also partner with its competitors: Sycamore, for example, has partner relationships with both GLW and JDSU. The crux of these interlocking relationships is that systems companies like Sycamore must, as chairman Desh Deshpande points out, “keep up with the technology advance made by the component makers while working with carriers to fill bandwidth needs and create revenue-generating services” (MarketWatch, 12/5/00). So long as JDSU continues to make those technological advances, the relationship with its customers will be symbiotic.
VII. Adoption potential
The need for increased capacity has caused the complexity and performance requirements of newly deployed fiberoptic networks to increase substantially while the product life cycle for these network systems have decreased. OEM system suppliers, under pressure from their customers (the carriers), cannot compete effectively by vertically integrating their own components and modules. The sheer number of components is also increasing.
Solution complexity and time to market constraints make it less cost effective for established systems companies to fill their component needs internally. At the same time, emerging NG systems providers typically have little or no internal component manufacturing capability and must rely on a robust supply chain to meet their needs.
The trend among OEMs is to reduce vertical integration at the component and module levels and to focus on overall system design and architecture. Rather than qualifying a different vendor for each component, module, or subassembly, OEM system suppliers are turning to fewer vendors and designing their products into the system solution.
VIII. Assessing growth
Throughout optical networking growth is demand-driven. Customers demand additional capacity and services from the service providers. In turn, the service providers demand cost effective solutions from the systems vendors, who then pressure the subsystem and components suppliers to produce the building blocks needed for those solutions.
Incremental costs:
Until this quarter, JDSU funded expansion out of operating revenues and maintained high ROIC, indicating that marginal returns on invested dollars are good. [I could not disentangle the financials sufficiently to get a handle on fixed/incremental cost relationship.]
Critical mass:
JDSU’s tipping point came with the first merger between JDS Fitel/Uniphase. With a strong position in both active and passive components, JDSU could pursue a module strategy. With subsequent acquisitions, broadening even further its product portfolio, it is rapidly approaching “escape velocity”—the point at which competitors can no longer catch up. Although a lot of venture capital money has gone to optical startups and competition is keen, JDSU operates in a large-scale commercial industry. As Tony Muller likes to point out, “It is not a business of science fiction.” Companies must execute in large volume and not offer just a single product/component with a low run rate. Execution is as important as the technology (WitSoundView conference, 11/14/00). Optical Micro-Machines, one of more than 60 MEMs startups, has only 300 employees (but up from 13), a single product that is not yet in prototype testing and no product qualified or shipping. It will face an uphill battle against JDSU’s MEMs.
Natural monopoly:
Juniper’s Sindhu believes that no single player will emerge to dominate the router market because it is simply too important. But it will be hard for more than two companies to get traction in the space because, given time to market constraints, service providers will not want to waste time evaluating or qualifying vendors #3, #4, or #5. The same holds true for modules and subassemblies 123jump.com.
Second-sourcing is, in fact, common in the optical components industry; but systems vendors remain reluctant about small unproven vendors, given the complexity of integrating components. By maintaining robust but not exorbitant margins through careful pricing, JDSU manages to pre-empt the emergence of some competitors, while also exploiting the high price elasticity of demand in the optical networking marketplace. Many of the optical component vendors have higher gross margins than their customers, an unusual price structure that reflects the degree of value added and the height of the entry barriers at the component level (Epoch, 7/27/00). Everyone playing in the optical space runs the risk that someone will build a better mousetrap or standardize around someone else’s product, but component makers sell to everyone and the diversification in product lines insulates them somewhat more than the systems vendors (http://www.fool.com/Server/FoolPrint.asp?File=news/2000/jdsu001215.htm)
IX. Threats to network effects
Disruptive technology. On the optics side as well as production. JDSU could get designed out, as well as designed in.
Industry consolidation. Had the NT/GLW deal gone through, it would have created an optical juggernaut with annual sales of more than $3b, twice the size of JDSU (Bloomberg, 11/27/00).
Increased competition: Not only may LU and Nortel spin-off their “captive” optics components divisions, but Avanex is developing a suite of modules. Bookham and Newport are also adapting semiconductor processes and automation, demonstrating that the optical components manufacturing can scale. (http://www.fool.com/Server/FoolPrint.asp?File=/news/2000/jdsu001215htm
Financing: Increasing vendor financing. Alcatel, for example, invested $1billion in 360networks (IBD 11/30/00). Cisco’s trebling of its loan loss reserve isn’t a big deal, but Ciena increased its provision from last year’s $250,000 to $19.2m for the current quarter, which is. Customers’ customers are also debt laden: Verizons went from $17b to $47b between 1996 and 1997, GBLX $1.4b in 1998 to $8b in 1999. (http://www.thestreet.com/telecom/1163145.html and economist.com
Commoditization: Current demand enables margins to be maintained by increased productivity and capacity. Were the supply/demand equation to change, margins would be squeezed and at the same time networks effects in the system would loosen.
Gorilla/king/prince?/pauper
JDSU is growing twice as fast as counterparts in the PC revolution 12 years ago. So the question of kingship or gorillahood naturally gets batted around. But the NG network is in its infancy and the question is premature. Building out the NG network will be a massive undertaking and along the way we can expect some rough terrain, marked by periodic bumps, shakeouts, creative destruction, and consolidation. If Paul Johnson is right, however, and the effort takes $1trillion over twenty years, the CAP is extraordinary.
GAP is where answers to the question get sticky. In this early stage we have proprietary technologies (and fierce competition for dominance among them), but no open architecture; high barriers to entry, but “equivocal” switching costs. To the extent that JDSU’s modules and sub-assemblies get imbedded into its customers’ systems solutions, switching costs increase. If it develops mission-critical OICs (optical integrated circuits) and can mass produce them, it will have lock-in.
End-to-end solutions require not just speed, reliability, and low costs at the core, but integration at the edge and in metro. These solutions will need another order of magnitude in software to manage traffic. Jozef Straus put the NG buildout squarely at the start of the TALC on the October earnings cc: “The metro market is starting to move. Juniper and Sycamore are creating “destructive” opportunities. JDSU is being designed into those solutions. Expect 40-gig end of next year and it is being moved forward because of customer demand. This is an inflection point in metro and we are getting traction.”
JDSU has achieved its growth so far through almost flawless execution, preemptive strikes at the competition, leveraging design wins and its portfolio. But it’s too early in the game to offer it either a crown or a banana. |
