Optical components

JDSU Posts Loss, Buys Startup

After posting sizeable losses on dwindling revenues, JDS Uniphase Inc. (Nasdaq: JDSU; Toronto: JDU) announced last night that it plans to purchase startup Scion Photonics Inc. for $43 million within the next few days (see JDSU Posts Q3, Buys Scion).

JDSU has held a minority stake in Scion for over a year, and execs say the startup offers key manufacturing advantages that will slash component costs and give JDSU a leg up when an upturn in the component market materializes.

"Everyone is very excited about the opportunity this brings," says Karen Markus, VP of JDSU's waveguide business unit, which will oversee Scion's San Jose, Calif., operation of 75-odd employees. "Scion gives us a foundation for creating a platform for waveguide integration."

This is really an acknowledgement that JDSU has so far failed to gain traction in the market for Arrayed Waveguide Gratings (AWGs) -- seen as key building blocks for developing a wide range of integrated optical products (see Photonic Integrated Circuits).
Unfortunately, JDSU can't say when demand for such products will materialize in a meaningful way. Indeed, analysts were chilled by JDSU's guidance last night, which calls for revenue reductions of up to 26 percent next quarter "as the downturn continues."

"June may not be the bottom; the low point has not been reached yet," JDSU CFO Anthony Muller told analysts last night.

JDSU's revenue shortfall was no surprise (see Next Sore Spot: JDSU?). Losses are mounting. This quarter, JDSU posted a GAAP loss of $4.3 billion ($3.19 per share) on revenues of $262 million. Last quarter saw a GAAP loss of $2.1 billion ($1.60 per share) on revenues of $286 million (see JDSU Reports Q2 Losses ).

On the upside, though, JDSU has a solid cash position. It's got $1.5 billion in "cash, money market and other highly liquid securities and over $100 million in marketable equity securities," according to last night's press release.

The company also has confidence, despite its guidance. JDSU CEO Jozef Straus told analysts last night he thinks North American carriers are experiencing saturation on key intercity connections that will compel them to seek new equipment within six months. And when equipment makers come calling, they'll be needing just the kind of integrated components -- those that combine multiple functions on a single chip -- that Scion can offer.

Indeed, JDSU says Scion's technical advances, which translate into cheaper manufacturing, are adding to the startup's customer list significantly -- which will be boosted even more when JDSU adds its clout to the pitch. On last night's call, CFO Muller told analysts that Scion's prospects are so good that JDSU expects to recoup the money it's spending on the startup "within two years."

Even before JDSU made its bid, Scion made headlines on its own. In an interview last week, Scion COO Venkatesan Murali told Light Reading that his company's expertise in working with eight-inch silicon wafers will help make AWGs for less than $100 per channel -- half the current lowest price (see Scion Seeks to Slice Components Costs).

But Scion may not help JDSU's fundamental outlook immediately -- particularly as some say the advantages of its manufacturing techniques are best realized at high volume production. To cope with the ongoing distress, JDSU plans to cut expenses even more, in part by reducing its 10,000-employee headcount by about 2,000 and closing facilities worldwide.

One facility set to close is the plant in Columbus, Ohio, acquired when JDSU bought PIRI (Photonics Integration Research Inc.) in May 2000 -- a move that got JDSU into the AWG business in the first place.

JDSU says Scion's use of chemical vapor deposition (CVD) technology is more advanced than PIRI's techniques. The Ohio plant will close by September. JDSU says some of the 111 employees may be given an option to move to Scion headquarters in San Jose.

Some say this adds to the value Scion brings: "JDSU needed a solution to their planar strategy quickly," writes James Jungjohann and colleagues at CIBC World Markets in a note today. "Privately held Lightwave Microsystems Corp. was stealing market share from JDSU's PIRI unit and Scion was already getting strong feedback from OEM customers designing Scion's devices against Lightwave's."

Analysts delivered mixed reviews of last night's news. While most viewed the Scion acquisition as a good one, JDSU's outlook of ongoing reduced revenues sounded an off note.

"Like almost every telecom equipment manufacturer, JDSU has little visibility beyond a few months' sales," comment the CIBC analysts. "[W]e feel sales may continue to fluctuate... for some time, as customer demand varies and OEMs continue to pull from existing inventory levels."

Others voice similar reservations. "We think valuation of JDSU stock is becoming increasingly attractive but is not compelling enough to offset weak telecom fundamentals," write David A. Jackson and David Siniscalchi of Morgan Stanley Dean Witter & Co. in a note today.

The market apparently concurs: At press time, JDSU shares were trading at $4.62, down $0.41 (-8.15%).

— Mary Jander, Senior Editor, Light Reading
Touch 12/4/2012 | 10:30:10 PM
re: JDSU Posts Loss, Buys Startup I think that this is another term for "sol-gel" deposition. You can find lots of references on the web for this technique. I am a little familiar with CSD for growing crystalline materials (like GaN). Another term for it is "liquid phase epitaxy".

The trouble with sol-gel or "spin-on glass" for making waveguides, is that they can have LOTS of hydrocarbons in them (in addition to the SiO2). So you can get a lot of absorption in the near-IR. The way around this is to anneal out all the muck, hopefully leaving just SiO2.

I believe there is a start-up in Canada that is using sol-gel for waveguides. Interesting technique, but I do not believe that this process can yield nearly as well as FHD or PECVD.
dwdm2 12/4/2012 | 10:30:22 PM
re: JDSU Posts Loss, Buys Startup Touch, thanks for the descriptions. I know some are using CSD (chemical solution deposition) to grow waveguide on Si substrate. Can you possibly describe how this method work?

Touch 12/4/2012 | 10:30:24 PM
re: JDSU Posts Loss, Buys Startup PIRI does indeed use FHD (flame hydrolysis deposition) to deposit the core layer. FHD uses an oxy-hydrogen torch to blow glass soot onto the wafer. They also add a dopant (Ge from GeCl4 I think) to increase the refractive index The soot is then consolidated at high temperatures (1000C or so). They then use RIE to etch the waveguide pattern into the core layer.

PECVD (plasma-enhanced chemical vapor deposition) on the other hand uses a plasma chamber to react silane (SiH4) and N2O (typically) to form layers of SiO2. This process also uses a dopant to raise the refractive index of the core (in order to get waveguiding).

Based on my experience with both processes, PECVD has better thickness and refractive index uniformity than FHD.

To cover the waveguide cores, one needs to use a glass with a low melting point. For PECVD, most people use BPSG (boron-phosphorous doped silica glass). You can get good step coverage, but one may need to deposit the top cladding in several steps. FHD however can deposit top claddings in a single process step.

Hitachi uses a different technique to deposit their core layers. I believe that they use e-beam evaporation of a titanium-doped silica onto fused silica substrates. They then consolidate and etch with RIE.

LightwaveMicro uses PECVD. PIRI (and NTT) uses FHD. Hitachi uses e-beam evaporation.
dwdm2 12/4/2012 | 10:30:25 PM
re: JDSU Posts Loss, Buys Startup I thought they use CSD (not CVD) method while most others use FHD + RIE process to grow the waveguides. Can anyone shed more light on this?

Light_wdm 12/4/2012 | 10:30:57 PM
re: JDSU Posts Loss, Buys Startup "JDSU says Scion's use of chemical vapor deposition (CVD) technology is more advanced than PIRI's."

PIRI's tech is FHD which NTT also uses. JDSU should know this.

Sign In