What's Happening in Taiwan?
August 23, 2004...Continued...

One of the primary things that's now happening, is that the Taiwanese blue spectrum LED players are in a definite move up the value chain in terms of quality.

In 2002, basically all the Taiwan manufacturers were regarded, by themselves and others, as small players in the GaN-based blue spectrum LED production business, whether it be epitaxial wafers, LED die production, and/or packaging of the die. Less than 10% of the world’s entire number of HB-LED chips came from Taiwan in 2002. Their ramp of MOCVD GaN platforms continued to shock even those selling MOCVD equipment to them, which was primarily Aixtron, and Emcore before Emcore's TurboDisc platform line was sold to Veeco. It was almost like a contest between the two MOCVD equipment leaders to see which could load up the island faster. Veeco's VP of TurboDisc Sales and Marketing, Tom Miehe, once joked about there being so many reactors moving onto the island that it was likely to sink. His quip remains one of the great insider chuckles.

By 2003, that massive reactor buildup started to kick in and Taiwan came on-line in a big way. All of a sudden, 40% of word’s chip count (but not 40% of the world's revenue) was concentrated in blue LED-based cellphone keypads, which required brightness levels of around 3-4 mW @ 465 nm. Constant arm-wrestling for marketshare in the apparently lucrative blue keypad business was the highlight of the year. Cellphone manufacturers were cropping up faster than new LED manufacturing company names. The worldwide consumer appetites for colorful fun phones became insatiable. Some systems integrators required top of the line quality and the highest brightness components, with price being their second concern. Other's didn't seem to care about quality and brightness levels that much, with price being their only issue. Throwaway cellphones, cellphones with constantly added features... high end, low end... there seemed, and still seems, to be no end to the various flavors and hues.

Now, in 2004, the big news on the quality side of the business is that many more players than expected have come on-line with much higher brightness chips. Epistar was the first. But now at least 5 to 6 players have these similar, higher brightness chips available, reporting brightness levels of around 6-7 mW @ 460 nm. These brighter LEDs are targeted at the higher-end white LED backlight markets. Those who tour the plants routinely and see it for themselves, have reported that some Taiwanese manufacturers even have 8-9 mW available in the lab already, and by reaching those brightness levels, they're now in a position to pursue the same markets currently dominated by Cree’s MegaBright and even XBright devices.

The conclusion of those in the field is that the Taiwan manufacturers, as a body, have indeed now achieved the quality levels required to turn their blue LED lines into true white. (Non technical readers new to the field, may appreciate being reminded that there really aren't any "white" LEDs... the color white is a fooling of the human eye. White LED are produced by either taking a blue LED and adding some kind of coating to make them appear white, or they're a combination of red (R), green (G), and blue (B)... the latter two being part of the blue spectrum, and when you take a magnifying glass to your TV or computer screen, discover that R + G + B = white.

So... What's Happening in Taiwan regarding the white LED business? Is consolidation taking place? Or... are licensing agreements by The Big 5 starting to kick in? It's textbook tech-growth standard practice at this stage of product development for a combination of consolidation and licensing. The evolution has been proved countless times in silicon land and it always ends up some form of what many simply call Second (2nd) Sourcing. Osram Opto was the first of The Big 5 to make its play in Taiwan. (Ref our Oct. 19, 2003 coverage Osram Welcomes White LED Licensees) which began with a license to Everlight. And now we have the recent Nichia deal with Opto Tech ((ref. our Aug. 5, 2004 coverage, : Nichia Teams with Opto Tech and First Wave Taiwan Allies on Blue Spectrum LEDs)) which obviously involved Highlink. Highlink provides a key element in the deal since the GaN expertise involved largely lies with Highlink's Chuong Tran and his team of outstanding MOCVD-based blue spectrum LED growers. Chuong was an important part of helping develop the original GaNzilla TurboDisc MOCVD platform when he was with the Emcore Apps Lab.

Never Say Never, as the old saying goes. Initially, Nichia said it would never license. That's another textbook tech-growth standard practice. World dominance is always a fun philosophy to kick-start any field, especially when your company is first with the brightest product. Nichia's team of gifted researchers, which was originally headed by Shuji Nakamura, did not "invent" the blue LED (Cree had blue SiC LEDs in the field years before) but Nichia certainly broke through the true high brightness barrier layering GaN on sapphire and had the IP to prove it. ATMI (who's IP is now owned by Cree) and Osram and TG also had strong positions in early stage GaN-based white LEDs. But originally, Nichia appeared to have a lock on blue spectrum LED intellectual property (IP). Then when Osram, Cree, Toyoda Gosei (TG) and Lumileds all began to play their white LED IP cards at what amounted to a virtual Big 5 poker game, Nichia ended up cross-licensing, in various manners, with the other four. Thus, The Big 5 came into being.

With the advent of first Osram, and now Nichia reaching out to the leading Taiwan manufacturers (all of whom have obviously worried more than just a little bit about IP issues during their MOCVD buildup days) the IP wall originally created by The Big 5 appears to be crumbling a bit.

That brings us to the present and the world of speculation, so here's our take on some interesting interpretations that have come across the screens regarding what is currently going on:

Take 1: Osram’s yellow phosphor licenses for creating white LEDs directly targets the business Nichia cares most about. Nichia is the world’s undisputed leader in terms of standard-size packaged white LEDs . Osram's licensing to second tier players proved that by initiating that practice, Osram isn't hurt in the markets where Osram plays best (automotive) but simply generates additional revenue via licensing. Osram's initiative in reaching out with their IP licensing program does challenge Cree and TG, who are currently the major competitive blue LED chip (die) providers. Why? Observers close to the situation say that Osram's reaching out automatically opens up more markets for their customers’ competitors but doesn't touch Lumileds current dominance on the high power market and Osram's current dominance of the automotive market... much, since the Taiwanese don’t really play in those markets... yet.

Take 2: Nichia’s deal with Opto Tech (aka "Optotech") is similar to Osram's with their licensees , plus sources say that the relationship is not new, that Optotech has supplied Nichia with Red, Orange and Yellow (ROY) HB-LEDs in the past. And with the new agreements, Nichia essentially gives nothing up. The agreements let Optotech/Highlink (and others on the Optotech team yet to be publicly disclosed) to play in the markets the Taiwanese already dominate, namely blue LEDs for keypads in Asia. And ... it gives Nichia an immediate second source for providing LEDs to a growing number of hungry customers looking for blue product. Nichia can can now take advantage of Optotech blue chips, which are priced considerably lower than Nichia's own products, and have them packaged by Citizen (with whom Nichia has licensing agreements) for sale in Japan.

Take 3: That would mean that Nichia gets an inexpensive source of LED chips, and licensing revenue to boot... without the overhead of having to expend capital to expand. In doing so, they help a “friendly Taiwanese” manufacturer (Optotech) to compete with the Taiwanese they would like to maintain dominance over, such as Epistar. Nichia also gains a degree of control over the Taiwanese. Why? Because the agreement limits the maximum chip capacity of Optotech’s factory. Optotech, which our sources assess has shown disappointing results over the last 2 years, gets a new sales pitch and hopefully increased revenue. Optotech's mantra could become “We are the IP safe Taiwanese – Nichia says so”. That carries a great deal of weight in these IP saber rattling times. And it also marks, FOR THE FIRST TIME, a Taiwanese blue LED chip provider getting access to the Japanese market... under Nichia’s control, of course. This interpretation would whack another huge crack in the IP wall! And who gets hurt? Well... the providers of blue LED chips to Japan amongst The Big 5: particularly Cree and TG.

Whether you fully agree with the above scenarios, or not (and remember, these are collected interpretations. I'm simply the 'clairvoyant' industry compiler and public messenger/historian). Hopefully those of you reading this have more to add... or subtract (and we always appreciate response and dialog). Among the myriad of questions that logically follow are...

What will Cree, TG, and Lumileds now do in response? Which companies are hurt by the crumbling of the IP wall? Which companies are helped? Are we watching The Big 5 hold on the industry loosen? Are the Taiwanese being given an opportunity to see just how brightly they can shine? Are geographical boundaries changing and some of the second tier players in Asia providing gate-keys to their nationalistic minded kingdoms?

My Take: From what I can conclude from all this is that the Taiwanese manufacturers get helped at every turn. But, what actually results is that product moves, and the huge buildup in Taiwan of MOCVD platforms gets more fully utilized. And isn't that the overall goal of the SSL industry? To field product faster to a blue/white, LED hungry world? The environmental ramifications of substituting LEDs for conventional, energy-wasting conventional lighting are too important not to keep that as the SSL industry's prime directive. So I see both the environment and economics winning. I clearly see those sincerely wanting to get energy saving blue spectrum LEDs into the field faster... a feat that will require teamwork and unprecedented industry cooperation (dare we start using the "S" word... Standards? in addition to the "C" word... Consolidation?), as the winners in 2004 and beyond. That way, everyone wins.

So that brings us to the end of this chapter of the ever-evolving Solid State Lighting story. The anticipated and oft-predicted "overcapacity problem" that was said to kick in around Q-3 of this year is happening, right now, on schedule. Many of the Taiwanese suppliers have quietly reported that, although they didn't actually sink their island after all with all those massive MOCVD platforms, they indeed currently have unused capacity. Inventory difficulties and a bit of a slowdown in handset production are indeed causing the market to flatten at precisely the time when even more capacity is continuing to be brought on-line, which means the overcapacity "problem" could worsen. But... it doesn't have to, in my opinion. Not if The Big 5 continue to move in the direction of teamwork and true 2nd sourcing, through solid, amicable licensing agreements. That way a worsening problem becomes a giant opportunity to get the job done, better, faster, with more and better blue spectrum LEDs fielded in the directions the industrial and consumer applications appetite appears to either need or desire.

Densen Cao
CAO Group, Inc.
8683 South 700 West
Sandy, UT 84070
801-256-9282 (tel)
801-256-9287 (fax)
www.caogroup.com (web)