Fragmentation depends on your point of view

A recent poster on the Photonics LinkedIn group was impressed by the stratification and fragmentation in the HB-LED business. The self-described newcomer can be excused for being naïve, but there are some things I think are worth repeating here. Most importantly, whether fragmentation is good or bad depends on where you sit.

A lot of people like to compare photonics with some industry X just before the industry took off. The most common example is silicon electronics in some early stage. Who wouldn’t want to repeat that ride?

But that’s a poor analogy. Photonics components are more like airplane parts than CMOS. Most photonics parts are highly specialized, and the suppliers can often make a nice profit because of the complexity or service they provide. That's good for a lot of companies, including small and medium size companies. For them, fragmentation is good. It means niche opportunities for them.

However, it turns out that a lot of photonics parts require an expensive clean room—a fab. Anyone who has an expensive capital investment, like a fab, wants to get volume through it to pay for the lights, and that favors consolidation. These suppliers want everyone else to get out of the business and leave it to them. Who wouldn’t? So, many of these companies look to minimize differences in products, even if it means standardizing the parts a little and giving up some of the profit margin. For them, fragmentation is bad because it limits their volume, and therefore their profitability.

If you are a customer or an end-user, there is occasionally a segment where consolidation is needed to lower the price to the customer, and move the market forward. But these opportunities are rare. When they appear, the customers usually have a way of forcing standardization onto the suppliers, not the other way around. Solid-state lighting might be an example where consolidation might help the end-user. (Or not, since it's not the only factor in its adoption.)

But watch out what you wish for. What's good for the customer, or for one supplier, may not be good for everyone. Whether the consolidation is good or bad depends on whether you survive the shake-out or not.

In summary:
* Photonics markets are notoriously fragmented.
* They always will be.
* That’s good for some companies.
* That’s bad for some companies.
* Consolidating suppliers can sometimes help grow the industry, but watch out what you wish for.

3 countries make 87% of all lasers. Wow!

Bet you didn't know this: about 87% of all laser revenues attribute to companies headquartered in only three countries: the U.S., Japan, and Germany. Wow! Who'd have guessed?

Don't believe me? Consider that about 1/2 of all laser revenues are for diode lasers for communications and optical storage. These are mostly made by Japanese companies, some U.S., some Taiwan, and a few others.

Then, consider that several big laser makers hail from Germany and Japan: TRUMPF, Rofin-Sinar, FANUC, Gigaphoton, and Mitsubishi. Germany is also home to many smaller laser makers, like Jenoptik, Toptica, and (despite the name) Menlo Systems. The U.S. is the official home to many familiar names: Coherent, GSI (including Synrad, Quantronix, and Continuum), Newport, IPG, Cymer, JDS Uniphase, Oclaro, and many, many smaller companies.


I'm counting revenues here, not units. A lot of commodity lasers are made in Taiwan, or even China, for laser pointers and such things. Oh, and of course, I'm talking just the laser, not the system or end-use.

True, the assembly may be anywhere from Russia to China, but the companies are headquartered in only a few countries. This means that at least a large part of the revenues (including the profits) flow back through headquarters. (More on that in a later post.) Oh, and of course I'm talking just the laser, not the system or end-use.

For the record, this all came about from a question I got from Breck Hitz, the Executive Director of the Lasers and Electro-Optics Manufacturers Association, LEOMA, who is trying to advance laser standards at the ISO.