We like to identify macro technology trends and disruptions then determine the best way (if any) to invest in them by following up with individual technical (as in technology, not charts) and financial analysis. Discussions with gentlemen like Giorgio Anania help me start this process. At this time I’m far from finishing it.
One trend I see unfolding is the emergence of Chinese optical component companies. Every company is moving their labor intensive component manufacturing to Asia (including Bookham) either through outsourced manufactuing or directly owned facilities in low-cost labor areas.
There are also new Chinese competitors emerging that not only manufacture in China but also have Chinese SG&A and R&D. These new companies are direct competitors to companies like JDS Uniphase (NASDAQ:JDSU), Avanex Corp (AVNX), Bookham, and Finisar Corp (NASDAQ:FNSR).
I walked away from my meeting with Giorgio confident that these new Chinese companies posed little short term threat to Bookham. But this additional investment must pose a threat to someone, and if not Bookham then who? It also led me down the path of analyzing the merits of owning your own fab, as all of these Chinese manufacturers do.
This is our view of the optical component industry business models.
* Vertical - Company owns virtually all resources required to deliver product; from R&D, to some/all component manufacturing, test and assembly, and sales and marketing. Intel is a vertically integrated manufacturer of CPUs, and TI makes their own DSPs.
* Platform - Company owns R&D and Sales/marketing but outsources majority of the manufacturing and test process. Similar to semiconductor fabless model. Platform is an excellent term coined by GaveKal.
The Red Vertical model doesn’t yet exist in the semiconductor industry and has appeared first in optical components because R&D barriers to entry in some markets are low, and labor is a high input cost given current optical manufacturing technology.
The optical market itself is spread into two distinct areas:
* High End: Telecom, lower volume, high specification variability based on end customer and market, boutique, significantly more R&D input. Tunable lasers, EDFA’s, 300 pin MSA’s, Long Reach optics.
* Low End: Enterprise, much higher volume, low/no specification variability, commodity, cut-and-paste R&D. Gigabit Ethernet and Fibre Channel SFPs, 10GbE XFPs, FTTH SFP’s.
Bookham is only focused on the higher end telecom optical modules and assemblies. Most of these require crucial, high IP components within. Bookham makes these components, and even supplies them to some competitors.
The Chinese companies, without exception, lack the capability to build these components and must buy them. They therefore have no pricing advantage in these markets. The InP manufacturing capability Bookham has allows them to differentiate their high-end components. JDSU has this capability as well, though I’ve heard from anonymous sources that they would like to unload it.
The Chinese companies do have a business opportunity in the lower end, where R&D is not as important and components are multi-sourced and readily available. Several of my contacts in the semiconductor industry have indicated that the Chinese simply copy reference designs from their component suppliers and then manufacture them. Their only advantage is cost.
The Japanese suppliers are tough competitors due to heavy R&D investment without regard for return.
Let’s look at how the companies are grouped as a function of target market and manufacturing model. I’ve added in a few examples from the semiconductor market, a market that I believe is identical to optical components in the long term.
The Vertical model is clearly better suited for commodity manufacturing where there are few product variations and high volume. Suppliers that outsource realize less manufacturing cost reduction as volume increases than suppliers that pay the fixed costs on their own manufacturing assets. The higher the volume, the greater the cost gap.
Price is always set at the margin, and the marginal costs of vertically integrated manufacturers are lower than platform companies that outsource manufacturing.
Therefore, any platform company outsourcing the production of low-end high volume products is going to get killed. You don’t see any fabless DRAM, Flash, or high-volume processor companies in the semiconductor business for this reason.
Most high-end optical companies ‘dabble’ in low-end products for the sake of having a broad product line, and I would wager they continue to lose money on these endeavors.
The Chinese vendors have intelligently focused on the areas where their low cost manufacturing will reap the most benefit, and the commodity areas where a lack of astute R&D and SG&A will penalize them the least. I expect Avago and Finisar have insightful opinions on the new Chinese companies.
The battle for the high end will not be West vs. China, but Platform vs. Vertical. Most of the suppliers in the high end market use outsourced manufacturing. Bookham does not, with the exception of some SFPs. As a result, Bookham occupies an odd quadrant.
Why use the Vertical model for lower volume modules with higher Intellectual Property content? Why not accept the higher variable cost model until enough market share is secured to harness the economies of scale of in-house manufacturing?
In the semiconductor market, vertically integrated suppliers that try to supply niche markets continually struggle. These suppliers or the markets they serve eventually go away.
Bookham must be anticipating that they will be the largest volume supplier in the high end either through growth or consolidation.
* If they are wrong, Bookham would need to enter the low-end markets to chew up spare manufacturing capacity- otherwise the fixed cost structure of owned facilities will sink the company.
* If they are right they’ll be the clear #1 supplier for high end telecom components with a manufacturing advantage based on critical mass.
Please share opinions and be constructive. Corrections are welcome as I’ve covered a lot of ground here.