The UN and Transceivers

Change is coming to this country with the Election of Barak Obama. Today I saw a headline calling it the end of the Baby Boomer Era. Minority President, Democrat, Centrist, and Diplomat with a new tack on Gitmo, Iran, the economy, and capability to use a Smart Phone amounted to an end to era in my book before adding baby boomers malaise. Obama is a technologist’s President, which is important to us at COTSWORKS. He uses Email, he used the Internet to raise money, get the word out, and generally politic his way to winning an election. And he's fighting giving up his Blackberry. As a small business owner, I’m very interested to see what he has to say about the “backbone” of America; more interested in this than the upcoming inauguration pageant. If I were invited to the ceremony, my toast would be: "here’s to the end of Q1…"

While our economy is stalling and worth his primary focus, he will have to look not only inward but outward. Our world is mostly connected (Madagascar still lacks an undersea fiber optic link so he can not call on a Monroe Doctrine as hemispheres matter much less these days. Central to a foreign policy is how he uses the United Nations. The UN is a problem because it is a broken system. There are no checks and balances to protect minority interests. There are no popular elections of judges in the World Court (not that a popular vote would be a good thing, but at least an expression of the people of this planet). The UN is made of countries where our standard of human rights is not applied. Women’s suffrage and a free press don't exist in most countries around the world. After living on this planet for a few million years have we not evolved enough to see that women have an equal say in how the world works? Or that if I want to complain about my elected leader I won't fear ending up in the Coliseum as a result? The USA is imperfect...but at least we now can control the fans at today's coliseums instantly and even anonymously (thank goodness for technology) just by texting.

The UN reminds me (in a non-human rights way) of the Multi-Source Agreement standards committees that govern how transceivers are built. Representatives from big companies who make parts get together and decide what an industry standard should be as long as they can benefit from it. Then they run back to their respective companies to try and get parts made the fastest and cheapest way possible and dominate a market segment. No one enforces the MSA or provides an industry neutral testing certificate. Just as the UN has no way to enforce its decrees, neither does that MSA committee.

The major transceiver vendors in the MSA, Finisar, JDSU, Avago, Source Photonics (MRV), and Emcore (the G8 or Security Council) among others keep plowing money into projects that don't yield any positive return. It’s mind boggling that billions of dollars can go into building what engineers ten years ago would never believed possible and is now sold at below 30% gross profit often leaving a negative or minimal net profit. Even with complete vertical integration of manufacturing, these companies probably make more on the sheet metal than they do once they put a laser inside. How did this mess happen? Why can't these companies reach a level of integration where they are profitable?

I got in a discussion with one of the researchers at Lightcounting about Innovation. I don't think that innovation is simply faster and smaller. When no one makes money and the only barrier to entry is when Chinese OEMs upgrade their design sets, you have a broken system that stifles innovation. Is it innovation to go from 100Base to 40G in 10 years? To the network equipment vendors, yes. To the optical and electrical scientists who shrink the components and solve dispersion, impedance, or BER issues, yes. To anyone making transceiver? I'm not sure.

10Gb/s transceiver shipments are growing at a whopping 92% CAGR and will probably continue for the next few years. Having created my own YouTube channel...I hope to contribute to even faster growth. The MSA, however, has imprisoned companies in rat race that leads to fewer companies in the short term and no clear path to profitability. The Chinese transceiver companies are now making 10G parts. The so called Tier 1 transceiver manufacturers wear the Blue Helmets and get shot at without ever seeing a return. Why shouldn’t Finisar have 55+% Gross Profit like its customers? Looking at this picture from a distance, it may turn out that public and private investors are spending several trillion dollars in Research and Development (operational losses) over some twenty years to enable optical networks where the end users benefit but the manufacturers never see a return. They take a 1/2 step to the wall each time. Ultimately, the MSA and the transceiver companies have to see a new direction soon or they will hit a bottom where 2-3 silver haired engineers at a committee meeting decide what the Chinese OEMs will make next year at the request of network OEMs who have booked profitable service contracts with end users.

When will this end? Somewhere down the road a transceiver will be built into part of a network chip or PHY just as it is on copper. The best analogy I can see is to look at the copper network drivers. And there, the market shrank to just a few, Broadcom, Marvell, and Intel for the bulk of the market with a few vendors doing high speed or alternate protocol (10, 40G and Fibre Channel). Wouldn't it be cool to look inside a PC and see a chip with duplex LC connectors sticking out of it? But whose name will be silkscreened on the chip is the trillion dollar question.The bottom, then, is when circuit boards have optical interconnects in them. USB 3.0 will help. VITA 42 will help. But somehow, the O to E has to be an O to O in order for transceivers to become chips on boards. I can’t tell you when that will be, but I don’t see it too soon. Less a technological problem, this barrier is one of the biggest problems in aerospace: the connector. You can not solder fiber.

The MSA has democratized transceivers and the associated standards which is good, but since they only seem to care about the current standard, say 40G, the old 1G standards get old and stale. The image below shows a readout of a Finisar SFP using our Capstone TM software and OptoTrek test board .

Just about every SFP made today surpasses the MSA spec’d numbers. The readout to the left is highly formatted so it hides the old ratings for output power or Rx sens but they are included in default text in the Serial IDs of the modules. The Vendor OUI listed isn’t dynamic or...regulated. Why couldn't transceivers connect to an online specification server to get updates and check their accuracy? The Vendor Programmable section is used by companies like Cisco to lock out non-Cisco parts in their switches without providing any real value. Any engineer with i2C knowledge can reprogram them and put their own company in (a Trademark violation) and go to market as an OEM. Supposedly, the FDA is watching how lasers are shipped in this country but where would they start to enforce regulations? Getting a part on the open market through a Google Search yields hundreds of companies that sell SFPs. Many of them are Chinese OEMs where they have an equal number of spelling errors as products…how can you trace it? If Fiber To The Desk ever arrives with 10G LAN servers, we may see local shops and nationwide chains of transceiver exchange or repair.

In any system, the strength and reliability of that system is its weakest link. With the price pressures that exist today in optical transceivers, the weakest link is not the single SFP out of 100 deployed that might fail, but the whole business of profitably manufacturing pluggables.