ISC 2010 Wrap-Up –What’s Hot in Supercomputing

This week’s blog post comes from Katharine Schmidtke, Finisar’s Strategic Marketing Manager.

I had the pleasure of attending the International Supercomputer Conference (ISC) event May 31-June 3, in Hamburg Germany. The show was well attended and appeared to have doubled in size from last year. In attendance were all the major industry suppliers and leading technology providers such as Mellanox, QLogic, Voltaire, NVidia, HP, IBM, Cray, Oracle, AMD, Intel, LSI, Supermicro and STEC. Key themes included supercomputing, storage and networking. Of those, hot trends discussed were cloud and parallel computing as well as the future developments to come in the next 10 years. The trend to higher bandwidth continues, with the IBTA announcing the latest roadmap which adds a new datarate at FDR (14 Gbps) in addition to EDR (26 Gbps). InfiniBand is targeting 300 Gbps by 2011 in a 12 lane format running at 26Gbps per lane.

And as I sat through the presentations and spoke to some of the folks on the show floor, it became evident that there are two central themes underpinning all the major business and HPC initiatives – latency and flexibility. Reducing latency is key for supercomputers, while flexibility is the driver in cloud computing. NVidia made quite a splash with their Tesla C2050 GPUs which are used to improve data crunching speeds in many supercomputers, including the new #2 supercomputer – the Nebulae system built by Dawning in China. Dr. Wilfried Oed from Cray shared some of the secrets in the new XE-6 supercomputer including the new Gemini network card which he admits is “more than just a router” by increasing processing speeds using a clever non-blocking routing system.

On the news front, the biannual 35th edition of the Top 500 supercomputer list was released. The U.S. continues to take the lead in the number one spot with Jaguar, the fastest supercomputer system used for The Department of Energy’s Oak Ridge Computing Facility. Trailing close behind and ranked second on the list is Nebulae, China’s fastest system worldwide.

Other noteworthy technical initiatives point to Intel, who dominates the high-end process market with roughly 82% of all systems and over 90% of quad-core based systems. IBM recoups the lead in market share by total systems and overall performance from Hewlet-Packard (HP). And exciting for Finisar – we were part of the 120 Gb/s InfiniBand demonstration at this year’s event. Check out the press release issued by the HPC Advisory Council to learn more.

Until next year’s ISC show, Auf Wiedersehen, as they say in German!

K.Schmidtke 2009

Finisar at ISC Hamburg Germany 2010

Finisar at ISC 2010

Growth Through Murders and Inquisitions

This week’s post comes from Finisar CFO, Steve Workman.

I was invited by Rafik to do a guest post on Lightspeed. I considered various topics including the adrenaline rush we all get working on the “ragged edge” of accounting, but then it occurred to me that you can’t truly know Finisar without understanding our views on M&A. Finisar has used M&A extensively in developing its vertically integrated business model. In doing so, it has targeted a variety of companies, both private and public, big and small, foreign and domestic. To date, 12 transactions have been consummated in building our optics business although we looked at many more along the way and even backed out of a couple. And we have done so almost irrespective of the macroeconomic environment or stock price at the time of the transaction. As long as we felt the deal could be accretive within a relatively short period of time (generally one year to allow for restructuring although manufacturing synergies can take longer), then the transaction qualifies for consideration. Of course, pure technology deals can take longer to pay off and generally require that we take a healthy dose of skepticism before evaluating the opportunity.

How successful have we been with our M&A strategy? We ask ourselves that question all the time. We had a couple of false starts with respect to establishing an internal capability for building lasers (this is hard). And we had the chutzpah to announce a hat trick (3 deals at one time for those who don’t follow hockey) back in the internet bubble when we probably didn’t fully appreciate some of the integration issues you can run into. While some transactions have definitely worked better than others, we have never been shy about using M&A to sharpen our competitive edge. As noted in a recent Harvard Business Review article, “while M&A activity has been severely depressed since 2008 and fell dramatically in early 2009, acquiring companies during that period tended to outperform their industry peers in market valuation, according to a global study by Towers Perrin and Cass Business School examining 204 deals, each worth more than $100 million.”

In Finisar’s case, we have greatly expanded our product lines, patent portfolio and R&D capabilities through M&A. Our optics revenues have grown from just $47 million in fiscal 2000 when we became a public company to an annual run rate of over $580 million as of the most recent quarter reported and was headed higher based on our guidance at the last earnings call. In the process, we have become the world’s largest supplier of optics for communications applications. But “being big” is not what drove our M&A strategy. To understand what did, we need to take a walk down Finisar M&A Lane.

In the heady days of the internet bubble, we should remember that did not make any of the key components that were used to build our transceivers such as lasers or ICs. Furthermore we built all of our products using outside subcontract manufacturers instead of doing it ourselves. During that time, we were at the mercy of some of our competitors in terms of obtaining key components. Without an adequate source of supply, market share can suffer as key customers direct more of their orders to those who are most likely to supply enough product in the timeframe needed. In addition, attempts to introduce new products earlier than our competitors were compromised as a result of not having early access to new versions of those key components. Our experience from the bubble underscored the importance of vertical integration for certainty of supply and for accelerating product development while having our own off-shore assembly and test operation was going to be important in order to control access to our technology and exercise greater control over product quality while keeping costs low. It wasn’t until later that we realized that having an internal manufacturing assembly and test capability also gave us the ability to respond more quickly to upside surprises which happens almost irrespective of the economic environment at the time.

If we just look back at the goodwill we wrote off in fiscal 2009, you could reach the conclusion that we sometimes venture off the reservation. But goodwill impairments are almost inevitable given the macroeconomic cycles we must work through periodically. In the case of Optium, we entered into a merger agreement where we agreed to issue shares for a certain percentage of the combined company in May 2008 which determined the amount of goodwill that would be booked. The transaction was finally approved by shareholders in the fiscal quarter ended October 2008. During that same quarter, Wall Street went into a meltdown and the economic consideration of what we had issued was no longer supporting the amount of goodwill that was generated according to the rules of GAAP accounting. As a result, we were forced to write off all of the goodwill in the same quarter that it was generated. Would we have done the same deal had we fixed the exchange ratio much later when our stock price was lower? Of course we would, because Optium’s stock price would have been lower as well. A falling stock price does not mean the relative contribution of each party to the value of the total has changed. If we had done the same transaction later in the year when stock prices were lower, the amount of goodwill written off would have been considerably less or perhaps none at all. The fact that companies record a goodwill impairment is sometimes more a statement about the economy at the time the transaction was undertaken rather than a problem with the transaction itself.

So how successful have we been with our M&A strategy? I’d say good enough so we aren’t shy about considering new opportunities, but have earned just enough credits from Hard Knock U. to be very careful as we evaluate them.

Next Stop: The Land of Down Under

In one week, you can catch Finisar’s Simon Poole as a plenary speaker at the Australian Conference on Optical Fibre Technology (ACOFT) – set to occur November 30 – December 3 in Adelaide, South Australia. The session will take place at The University of Adelaide on Tuesday, December 1st at 9:35a.m. ACDT. Simon will share his thoughts on what the future holds for photonics-based technologies, the lessons to be learned from the technological and economic changes over the past couple of decades and engage in some crystal-ball gazing about what might happen over the next 10 years. Simon has been a regular presenter at conferences around the world over the past two decades and this presentation will be 22 years, almost to the day, since Simon’s first ACOFT paper when, as a newly-minted PhD, he gave an invited paper on Rare-earth doped fibres and EDFAs at ACOFT’87 in the wonderfully-named Surfers Paradise. Check back on “Out of His Depth” (coming soon) to hear more from Adelaide.

150 Years Under The Sea

This week’s blog post was written by Finisar Finance Director, Tom Downey.

A recent summary of undersea cable construction reveals the continued investment in infrastructure for delivering bandwidth via optical fiber. Before looking at recent developments, it is interesting to look back at a bit of undersea cabling history.

FNSRblog_UnderSea_Table 1

Before the late 1990’s, subsea cables carried telegraph and voice traffic. Then the dot-com and telecommunications frenzy created unprecedented construction under the sea as well as on land. In 2001, $13.5 billion was spent. However, with the boom, came the bust; only about $2 billion was spent between 2004 and 2007. In the boom, spending was speculative and thus, very high. However, with the creation of bandwidth, because customers could transport more traffic into the network clouds, they did! Even during the lean cable spending years, demand never slaked, growing at an average compound annual rate of 54 percent in three years from 2002 to 2008. The table below shows compound annual growth rates of lit submarine capacity vs. bandwidth use.

Now, carriers are adding capacity where new demand had soaked up the earlier glut. Spending will be lower because new cables will cover shorter distance, regional runs. With the continued demand, over 25 new submarine cables should enter service in 2009 and 2010 with construction spending over $4 billion.

FNSRblog_UnderSea_Table 2

The continued investment in undersea capacity underscores the modern day story of the optical networking revolution. New applications continue to drive ferocious demand for bandwidth and the migration of the applications from a client’s device into the network. With the demand, component manufacturers continue to face the overarching challenge – successfully meeting the world’s insatiable demand for connectivity and bandwidth.

Sources:
IEEE Spectrum July 2009
http://www.spectrum.ieee.org/telecom/internet/a-telecom-diet-rich-in-fiber
http://www.atlantic-cable.com/Cables/CableTimeLine/index.htm