There has been much dire commentary from experts as to how Japan’s earthquake could, might, or may be very disruptive to our industry. It is true that the earthquake could, might, or may have been very disruptive to our industry. It was certainly a disruptive weekend and week if you work in supply chain management. But historically, these sorts of shocks to the supply chain have not been disruptive on an annual basis. Even this time, while there was a spike in flash prices on the Monday right after the quake, they were already moderating by the end of the week.
As big as the quake was, the primary source of problems for the industry is not the earthquake, nor even the tsunami. It was the low preparedness of Japan’s power industry. This includes poor regulatory oversight, political payoffs and choice of site location, making this a disaster waiting to happen.
The biggest thing to learn from the table above is that recovery depends highly on how well prepared a company is in both the planning that went into siting and building the fab, as well as having ongoing safety processes in place. What made Taiwan’s Chi-Chi earthquake so bad was the fact that simple things like tying tools down and double containment of gases and chemicals were not in place. If the same earthquake were to happen today, Taiwan would come out far better. While the quake was severe, the vibration had largely dissipated by the time it hit the semiconductor production areas.
While the Sendai earthquake was one of the worst in history and ranks first in big shakers for the semiconductor industry, the actual severity was limited because it was so far out to sea. Contrast the USGS shake map for Sendai against that for the Loma Prieta earthquake. In the areas where the closest fabs were, Loma Prieta was more severe. As for Toshiba’s flash fabs, they were way down past the green zone on the chart. This is why they sustained so little damage and recovered on the same day of the quake.
Moreover, most of Japan’s semiconductor production was not affected by the Tsunami either. The biggest issues currently affecting Japan’s semiconductor production are the same as I wrote about last week: power and transportation. If a fab is subject to the 3-hour rolling blackouts, it can’t run. Moreover, Japan’s rail systems run on electricity and their scheduling systems didn’t incorporate plans for such an event.
As for Toshiba’s flash fabs, they have been able to get uninterrupted power, so this is not a problem there. This comes back to the point about planning for these disruptions: it’s always good to have your own source of power or at a minimum an exemption similar to what hospitals get where the power cannot be shut down. Another example is Globalfoundries' Dresden fab that was built with a co-generation plant. It paid for itself by selling power back into the grid during peak hours, just like a homeowner would do with solar panels. When a flood took out the local power plant, it went right on with no interruption.
Thank you Dan for a very informative article. It might be purely coincidental, but the magnitude of the earth quake has increased every time for the last three incidents!
Has the estimation of the damage due to the recent earthquake/ tsunami in Japan been completed? How long it might take to recover for the electronics industry?
@rspell2, thank you very much for the update. I know that modern reactors are ready for these incidents, but why did Japan not retire or retrofit the older facilities to be ready for this? Of course it was unexpected, but the new reactors are able to withstand it. They must have known the older ones were not.
The nuclear plants at Fukushima were designed for earthquakes, and even tsunamis - though apparently not for one of the size that actually occurred:
"While the media focuses on the serious problems in a single cluster of older reactors, the more modern reactors shut down gracefully. Of 54 reactors in Japan with a combined capacity of 48 gigawatts, five clusters were affected by the quake with four shutting down successfully despite sometimes significant damage.
The cluster closest to the epicenter of the quake was Onagawa. It experienced significant damage, including a fire, but shut down without problems. Fukushima Daiichi isn't far from its slightly newer sister plant, Fukushima Dainii. The four reactors at Dainii shutdown successfully despite significant damage caused by the quake.
The two plants were designed to withstand a 5.7 metre tsunami, but the tsunami that hit these two plants was 14 metres high."
Very good article. I enjoyed the learning. It is a shame that nuclear power will take the biggest hit from this disaster. I don't understand how Japan can have reactors that are not ready for earthquakes.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.