Tech mergers and military electronics obsolescence
Most industry folks I speak with agree that the current procurement climate is driving commonality and open architectures, which often means more investment in commercial off-the-shelf (COTS) technology. Another question: What about obsolescence? Have the Department of Defense and industry gotten any better at navigating obsolete components?
Unfortunately, the military no longer drives technology development. The military is a consumer of technology just like the rest of us and is also subject to the whims of commercial markets; it then passes these pressures to defense prime contractors and system integrators.
A perfect example dates back to 2008, when Apple Inc. bought PA Semi (Palo Alto Semiconductor), which made low-power 64-bit processors called PWRficient – high performance, low power, and loved by the military embedded computing community – but Apple only wanted the engineering talent, not the products. Defense electronics suppliers then had to scramble and relaunch product families with Intel and Freescale tech, but they still suffered losses as a result.
The point: Defense electronics suppliers get obsolescence. They have learned the hard way. As the conduit between commercial technology and military systems, they take that technology and enable it to work in mission-critical harsh military environments. They also convince their customers they have a plan for when these commercial components go end-of-life. However, not all of the customers have the volumes to survive long-term when they have to guess if military funding will be short-term, long-term, or happen at all. It can be like buying a lottery ticket.
They still have to guess when it comes to figuring out how much obsolescence the government is willing to pay for, says Dan Deisz, Director of Design Technology at aftermarket semiconductor supplier Rochester Electronics.
“The way the Department of Defense (DoD) goes about managing obsolescence could be improved. OEMs, when competing for a new contract, are required to keep costs low and flexibility high – planning for refreshes that may never get funded or for a contract that may not be won,” he notes. “This forces them to do a derivation of what they’ve already done before, which hints toward using older and proven semiconductor products.”
Today’s semiconductor market is echoing that of 2008 with mergers and acquisitions making news, albeit in much less subtle ways than Apple’s purchase of PA Semi.
“The problem is that the semiconductor industry is a bit like the [old] Pac-Man video game right now, with big companies gobbling up smaller ones,” Deisz continues. “Examples include Intel buying Altera, Avago Technologies buying Broadcom, and NXP buying Freescale. The DoD needs a better macro plan for dealing with these changes, as many electronics systems use the PowerPC architecture or Altera FPGAs. One or more of these could be dead in a year or two and those without a plan for dealing with that will be in trouble. Yet I don’t see any change in how the DoD partners with semiconductor companies to better manage these challenges. Altera raised their prices even before they were acquired to several companies this year. Xilinx is likely going to do the same because they can.”
He poses the question I had, which was “How will the DoD manage that change? Is that the right space for the DoD?” The government represents less than one percent of total semiconductor spending. In other words, it does not drive semiconductor product designs or prices and is thus at the mercy of consumer-market trends.
Deisz says he has a solution. “One way to fix this would be to focus on total life cycle cost at the start,” he says. “In other words, tell me the true cost to maintain the system for 30 years up front and put everything in place from a semiconductor perspective to make the same system for 30 years. The semiconductor content wouldn’t be a heck of lot of money compared to what they are spending now. Then every single change to the system would need to be uniquely justified in annual budgets. This way, contractors would be duly incentivized to design for obsolescence; funded refreshes would then be locked in and paid up front. Right now, both the DoD and DoD OEMs are jointly incentivized to design in flexibility and the potential for change to try and capture future technology capabilities. When that future is unpredictably funded, unfunded refresh will drive obsolescence and counterfeit demand as it is today.”
Admittedly, Deisz says, this would create more work for DoD personnel and OEMs, “but it could solve long-term obsolescence issues and at least bring full awareness to the costs up front. The way things are now, refresh cycles for many programs cost more than the platform itself. It could be better off up front if the DoD were to do mature technology ASIC designs instead of leading-edge FPGAs in several cases, and to lock it down for 20 years. They could add additional capabilities when necessary or design with more industry standard interfaces, but keep that core technology locked down.”