What Donald Trump’s presidency means for the flow of electrons.
This week the entire semiconductor market woke up with a severe political hangover. Aside from the initial shock of the election results themselves, the winning platform of “America First” could have far-reaching implications for an industry that has spent decades optimizing a global supply chain the way it has finely tuned other processes to reduce the cost per transistor.
There are many unanswered questions about what will happen, in what order, and to what extent. Still, it’s possible that a number of key pieces of this supply chain could be disrupted at the same time. And that could affect everything from how chips are architected, what materials are used, what IP is included, and where they are manufactured.
Let’s start at the top and work down. Right now, most large semiconductor companies—which is nearly everyone in the digital world excluding Intel—are fabless. That means they design chips with the most robust and least expensive bill of materials, and send those designs off to outside foundries. Much of this foundry work, minus a limited commercial operation by Intel, is done by companies that are headquartered outside the United States. Whether non-U.S. companies with U.S. operations (Samsung has a fab in Texas and GlobalFoundries has fabs in New York and Vermont) get a break has not been addressed. But no matter what happens, any regulations enacted in the United States will be met with similar regulations outside the United States.
It’s not just about who will manufacture these chips, though. It’s also a question of what they will use to make the chips. In August, SunEdison Semiconductor, the last major U.S. wafer maker, accepted a $683 million offer from Taiwan’s GlobalWafers. The deal is currently being reviewed by regulatory agencies. It may be viewed much differently now than it was earlier this week.
Dropping down a couple notches, many of the materials essential for leading-edge nodes are imported these days. Rare earth mining in North America has virtually disappeared, undercut by mining operations in other parts of the world. In interviews conducted over the past few months, semiconductor industry experts concluded those mining operations could be re-started, but it would take a couple years to get them rolling again. Without them, Moore’s Law becomes even harder and more expensive to continue and low-power, high-performance devices would be much more difficult to develop.
There are other elements that are important in the electronics supply chain, as well. Tantalum is a critical component of capacitors, which are widely used in mobile phones, computers and increasingly in cars. The largest producers of tantalum today are Australia, China, Ethiopia and Mozambique, although exploration is underway in other countries. The tantalum found in the United States is considered such low grade as to be commercially useless.
It’s not that chips cannot be designed and developed differently. There are many architectures and approaches that have been discarded in favor of less expensive options. If tariffs are enacted, or if supply chains are politically impacted, some of those alternatives may be necessary. It’s far too early to call, as any changes will likely take months or years to roll out.
But it’s not too early to at least start thinking about possible changes and reviewing everything from design strategies and to what IP and what materials are used, where they come from, and what can be substituted and for what cost. Change is coming, and it will almost certainly affect the flow of electrons around the globe and how much it costs to move them.
Trade War Looms Over Materials
EU, U.S. file complaints against China over raw materials.
Still Searching For Rare Earths
There’s good and bad news for buyers of rare earths.
Will There Be Enough Silicon Wafers?
Market pressures shrink number of providers of basic building blocks for chips.