The self-defeating trade policy affecting memory chips

Affordability is a major concern for many Americans. Whether you’re checking out at the grocery store, filling up at the gas station, or paying your bills, high prices have become a frustrating facet of everyday life. New electronics, in particular, are becoming more expensive as high demand and federal policy drive up production costs.

Many of the current and forthcoming price increases in electronics (particularly smart devices) are related to higher costs for a key component: memory chips. Memory chips are semiconductors that store data and program instructions in electronic devices. Different types of memory chips serve different functions, but some of the most widely used are Dynamic RAM (DRAM) and NAND flash. DRAM serves as a device’s working memory, holding the data and instructions a processor is actively using. DRAM is fast but volatile, losing its contents when power is cut. NAND flash is a non-volatile storage chip. Though slower, it retains data without a power source (which is why most devices rely on both DRAM and NAND flash chips).

Laptop, PC, and smart device buyers are not the only ones feeling the effects of rising prices for key memory chips. DRAM and NAND chips are also vital for the current 5G telecommunications infrastructure. Though far more attention (both in policy and in the media) is devoted to key logic chips (such as central processing units and graphics processing units), memory chip costs are quickly becoming a drag on the digital ecosystem.

To be clear, AI-driven demand is likely the most significant force behind rising memory chip prices. Accordingly, the only durable solution to address the demand-driven shortage of memory chips is greater supply. Yet current trade policy works against the very efforts that would expand memory chip supply.

Federal policy from the Biden administration’s CHIPS and Science Act to the current Trump administration’s trade policy demonstrates an overarching desire to reshore American semiconductor manufacturing. Putting aside any merits this form of industrial policy may have, successive presidential administrations have made clear their intentions to build upon and expand the productive and manufacturing capabilities of the US semiconductor industry. Unfortunately, the current administration’s trade policy undermines this goal.

Much of the relevant tariff regime relies on Section 232 of the Trade Expansion Act of 1962, which authorizes the Commerce Department to investigate whether imports of a given product threaten national security and empowers the president to impose tariffs in response. Crucially, Section 232 is an ongoing statutory authority rather than a one-time policy. The Trump administration has invoked it in separate proceedings covering different products, each with its own investigation and tariff schedule. Two of those proceedings have implications for American memory chip production, with one targeting semiconductors and the other targeting metals (such as steel, aluminum, and copper).

The rationale behind the semiconductor proclamation is what makes the tariffs on metals difficult to reconcile. The Commerce Department found that the United States manufactures only about 10 percent of the chips it needs and treated that dependence on foreign supply as a national security risk (the very risk reshoring is meant to address). Yet the same Section 232 authority, invoked against steel, aluminum, and copper, raises the cost of the domestic production that the semiconductor finding says the country needs.

Semiconductor fabrication plants (“fabs”) are particularly metal-intensive industrial structures, built with heavy steel frames, extensive copper wiring, and large cooling systems. Micron, the only major American manufacturer of memory chips, reports that its Idaho fab has a structural backbone framed with some 70,000 tons of steel. The recent Section 232 metals tariff raises the cost of that material across the board. As of April, those tariffs bite harder as the duties are now assessed on the full customs value of covered steel, aluminum, and copper articles and their derivatives, rather than only the value of the metal content. Thus, the tariff falls on fabricated components and finished structural inputs, not just raw metal. Even when sourcing domestic material, fab developers face inflated prices. Micron is building new DRAM fabs in Idaho and New York under precisely these conditions. If the goal is to expand domestic memory chip production, taxing the plants that produce those chips is counterproductive.

The tariff regime itself all but acknowledges the problem. The same proclamation carves out metal-intensive industrial and electrical-grid equipment, capping the combined duty on those goods at 15 percent through the end of 2027, well below the 50 and 25 percent rates that fall on other covered goods. It is hard to explain the existence of a special, lower tier for exactly the equipment a domestic buildout requires as anything other than a tacit admission that these tariffs raise the cost of building American industrial capacity.

Both the Biden and Trump administrations have made reshoring American chip production a national priority. But current trade policy runs counter to that goal. It raises the cost of building the fabs that reshoring requires, under the same statutory authority invoked to protect the industry (a contradiction the administration effectively concedes by capping the tariff on the equipment those fabs need). By taxing the construction needed to expand domestic memory chip production, this tariff regime delays the very supply increases needed to ease the shortage.