Join War on the Rocks and gain access to content trusted by policymakers, military leaders, and strategic thinkers worldwide.
The U.S.-Israeli war with Iran, now in an unstable ceasefire, has exposed a structural failure in the global semiconductor memory supply chain, and it is not the one analysts seem to be tracking. The story receiving attention is helium: Qatar’s Ras Laffan facility went offline, a 45-day inventory clock started running, and spot prices doubled within days. The story receiving almost no attention is bromine, and it is potentially the more dangerous one. Bromine is the raw material from which specialized chemical suppliers produce semiconductor-grade hydrogen bromide gas, the etch chemical that South Korean fabs use to carve the transistor structures in every Dynamic Random-Access Memory (DRAM) and NAND flash chip on earth. A DRAM chip powers active computation and loses its contents the moment power cuts. A NAND chip retains data without power and underlies every form of digital storage. Together they underpin every modern computing device, from the phone in your pocket to the data center running your AI applications.
South Korea sources 97.5 percent of its bromine imports from Israel. Beyond that vulnerable concentration, converting bromine into semiconductor-grade hydrogen bromide gas requires dedicated purification infrastructure, and producers outside Israel are already fully committed to existing customers and stretched too thin to absorb additional demand. Building new conversion capacity takes years of permitting, equipment procurement, and fabrication qualification.
ICL Group, the Israeli multinational formerly known as Israel Chemicals Ltd., currently continues Dead Sea operations. Israel routes most trade through Mediterranean ports at Haifa and Ashdod, bypassing the Strait of Hormuz entirely. But Iran has been striking the Negev — Israel’s southern desert and the heart of its defense and industrial infrastructure — with ballistic missiles for three weeks, hitting Dimona and Arad, both within 35 kilometers of ICL’s Dead Sea extraction and conversion complex. If Israeli bromine production is displaced, there are no conversion facilities outside Israel capable of immediately producing semiconductor-grade hydrogen bromide gas at the scale required to replace it, and policymakers have not yet acted on that fact.
The vulnerability sits in plain sight, within missile range and outside any meaningful policy response. A disruption would be immediate and global. Within weeks, shortages would propagate across everything from consumer devices to military systems.
Bromine’s role in semiconductor manufacturing is specific and non-substitutable. Its primary derivative, hydrogen bromide, is consumed at the polysilicon etching stage foundational to both DRAM and NAND flash production. Each DRAM memory cell requires a polysilicon gate electrode etched with extreme precision over a silicon oxide layer as thin as 20 angstroms. Hydrogen bromide gas plasmas achieve a polysilicon-to-oxide selectivity ratio of 100 to 1, while chlorine-based alternatives achieve roughly 30 to 1. At advanced DRAM node geometries, that is the difference between a functional transistor and a destroyed one. Bromine also appears in chemical vapor deposition processes and chip packaging. There is no viable near-term substitute in any of these applications.
Three structural realities determine why the gap cannot be bridged through market reallocation. First, bromine already converted for industrial use such as flame retardants and drilling fluids cannot be reconverted. Those processes are chemically irreversible at any industrial scale and the resulting compounds cannot meet the parts-per-billion purity specifications that fabrication facilities require. The two supply chains draw from the same raw material but diverge permanently at the point of conversion. Second, converting raw bromine to semiconductor-grade hydrogen bromide gas requires dedicated purification infrastructure, specifically gas-phase distillation columns capable of lowering trace metals to parts-per-billion contamination levels. That infrastructure does not exist at scale outside the existing semiconductor chemical supply chain, and building more facilities requires permitting, equipment procurement, testing, and fabrication qualification measured in years. Third, producers such as Resonac, Air Liquide, and Adeka manufacture semiconductor-grade hydrogen bromide gas outside Israel, but their combined capacity is already committed to existing customers: Taiwan Semiconductor Manufacturing Company, the world’s dominant contract chipmaker; Samsung, the leading producer of DRAM and high-bandwidth memory; and Semiconductor Manufacturing International Corporation, China’s largest state-backed foundry. Critically, those customers are not holding steady: AI infrastructure buildout is accelerating demand across the board, meaning outside producers are stretched thin against a growing baseline. Even if outside producers could expand output, South Korean facilities would be competing for that capacity with Taiwan, Samsung’s own logic plants, and China, all of whom face the same accelerating demand.
The Dead Sea is among the most bromine-rich bodies of water on earth. ICL Group, which extracts at the lowest cost of any producer globally, dominates a supply that Israel and Jordan together account for roughly two thirds of globally. Critically, ICL’s hydrogen bromide gas production, including the semiconductor-grade output supplied to South Korean fabrication plants, is manufactured at the same Sodom facility where extraction occurs, meaning extraction and conversion infrastructure are co-located in the same vulnerable corridor. Iranian missiles have already penetrated Israeli air defenses in the Negev on multiple occasions, wounding nearly 200 people in Dimona and Arad, both in the same geographic corridor as ICL’s production and conversion sites.
The mechanism of disruption does not require a direct hit on an ICL facility. War risk insurance for vessel calls at Israeli ports has already risen from 0.2 percent to between 0.7 and 1.0 percent of vessel value per seven-day call, adding up to $500,000 in costs per voyage on a mid-sized cargo ship. Even for ships routed through the Mediterranean rather than the Red Sea, those insurance costs apply the moment a vessel calls at an Israeli port. The war risk premium follows the port, not the route. ZIM, Israel’s primary shipping line, has implemented a “war risk premium surcharge” on all cargo to and from Israel. Haifa oil refinery — the country’s largest — was shut down after its power station was damaged in an Iranian attack, demonstrating that critical industrial infrastructure does not require a direct strike to be forced offline. The downstream consequences of even a partial disruption to that corridor would propagate immediately across the global memory supply chain.
Samsung and SK hynix together dominate approximately 70 percent of the global DRAM market. SK hynix alone holds roughly 57 percent of the high bandwidth memory market. Since DRAM and NAND underpin every modern computing device, a supply disruption would propagate across the full consumer and industrial electronics stack, not only AI infrastructure. High bandwidth memory — a specialized form of DRAM stacked vertically to deliver the data speeds that AI accelerators such as Nvidia’s graphics processing units require — is sold out through 2026, and DRAM suppliers hold only two to three weeks of inventory. A shortage would force both companies to allocate scarce hydrogen bromide gas to their highest-value lines — high bandwidth memory for AI accelerators — at the expense of commodity DRAM and NAND used in phones, personal computers, laptops, and data storage. The consequences fall hardest across Africa, South Asia, and Latin America, where memory already accounts for 15 to 20 percent of the bill of materials for a mid-range smartphone. That share rises sharply for budget devices, the primary gateway to digital participation across Africa, South Asia, and Latin America. Smartphone prices in Bangladesh have already risen 10 to 25 percent in 2026 as a direct result of DRAM and NAND inflation, with similar increases reported in Nigeria and South Africa. Budget smartphones are reverting to 4 gigabytes of RAM in 2026, precisely as on-device AI features demand more, not less. A bromine supply shock would price hundreds of millions of people out of the devices through which they access banking, education, healthcare, and economic opportunity.
The exposure extends beyond commercial technology. The majority of guidance systems, radar modules, and electronic warfare packages fielded by the U.S. military run on DRAM and NAND flash chips sourced from the same commercial facilities, on the same allocation logic, with less procurement flexibility than commercial customers. Since the Defense Department shifted to commercial off-the-shelf procurement in the 1990s, there is no separate defense-grade memory supply chain. A shortage that forces Samsung and SK hynix to prioritize high-margin high bandwidth memory for AI customers would deprioritize the commodity DRAM used in precision-guided munitions, intelligence platforms, and shipboard radar systems, with no government visibility into how that allocation decision gets made. The same war straining ICL’s operational continuity is simultaneously depleting munitions stockpiles whose guidance systems depend on the same memory supply chain. The supply stress and the demand spike are running in the same direction at the same time.
The consequences for American AI follow directly from the South Korean exposure but run through a supply chain that most U.S. policymakers have never traced. Every Nvidia Blackwell and Rubin graphics processing unit requires high-bandwidth memory stacks that come almost entirely from SK hynix and Samsung, as SK hynix is Nvidia’s primary high-bandwidth memory supplier for both platforms. Microsoft, Amazon, Google, and Meta are deploying hundreds of billions of dollars in AI infrastructure on delivery schedules that assume South Korean plants will have uninterrupted access to the etch chemicals they need. A bromine disruption produces delivery slippage, renegotiated contracts, higher spot prices, and delayed server deployments.
Three levers are available, and they require action simultaneously. First, the most immediate is physical pre-positioning. Arkansas bromine from Albemarle and TETRA Technologies cannot be used directly in chip production, but it could serve as feedstock for semiconductor-grade hydrogen bromide gas conversion if that infrastructure existed, which is precisely the gap that ought to be closed. South Korean companies could also establish bromine forward contracts locking in supply and price for 12 to 18 months.
Second, the single most important structural action is the one with the longest lead time: building semiconductor-grade hydrogen bromide gas conversion capacity outside Israel. The Chip 4 framework should be extended to include a critical materials annex with a coordinated allied program to site, permit, and fund dedicated gas-phase distillation infrastructure capable of achieving parts-per-billion purity in geographically diversified locations — particularly in South Korea, Japan, and the United States. Private firms will not build conversion infrastructure at this scale and speed without government mandate, offtake guarantees and permitting priority.
Third, each government should take action in their own lanes, but in a coordinated fashion. South Korea should designate bromine a critical mineral, mandate minimum inventory levels, and fund domestic conversion infrastructure jointly with Samsung and SK hynix. The United States should add bromine, semiconductor-grade hydrogen bromide gas, and the full range of specialty gases derived from bromine to the critical minerals list, and use Defense Production Act authority and CHIPS and Science Act funding to co-invest with allies in purification capacity on allied soil. Israel should formalize bromine as a strategic export commodity, harden ICL’s production sites against missile attack, and use the 2030 Dead Sea concession expiration to bring in allied capital in exchange for long-term supply priority.
In sum, the bromine risk sits outside every dashboard anyone is monitoring. The structural failure is not the war: It is that the global memory supply chain has built itself around a conversion chokepoint with no redundancy and no fallback. If ICL’s Sodom facility goes offline, the gap does not get filled. The action that matters most — building semiconductor-grade hydrogen bromide gas conversion capacity outside Israel — takes years. The actions available now — forward contracts, inventory mandates, and Arkansas feedstock development — buy months at best, not years. That gap is precisely why these three countries should move now, before an Iranian ballistic missile makes the answer irrelevant.
Alvin Camba, Ph.D., is lead scientist and director of research at Lyvi. He is also a nonresident fellow in the Indo-Pacific Security Initiative at the Atlantic Council’s Scowcroft Center for Strategy and Security, and a senior research fellow at Associated Universities, Inc. His book on Chinese megaprojects and coalition politics in Southeast Asia is in production at Cornell University Press.
**Please note, as a matter of house style, War on the Rocks will not use a different name for the U.S. Department of Defense until and unless the name is changed by statute by the U.S. Congress.
Image: NASA/GSFC/METI/ERSDAC/JAROS and U.S./Japan ASTER Science Team via NASA.
