One Missile, One Plant: 70% of the World's PCB Resin Offline

Table of contents

The Single Point of Failure Nobody Talks About

Every AI server needs a printed circuit board. Every GPU needs a PCB. Every smartphone, every laptop, every piece of networking equipment, every EV controller, every medical device. PCBs are the substrate on which the entire electronics industry is built.

The high-performance PCBs used in AI servers, 5G infrastructure, and automotive radar require a specific material: high-purity polyphenylene ether resin. PPE resin gives the laminate the electrical properties needed at high frequencies. Standard FR-4 epoxy works for consumer electronics. It does not work for the hardware running inference at hyperscale.

SABIC, the Saudi Arabian Basic Industries Corporation, supplies approximately 70% of global high-purity PPE resin for PCB laminates.

One company. One facility. Seventy percent of a critical material for the global electronics supply chain.

On April 7, 2026, Iranian ballistic missiles struck SABIC's petrochemical complex at Jubail Industrial City.

What Happened

The strike was part of a broader Iranian attack on Saudi infrastructure, described by the IRGC as retaliation for strikes on Iran's Asaluyeh petrochemical facilities. Missile debris ignited a fire at the heart of SABIC's Jubail complex, which sits at the center of Saudi Arabia's $69 billion downstream petrochemical empire.

Production of high-purity PPE resin halted immediately. As of late April 2026, SABIC had not resumed output. No public timeline for restart has been provided. The facility remains offline in an active conflict zone, with Gulf shipping severely disrupted by the wider war.

This is not a temporary blip. This is a structural supply shock to a material that has no rapid substitute at the purity levels required for advanced PCB laminates.

The Cascade

The numbers moved fast.

PCB prices surged 40% in April alone, according to Goldman Sachs analysts. Copper foil, another critical PCB input, was already up roughly 30% year to date before the strike. Lead times for key chemical inputs like epoxy resin expanded fivefold, from three weeks to fifteen.

The global PCB market crossed $100 billion in 2026, driven overwhelmingly by AI server and high-performance computing demand. Hyperscale operators deployed approximately 1.2 million AI-optimized servers in 2025 alone, each integrating 8 to 16 GPU accelerators. The average AI server PCB layer count rose from 18 in 2023 to 32 in 2025. These are not simple boards. They require the exact class of high-frequency laminate material that SABIC produces.

PC makers and system integrators reported surging prices across PCBs, plastic materials, and passive components. AMD and Intel CPU shortages worsened as manufacturers allocated more capacity to server CPUs for AI data centers. TSMC's CoWoS packaging capacity remained fully allocated through mid-2027.

And underneath all of it, the PPE resin supply that feeds the laminate that feeds the PCB that feeds the server that runs the AI was sitting in a bombed-out plant on the Saudi coast.

The Concentration Problem

This is the story the tech industry does not want to tell.

The entire AI infrastructure buildout depends on a supply chain that concentrates critical materials in single points of failure. SABIC at 70% of PPE resin is the most dramatic example, but it is not the only one.

Ras Laffan Industrial City in Qatar, which supplies roughly 30% of global semiconductor-grade helium, was also disrupted by the wider conflict. Helium is used in semiconductor fabrication for cooling and leak detection. Without it, chip production slows.

The neon supply chain runs through Ukraine. Palladium comes primarily from Russia. Gallium and germanium are dominated by China. Every critical material in the semiconductor and electronics supply chain has a geographic concentration that a single geopolitical event can disrupt.

The PCB resin crisis is not an aberration. It is the pattern revealing itself.

What This Means for AI

The timing is brutal. The AI industry is in the middle of the largest infrastructure buildout since the construction of the internet backbone. Every major cloud provider is building data centers as fast as they can source the hardware. Nvidia cannot produce GPUs fast enough. TSMC cannot package them fast enough. And now the PCBs that the GPUs mount on are constrained by a material shortage caused by a war 8,000 miles from Silicon Valley.

Every AI server that does not get built because a PCB could not be manufactured is inference capacity that does not come online. Every month of delay in GPU production ripple-effects through model training timelines, product launches, and competitive positioning.

The companies that pre-ordered materials, locked in long-term supply agreements, and diversified their PCB sourcing before April are fine. The companies that were running just-in-time procurement on a $100 billion market with a 70% single-supplier dependency are scrambling.

The Uncomfortable Question

The tech industry spent the last three years building AI as fast as humanly possible. Billions in GPU purchases. Tens of billions in data center construction. Hundreds of billions in model training compute.

Almost none of that investment went into securing the physical supply chain that makes the hardware possible.

Nobody asked what happens when the plant that makes 70% of a critical PCB material gets hit by a missile. Nobody stress-tested the supply chain for a scenario that, in hindsight, was entirely predictable given the geopolitical environment.

The AI arms race was built on an assumption that hardware would always be available if you could pay for it. That assumption broke on April 7.

PCB prices are up 40%. Lead times are up 5x. The plant is still offline. And the war is not over.

The next time someone talks about AI scaling, ask them where their PCB resin comes from.

Stay in the loop New teardowns and guides, straight to your inbox. No spam.