Hormuz disruption exposes AI’s hidden supply-chain risks

• Essential materials vulnerable
• Costs drive investment rethink
• Asian inventories depleted

The most pressing supply-chain risk with artificial intelligence is not advanced chips but the industrial materials and chemicals that underpin them, according to a new report.

Washington-based consultancy The Asia Group warned that prolonged disruption in the Strait of Hormuz is raising costs for industrial inputs needed to build AI infrastructure.

“AI may live in the cloud, but it’s very much built on physical supply chains,” said Kelly Magsamen, a co-author of the report and former chief of staff to US Defense Secretary Lloyd Austin.

The study highlights vulnerabilities across a range of industrial inputs, including copper, sulphuric acid and helium. These materials are essential for semiconductor manufacturing, batteries, data-centre construction and the wider equipment used to build AI systems.

Before the Iran conflict, about one-third of the world’s helium transited the strait, the report said, while nearly half of seaborne sulphur passed through the waterway.

Helium is critical for semiconductor manufacturing, while sulphur is an essential input for copper refining, magnets and electrical infrastructure.

The report, which assesses Asia’s exposure to the Hormuz disruption, said South Korea’s semiconductor industry produces the leading-edge memory products underpinning the global AI boom.

The country accounts for more than 60 percent of the global market, according to Invest Korea, its national investment promotion agency.

While Hormuz has long been viewed as the world’s most important oil and liquefied natural gas chokepoint, the report said its closure showed that it is also “a complex distribution centre for the global economy”.

Magsamen, a senior adviser with The Asia Group who served under President Biden’s administration, described sulphur as the “sleeper cell of the Hormuz crisis” because sulphuric acid is essential for refining copper, which is used in power grids, data centres and electric vehicles.

She said the strait had exposed vulnerabilities extending far beyond energy markets.

“This is not just an oil story. This is really about the inputs to all of the major economies in the world. This is an industrial supply-chain shock that’s going to continue to unfold over time and across many different sectors.”

Sulphur prices jumped during the Iran war as disruption to Qatari exports tightened supply. They have eased since shipping resumed through Hormuz, but remain above pre-conflict levels on lingering supply concerns and strong fertiliser demand.

Rather than causing an immediate shortage of AI chips, the report warns that higher costs could force companies to rethink where they invest their money.

Magsamen said she would be watching whether “higher input prices and inflation” begin to affect investment decisions, “especially around technology” and AI companies.

Further reading:

• MGX catapults Abu Dhabi to centre of global AI evolution
• Turkey’s $10bn AI strategy depends on reversing brain drain
• Hormuz closure threatens Gulf’s AI goals, warns OECD

One of the biggest risks, she said, was the opportunity cost – the trade-off between current spending on resilience and investing in future AI capacity.

Companies could divert capital towards inventories and alternative supply chains instead of expanding AI infrastructure. “When you’re investing in resilience you can’t necessarily focus on the build-out,” Magsamen noted.

Asia is likely to bear the brunt because of its concentration of advanced manufacturing, the report said.

Companies initially relied on stockpiles, expecting the disruption to be short-lived. Months later, those buffers are being depleted as the inventories many businesses built up after the Covid-19 pandemic begin to run down.

Kurt Campbell, chairman and co-founder of The Asia Group and a former US deputy secretary of state, said Asian businesses had largely exhausted inventories in a number of sectors, leaving “no more slack left in the system”.

For AI companies, that could mean fewer resources available for building new data centres, semiconductor capacity and other computing infrastructure.

“You’re going to see the impact on the AI sector unfold over time through trade-offs around investment,” Magsamen said. “AI relies on broad access to capital and certain physical components. Both have proven to be vulnerable in this context.”