AI Boom Reshapes Gold Demand in Electronics

MONTREAL (Scrap Monster): Artificial intelligence is creating new demand for gold in advanced processors, memory, networking equipment and data-centre infrastructure, but the rapid expansion of AI hardware has yet to produce a comparable increase in overall industrial gold consumption.

Global gold demand from technology applications — including electronics, other industrial uses and dentistry — totaled 322.8 tonnes in 2025, broadly unchanged from 326.2 tonnes a year earlier, according to the World Gold Council.

Electronics remained by far the largest industrial use, consuming 270.4 tonnes of gold during the year.

AI Is Increasing Gold Use in Advanced Electronics

Gold remains important to electronics because it combines high electrical conductivity with resistance to corrosion and oxidation.

It can also be formed into extremely thin wires, films and coatings without losing the performance characteristics required in high-reliability applications.

These properties make gold valuable in microprocessors, memory chips, connectors, sensors, communications equipment and other electronic components where long-term reliability is critical.

The rapid build-out of artificial-intelligence infrastructure has strengthened demand for many of those applications.

AI servers require increasingly powerful processors, memory systems and high-speed networking hardware, while hyperscale data centres are deploying larger quantities of advanced electronics to handle growing computing workloads.

Why AI Has Not Yet Produced a Gold Demand Surge

Despite the enormous growth in AI hardware, overall technology demand for gold was essentially unchanged in 2025.

The World Gold Council describes the market as a tug-of-war between two opposing forces.

On one side, AI is increasing demand for sophisticated, high-performance electronic components. On the other, high gold prices and continuing advances in manufacturing are encouraging companies to reduce the amount of gold used in each device.

Manufacturers have increasingly adopted thinner bonding wires and coatings and, where performance requirements permit, substituted less expensive materials.

Miniaturization is another factor. Smaller chips and more compact component designs can reduce the amount of gold required per unit even as the number and complexity of devices continue to grow.

Electronics Used 270 Tonnes of Gold in 2025

The figures show why AI growth has not translated directly into a large increase in industrial gold consumption. Stronger use in advanced electronics has been offset by weaker demand in some traditional applications and by continuing efforts to reduce gold content.

2026 Data Shows AI Beginning to Add Support

The first-quarter 2026 data show a somewhat stronger AI effect.

Total technology demand reached 81.6 tonnes during the quarter, compared with 80.4 tonnes a year earlier.

Electronics demand increased to 69.3 tonnes, the strongest quarterly level since late 2021.

The World Gold Council said expanding AI infrastructure was a key contributor, particularly for high-reliability chips, memory, wireless components and advanced power modules.

At the same time, manufacturers of more price-sensitive consumer electronics continued trying to reduce or replace gold because of elevated metal costs.

Co-Packaged Optics Could Create a New Gold Market

One emerging technology attracting attention is co-packaged optics, or CPO.

Next-generation AI systems need to move enormous volumes of data between processors, memory and networking equipment. Traditional electrical interconnections are increasingly challenged by bandwidth limits, power consumption and heat generation.

Co-packaged optics brings optical communications components much closer to the processor, potentially increasing bandwidth while reducing transmission bottlenecks and energy consumption.

Gold can play several roles within these systems, including in wires, coatings, optical and photonic components and high-frequency electrodes.

Major semiconductor and networking companies are investing in CPO development, but the technology remains at an early stage of commercial adoption.

If it becomes widely deployed across hyperscale AI infrastructure, the World Gold Council believes it could become a small but meaningful additional source of electronics demand for gold.

Gold Thrifting Is Reaching Technical Limits

The electronics industry has spent decades reducing the amount of gold used in individual components.

Bonding wires and coatings have become progressively thinner, while gold has been removed entirely from some lower-cost applications.

But the World Gold Council notes that much of the easiest gold reduction has already taken place.

Many existing wires and coatings are now close to their technical minimum thickness, limiting how much further manufacturers can reduce gold without affecting performance.

High-End AI Chips Still Need Reliability

This distinction is particularly important in AI applications.

Advanced AI processors and networking devices are expensive and operate under demanding conditions where component failure can have substantial consequences.

In those applications, manufacturers may place greater emphasis on reliability and longevity than on minimizing the cost of relatively small quantities of gold.

That is why gold remains difficult to replace in high-end and safety-critical devices even as manufacturers continue searching for alternatives in less demanding products.

Gold Remains Important Beyond AI

Gold's industrial role extends beyond computing and telecommunications.

The metal is used in aerospace systems, optical applications, chemical catalysis and healthcare technologies.

Medical diagnostics are one example. Tiny gold particles are used in large volumes of diagnostic tests where their physical and chemical properties help deliver rapid and reliable results.

These applications account for a relatively small share of total global gold demand but highlight why gold's industrial value cannot be measured solely by tonnage.

Market Read

The AI boom is clearly supportive for gold demand in electronics, but it has not yet produced a broad industrial-demand surge.

The reason is straightforward: more AI hardware is being built, but manufacturers are also using gold more efficiently and eliminating it from applications where cheaper substitutes can perform adequately.

The more important long-term question is whether expansion in high-performance AI infrastructure — including advanced processors, memory, networking and co-packaged optics — eventually grows faster than manufacturers can reduce gold use per component.

For the recycling industry, the development is also worth watching from the opposite end of the lifecycle. Today's investment in sophisticated, gold-bearing computing hardware will eventually become tomorrow's high-value electronic scrap.