Inside Silver's Industrial Engine: Solar, EVs, and AI Data Centres

Silver's industrial uses hit 680 Moz in 2024, led by solar, EVs, and AI data centres. The full breakdown of why the structural deficit keeps widening.

Published May 21, 2026 · 9 min read

Key Takeaways

• Silver's industrial demand hit a record 679.0 Moz in 2024 — roughly 59% of total demand — before easing 3% to 657.4 Moz in 2025, as record prices cut into photovoltaic offtake (World Silver Survey 2026).
• Solar PV alone consumed an estimated 232 Moz of silver in 2024, up from 140 Moz in 2022. At ~20g per panel and no commercially viable substitute in current cell architectures, solar demand scales directly with every gigawatt installed.
• Electric vehicles use 25–50g of silver per car across battery management systems, high-voltage contactors, and power electronics. At 17+ million EVs produced globally in 2024, automotive has become a locked-in, growing structural demand floor.
• AI data centre infrastructure has emerged as a material new demand vector: GPU-dense server racks are silver-intensive at the PCB, connector, and thermal management level. Analysts estimate AI infrastructure added 10–15 Moz of incremental demand in 2024 alone.
• The substitution argument does not hold at commercial scale: aluminium alternatives in solar cells reduce power conversion efficiency by 5–15%, making them uneconomic at scale. Mine supply grew just 8% from 2020 to 2024 while industrial demand grew 22% — the gap is structural, not cyclical.

Most coverage of silver as an investment centres on the gold-silver ratio, the supply deficit, and what central banks are buying. The industrial demand story — the engine actually driving that deficit — gets far less analytical attention. That is a mistake. Silver's industrial demand is now so large, so structurally embedded in the global energy transition, and so resistant to substitution that it has fundamentally changed the metal's character from a volatile commodity into something closer to a strategic industrial resource that also trades as a monetary asset. Understanding exactly where the 680+ million ounces consumed industrially each year actually goes — and why that number will keep growing — is the most important due diligence a silver investor can do. All prices in this article are quoted in US dollars (USD) per troy ounce.

679.0 Moz Silver's Record Industrial Demand — 2024 (World Silver Survey 2026). Up from 511.9 Moz in 2020 — a 33% increase in four years, and about 59% of total silver demand in 2024, making it the dominant driver of the structural supply deficit. It eased to 657.4 Moz in 2025 as record prices cut into photovoltaic offtake.

The Demand Breakdown: Where Silver Actually Goes

Total global silver demand runs at approximately 1,200 Moz per year, against mine and recycling supply of roughly 1,030 Moz — producing a structural deficit of ~190 Moz annually. Industrial applications dominate on the demand side by a wide margin. The table below shows the full demand picture for 2024.

Solar PV: The Largest and Fastest-Growing Consumer

Solar photovoltaic panels are now the single largest industrial consumer of silver on the planet — consuming an estimated 232 Moz in 2024, up from 140 Moz in 2022 and just 85 Moz in 2019. The driving force is straightforward: global solar installations have set records every year. The International Energy Agency confirmed that 2024 saw the addition of approximately 593 gigawatts (GW) of new solar capacity globally — the largest single-year addition in the history of the energy transition. China alone installed over 300 GW. Each gigawatt of solar capacity requires approximately 380–420 tonnes of silver, meaning 2024's installation record consumed roughly 225–250 Moz of silver for solar alone.

593 GW New Solar Capacity Added Globally in 2024 — A Record (IEA). At roughly 400 tonnes of silver per GW of solar capacity, 593 GW of new installations required approximately 237,000 tonnes (~232 Moz) of silver — equivalent to ~23% of annual global mine supply consumed by a single end use.

The silver in a solar panel is not incidental — it is functional. Silver paste is screen-printed onto the photovoltaic cells to form the busbars and fingers that collect and conduct electricity. Silver's electrical conductivity (the highest of any element at room temperature) and its resistance to corrosion make it optimal for a cell that must operate outdoors for 25–30 years in all weather. The question silver bears most frequently raise is: can silver be substituted in solar cells? The answer at commercial scale in current cell architectures is no — at least not without a meaningful efficiency penalty.

The Substitution Problem in Solar Cells

The primary proposed substitute for silver in solar cell metallisation is aluminium, which is far more abundant and dramatically cheaper. The problem is efficiency. Aluminium has lower electrical conductivity than silver — roughly 60% — which means more energy is lost in transit from cell to inverter. In practice, substituting aluminium for silver in standard PERC (Passivated Emitter Rear Cell) and TOPCon (Tunnel Oxide Passivated Contact) architectures reduces the cell's power conversion efficiency by approximately 5–15 percentage points. For a solar developer financing a utility-scale project, that efficiency reduction translates directly into fewer megawatt-hours per dollar of capital expenditure — making the project materially less economic. Industry trials of silver-reduced pastes have achieved some reduction in silver intensity per panel (from ~20g toward 15–16g), but full substitution remains technically and economically impractical for mainstream cell designs.

The solar demand trajectory is not slowing. The IEA's 2025 forecast puts annual solar installations exceeding 800 GW by 2027 under its stated-policies scenario — more than doubling 2022's installation pace in five years. The EU, India, the US (Inflation Reduction Act), and China all have mandated solar targets requiring sustained high installation rates through the end of the decade. Under the current silver-intensity curve, global solar demand for silver could exceed 300 Moz per year by 2027 — roughly 30% of current annual mine supply consumed by a single end use.

Electric Vehicles: Locked-In Demand Scaling Rapidly

Electric vehicles use 25–50 grams of silver per car, compared to approximately 15–28 grams in a conventional internal combustion engine vehicle. The higher silver content in EVs is not incidental — it reflects the fundamental architecture of an electric drivetrain. The battery management system, onboard charger, power conversion electronics, rapid-charging contactors, and multiple motor control units all require silver-containing components: silver-coated contacts in high-voltage switching circuits, silver-brazed connections in battery cell assemblies, silver-filled thermally conductive compounds in power electronics, and silver-paste circuit boards throughout. Global EV production ran at approximately 17.1 million units in 2024 — meaning EVs consumed roughly 425–850 tonnes (~14–27 Moz) of silver in automotive electronics alone that year, separate from the broader automotive sector.

The important distinction in automotive silver demand is that EV growth adds to total automotive silver consumption rather than replacing it. ICE vehicle production has not collapsed at the pace EV bulls predicted in 2021–2022 — global conventional vehicle production remains at approximately 70 million units per year, each consuming 15–28g of silver. The net effect is that total automotive electronics silver demand has been growing structurally: more silver per vehicle on the EV side, sustained conventional vehicle volumes on the ICE side, and the overall mix shifting toward the higher end of the per-vehicle silver consumption range. By 2030, at current adoption trajectories, EVs could represent 25–35 million units of global production — consuming in excess of 50–80 Moz of silver annually from the automotive sector alone.

AI Data Centres: The Newest Structural Demand Driver

The emergence of AI as a major infrastructure investment theme since 2023 has added a demand vector to silver that was not in the market's projections even three years ago. GPU-dense server racks — the hardware foundation of large language model training and inference — are among the most silver-intensive computing architectures ever deployed at scale. A single AI server rack contains hundreds of printed circuit boards, each manufactured with silver-bearing solder pastes; thousands of silver-plated connectors and contact surfaces; silver-coated cooling components including vapour chambers and heat spreaders in high-power GPU modules; and silver-brazed interconnects in high-current power delivery systems. The cumulative silver content per rack is not trivial — and the AI infrastructure build-out is deploying hundreds of thousands of such racks globally.

10–15 Moz Estimated Incremental Silver Demand from AI Data Centre Infrastructure — 2024. Industry analysts estimate the AI data centre build-out added 10–15 Moz of incremental silver demand in 2024 as hyperscalers deployed GPU infrastructure at unprecedented scale. This figure is expected to grow as announced $300B+ capex plans are executed through 2025–2026.

Microsoft, Google, Amazon, and Meta collectively announced data centre capital expenditure plans exceeding $300 billion for 2025 alone. Nvidia's GPU clusters deploy in racks of 8 or more GPUs per server, with hundreds of servers per rack and thousands of racks per campus-scale data centre. The silver content in a single large AI training cluster is estimated by materials researchers at several hundred kilograms — not per server, but per full-scale deployment. At the pace of the current AI infrastructure cycle, the cumulative silver demand is material and growing into a market that was not modelling it at all in prior supply-demand forecasts.

5G Infrastructure: A Sustained Demand Source Already Running

5G network deployments — still in full rollout globally as of 2026 — are silver-intensive at every layer of the radio access network. Massive MIMO antenna arrays at 5G base stations use silver-plated copper elements for RF performance; switching and routing equipment in 5G core networks contains silver-bearing PCBs and connectors; and the consumer devices connecting to 5G networks each contain silver-coated contacts and connectors. The GSMA estimates global 5G deployments will require the installation of 15–20 million small cell antennas and macro sites by 2030 — each site silver-intensive in its antenna arrays, backhaul equipment, and power systems.

Electronics and Electrical: The Stable Demand Foundation

Beyond the high-growth sectors, silver has a broad and stable industrial base in traditional electronics and electrical applications — approximately 220 Moz in 2024. This covers semiconductors (silver bonding wire in chip packaging), consumer electronics (silver contacts in switches, displays, and keyboards), and industrial electrical equipment (circuit breakers, switchgear, and relay contacts, where silver's arc-resistance and conductivity are essential). These are not growth drivers, but they represent a stable demand floor that is unlikely to contract materially: the use of silver in electrical contacts is determined by the physics of electrical conductance and arc suppression, for which copper and aluminium are inferior substitutes in high-cycle, high-reliability applications. A relay contact that must open and close millions of times without welding shut requires silver. Aluminium simply fails at this task.

Brazing Alloys, Medical, and Water Purification

The remaining ~158 Moz of industrial demand is distributed across applications that are individually smaller but collectively important and durable. Silver brazing alloys — used to join dissimilar metals in HVAC systems, aerospace components, and industrial piping — are a price-inelastic demand source: the structural integrity of the brazed joint depends on the alloy composition, and the cost of silver in a single joint is negligible relative to the cost of failure. Medical applications include silver-coated wound dressings (silver's antimicrobial properties are exploited in advanced medical textiles), silver-ion water purification systems used in hospitals, aircraft water systems, and swimming pools, and silver-containing dental alloys. These applications will not double the way solar has — but none are going away, and several are growing as healthcare infrastructure expands in emerging markets.

Why Supply Can't Keep Up — And Won't

Understanding that industrial silver demand now runs around 660–680 Moz a year — a record 679.0 Moz in 2024, easing to 657.4 Moz in 2025 as record prices bit into photovoltaic offtake — is only half the equation. The other half is why supply cannot respond, even as prices have been significantly above long-run averages. The fundamental constraint is structural: more than 70% of silver mined globally is a byproduct — it comes out of copper, zinc, lead, and gold mines as a secondary product. The decision to mine more silver requires the primary metal's economics to justify a new mine, and takes 10–15 years from exploration to production. No amount of silver price appreciation makes that faster. New dedicated silver mines are rare because silver grades high enough to justify a standalone project are uncommon — and the three countries that dominate global production (Mexico, Peru, and China, together ~50% of supply) add geographic concentration risk on top of the structural inelasticity. The balance below is the Metals Focus series as published in World Silver Survey 2026 (April 2026); total supply there includes net hedging supply and official-sector sales, so it runs a little above mine production plus recycling.

What the Industrial Demand Story Means for Price

Silver trades at approximately $82.40 per troy ounce as of May 21, 2026 — up from $74.55 in late March but still roughly 32% below its all-time high of ~$121, reached on 29 January 2026. The industrial demand story does not tell you precisely when silver will revisit those highs — no fundamental analysis can do that with precision. What it does tell you is three things about the structural setup. First, demand is growing from sectors that are policy-mandated, economically self-reinforcing, and largely price-insensitive: a solar developer does not cancel a project because silver costs $80/oz instead of $30/oz. Second, supply cannot respond meaningfully within any 3–5 year investment horizon because of the byproduct mining constraint. Third, the deficit being filled by above-ground stocks is drawing down a finite buffer — and history shows that when that buffer reaches critical levels, price moves sharply and quickly.

• Solar demand keeps growing: 800+ GW of annual installations expected by 2027 versus 593 GW in 2024. At current silver intensity, that implies 280–320 Moz consumed by solar alone — 27–31% of annual mine supply from a single end use.
• EV production is a long-runway driver: conservative projections see EV production at 25–35 million units by 2030, adding 50–80 Moz of annual automotive demand against today's ~14–27 Moz — a multi-year structural step-up.
• AI infrastructure is an accelerant not yet in most models: hyperscaler capex plans are multi-year and are not price-sensitive to commodity inputs. The AI demand vector is nascent but material and growing — and most silver supply-demand forecasting models were built before the AI infrastructure cycle began.
• Mine supply is structurally capped: world mine production grew from 790.3 Moz in 2020 to 846.6 Moz in 2025 — a 7% increase over five years, against a 28% increase in industrial demand over the same period. There is no credible scenario where mine supply grows fast enough to close that gap within this decade.
• Recycling is not the cavalry: silver recycling is growing modestly — from 181.5 Moz in 2020 to a 12-year high of 197.6 Moz in 2025, under 9% over five years. But the majority of silver consumed industrially (particularly in solar cells, electronics, and brazing) is consumed at concentrations too low to economically recover. Most of it does not return to the market.

How to Think About This as an Investor

The industrial demand story argues for patience and position sizing over timing. The structural deficit is real, growing, and persistent — but silver is also one of the most volatile commodity markets globally, capable of falling 30–40% in months even within a multi-year bull context. The industrial thesis does not insulate you from that volatility; it is the reason to hold through it. The investors who captured the full 2025 move from $34 to $121 understood the supply-demand imbalance long before $100 was a conceivable price target — and sized their positions to survive the drawdown from $121 to $74. The thesis has not changed: solar installations are hitting new records each year, EV production is scaling, AI infrastructure is being built at a pace the market did not anticipate, and the mine supply response is structurally limited. The above-ground stocks filling the annual deficit are finite. Use the portfolio builder to model how a silver allocation fits your current precious metals exposure at today's live spot prices.