Chaos and Order

💡 왼쪽 원문을 읽으면서 오른쪽에 따라 써보세요. Tab 키로 힌트를 받을 수 있습니다.

원문 렌더가 준비되기 전까지 텍스트 가이드로 표시합니다.

Introduction: The Center of Gravity in the AI Rally Is Shifting

AI remained the protagonist of markets in the first half of 2026. Nvidia reportedly crossed a market capitalization of 5 trillion USD for the first time ever, and the race to build data centers shows no sign of cooling. Yet in early June 2026, the semiconductor sector swung sharply in a single session, with the Nasdaq reportedly falling roughly 4 percent and about 1 trillion USD in market value evaporating. Nvidia and Micron then rebounded by around 5.6 percent, underscoring how violent the volatility had become.

Amid this turbulence, more and more investors are asking the same question. If chip valuations feel stretched, is there another way to participate in the AI boom?

One answer lies in physical infrastructure, especially power grids and commodities. No matter how fast a GPU runs, a data center does nothing without electricity. To generate and move that electricity, you need power plants, transmission lines, transformers, and the essential material behind all of them: copper. AI is, at its core, an electricity-hungry industry, and electricity is built from raw materials.

This article analyzes how the broad current of the AI power buildout affects power grids, copper, and the commodity cycle. It lays out both the bull and bear cases, then closes with a list of risks and checkpoints.

> This article is analysis for informational and educational purposes only. It is not investment advice and does not recommend buying or selling any specific security or asset. All investment decisions and their outcomes are solely your own responsibility, and you should consult a qualified professional if needed.

Core Analysis 1: Just How Much Electricity Does AI Consume

Start by sizing the demand. The International Energy Agency (IEA) has said it expects data center power consumption to rise rapidly. The IEA reportedly projected that global data center electricity demand could exceed 1,000 terawatt-hours (TWh) around 2026, a figure comparable to the entire annual electricity consumption of Japan.

The shift is large even when you look only at the United States. Multiple institutions have projected that data centers, currently around 4.4 percent of total US electricity use, could rise to between 12 and 20 percent by around 2030.

Conceptual Flow of Data Center Power Demand

Below is a simplified flow of how AI power demand translates into demand for physical materials.

Explosion in AI training/inference

GPU clusters expand ----> power consumption surges

New data centers built

+----+--------------------+

| |

v v

Secure generation Expand T and D infrastructure

(nuclear/gas/renewables) (lines/transformers/switchgear)

| |

+-----------+-------------+

Commodity demand rises

(copper, aluminum, steel,

rare earths, uranium)

The key takeaway from this simple chain is that the more AI demand grows, the more structurally the commodity demand at the bottom of the chain expands.

Comparing Power Sources for Data Centers

| --- | --- | --- | --- |

The widely reported case of Microsoft agreeing with Constellation Energy to restart the Three Mile Island nuclear plant under a roughly 20-year power purchase agreement shows just how aggressively Big Tech is moving to lock in stable, carbon-free electricity.

Core Analysis 2: Why Copper — The Metal of Electrification

Among commodities, copper is special, because it is the most widely used metal for carrying electricity. Wires, transformer windings, internal data center cabling, and power distribution equipment all rely on copper across nearly the entire electrical infrastructure.

The Structure of Copper Demand

Copper demand (conceptual)

==============================

Traditional demand New growth demand

------------------ -----------------

construction -> grid modernization

appliances/plumbing -> data center wiring

general industry -> EVs/charging infra

renewable generation

transformer/T expansion

The share of new growth demand

within the total pie is rising fast

Several miners and trading houses have argued that electrification (electric vehicles, renewables, and data centers) will drive a structural increase in copper demand over the coming years. Miner BHP and trading firm Glencore have reportedly published analysis suggesting that, over the long run, copper supply may struggle to keep pace with demand.

A Simple Sketch of Copper Price Direction

Below is a conceptual representation of the broad direction copper prices have been discussed in recent years. It is not actual data, only a sketch to convey the flow rather than precise levels.

Copper price direction (conceptual, not actual data)

High | . *

| . *

| . * swings

| . *

Low |. *

+--------------------------------> time

pandemic low recovery electrification hope

Copper has at times reportedly traded above roughly 10,000 USD per metric ton. That said, prices swing widely with the Chinese economy, the value of the dollar, mine supply disruptions, and many other variables.

Where a Unit of Copper Goes

| Use | Copper intensity | AI relevance |

| --- | --- | --- |

| One electric vehicle | About 3 to 4 times an internal combustion car | Indirect (electrification trend) |

| Data center wiring | Very high | Direct |

| Transmission/distribution grid | High | Direct (moving power) |

| Transformers | High (winding material) | Direct |

| Renewable generation | High | Indirect (power diversification) |

Core Analysis 3: Transformers and Wires — The Invisible Bottleneck

The most frequently cited bottleneck in power infrastructure is the transformer. Transformers are the essential equipment that converts electricity generated at plants to the right voltage for transmission and distribution. Yet transformers take a long time to manufacture, and the supply of their core materials, grain-oriented electrical steel and copper windings, is limited.

Multiple outlets have reported that lead times (the time from order to delivery) for large transformers in the United States and Europe have lengthened significantly. Some large transformers have reportedly faced delivery waits of several years. Such bottlenecks slow the pace of grid expansion and, at the same time, draw attention to transformer manufacturers and the suppliers of their core materials.

The Power Grid Value Chain at a Glance

[generation] -> [step-up transformer] -> [transmission] ->

[substation] -> [step-down transformer] -> [distribution] ->

[data center/end user]

Each stage needs copper, steel, aluminum,

and electrical steel, with transformers

and transmission lines flagged as key bottlenecks

Building or fixing a power grid takes as much time and capital as building a chip fab. In the United States, replacing aging grids and clearing the queue of new connections (the interconnection queue) have been flagged as major challenges. All of this work consumes large volumes of copper, aluminum, steel, and electrical steel.

Core Analysis 4: The Commodity Cycle and Supply Constraints

One of the most important concepts in commodity investing is the cycle. When prices rise, new mine development and capacity expansion follow; when supply grows, prices fall again, and the loop repeats. But for a metal like copper, many analyses note that it takes more than a decade on average to discover and develop a new mine and reach actual production.

Why Supply Struggles to Keep Up With Demand

Demand signal rises -> prices climb

decision to develop new mine

(exploration -> permitting -> build -> production)

more than a decade

meanwhile demand has already grown further

=> potential for structural shortage

Because of this time lag, supply struggles to catch up quickly during phases of fast-rising demand. Some analysts describe this as a structural bullish factor for copper.

AI Buildout Relevance of Major Commodities

| --- | --- | --- | --- |

Uranium is especially interesting. As nuclear power regains attention as a stable source for AI data centers, analysts have noted that the demand outlook for uranium, the fuel for nuclear plants, has improved alongside it. That said, uranium prices are also highly volatile and exposed to geopolitical risk.

The Path From a Single Mine to Production

To understand commodity supply constraints, it helps to see the stages a single mine passes through.

Exploration : geology surveys, drilling, resource estimates

(years, low success rate)

Feasibility : economic assessment, mine plan

Permitting : environmental impact, community consultation

(a frequent source of delay)

Construction : mine, smelter, infrastructure build

(enormous capital outlay)

First output : early grade and recovery rates vary

Operation : a fight against falling grades and rising costs

Each stage requires capital, time, and the consent of local communities and regulators. Because this process is so long, it takes years for a price signal to translate into actual production growth. This is the root cause of the inelasticity of commodity supply.

Core Analysis 5: The Inflation-Hedge Angle

Commodities are traditionally cited as an inflation hedge, because as prices rise, the nominal prices of real assets tend to rise too. If the AI power buildout structurally lifts demand for copper and power infrastructure, that could act as a kind of demand-pull inflation factor.

A Simple Portfolio Comparison

| --- | --- | --- | --- |

Commodities are not always a good inflation hedge, however. In an inflation accompanied by recession (stagflation), weaker demand can actually push industrial metals like copper lower. The hedge effect varies by regime, and that is worth remembering.

The Bull Case: Seeing the Beneficiaries of the AI Power Buildout

The bull case can be summarized as follows.

1. Structural demand. AI data centers, electric vehicles, and renewables simultaneously lift demand for copper and power infrastructure. This is not a single theme but several mega-trends overlapping.

2. Supply constraints. New copper mines take more than a decade to develop, transformers have long lead times, and the expansion of core materials like electrical steel is slow. Supply cannot easily grow in the near term.

3. Policy support. US infrastructure spending, grid modernization policy, and national electrification targets underpin demand for infrastructure materials.

4. Diversification. For investors who find chip valuations stretched, commodities and infrastructure offer indirect exposure to the AI theme with a different risk profile.

5. Inflation hedge. As real assets, they may help defend nominal value during inflationary periods.

Bull-case proponents sometimes call this the picks and shovels of the AI boom, an analogy to the gold rush, where the people selling shovels often made money more reliably than those digging for gold.

The Bear Case: Watching for Overheating and Traps

The opposing view is plenty persuasive too.

1. Cyclical sensitivity. Copper is nicknamed Doctor Copper for its sensitivity to the economy. If the global economy, and Chinese property and manufacturing in particular, slows, prices can be capped even if AI demand is healthy.

2. Uncertain demand estimates. Forecasts for AI power demand vary widely by institution. If efficiency gains (more compute per watt) arrive quickly, the rise in power demand could be smaller than expected.

3. Already priced in. Because the electrification story is so widely known, expectations may already be substantially reflected in copper and related share prices.

4. Substitution and conservation. If prices rise too far, substitution toward aluminum and others increases, and efficiency naturally trims usage, a self-correcting adjustment.

5. Volatility and cycles. Commodities have large cycles and high volatility. Entering near a peak can mean enduring losses for a long time.

The unresolved debate over an AI bubble itself is especially worth noting. As with the early-June 2026 report of roughly 1 trillion USD evaporating in a chip selloff, when the broad AI theme shakes, its derivative themes of power and commodities can swing along with it.

Bull Versus Bear at a Glance

| Issue | Bull case | Bear case |

| --- | --- | --- |

| Demand | Structural, overlapping trends | Uncertain estimates, efficiency variable |

| Supply | Shortage possible from slow development | Capacity eventually expands as prices rise |

| Valuation | Attractive versus chips | Already largely priced in |

| Economy | Underpinned by policy | Doctor Copper, vulnerable to recession |

| Inflation | Real-asset hedge | Weak in stagflation |

Neither side can be declared absolutely right. The market sets prices through a tug-of-war between the two views.

Extended Analysis: Lessons From Historical Cycles

In commodity investing, the past does not guarantee the future, but cycle patterns tend to repeat. It is often cited that during the past supercycle (the phase of Chinese industrialization in the 2000s), copper and other industrial metals showed prolonged strength. Yet it is equally worth remembering that investors who entered near the peak in the latter part of that cycle had to endure years of price correction afterward.

The Four Typical Phases of a Supercycle

Phase 1: dormancy

demand signals exist but prices are quiet

supply is still ample

Phase 2: acceleration

demand overtakes supply

prices surge, media attention concentrates

Phase 3: overheating

speculative money flows in, volatility peaks

a flood of new supply decisions

Phase 4: correction

supply arrives just as demand slows

prices fall, the cycle resets

No one can declare with certainty which phase the AI power theme is in today. Bulls see it as still at the start of Phase 2; bears see signs of Phase 3 overheating already. What matters is acknowledging that judging the phase itself is hard, and not betting too heavily in one direction.

Comparing the AI Theme With Past Infrastructure Booms

| Dimension | Past case | AI power theme |

| --- | --- | --- |

| Demand trigger | Emerging-market industrialization | AI compute demand |

| Core materials | Copper, iron ore | Copper, uranium, electrical steel |

| Bottleneck | Mine supply | Transformers, grid, generation capacity |

| Main risk | Emerging-market slowdown | AI efficiency gains, bubble fears |

| Time horizon | More than a decade | Uncertain, ongoing |

This comparison is not a precise forecast but an organization of the structural similarities and differences between past and present. The differences, especially the pace of AI efficiency gains, are a variable that did not exist before.

Extended Analysis: Regional Power Grid Investment Flows

Grid expansion is not a US-only story. Each region carries its own drivers and constraints.

Comparing Grid Situations Across Major Regions

| --- | --- | --- | --- |

According to multiple reports, both the United States and Europe need large-scale investment to modernize aging grids and accommodate new demand. These investment flows spread demand across a broad supply chain of transformers, wires, cables, and distribution equipment.

That said, because regulatory environments and financing conditions differ by region, it is hard to assume that strong demand in one region immediately translates into a global price increase. Looking at regional flows together aids a balanced judgment.

Risks and Checkpoints

Here is a list of items worth monitoring when observing or reviewing this theme. The following is a checklist for analysis, not a recommendation.

Macro Checkpoints

[ ] Chinese economy and property indicators (a major axis of copper demand)

[ ] US dollar value (a strong dollar is a headwind for commodities)

[ ] Fed rate path (real rates and the cost of capital)

[ ] Global manufacturing PMI (a gauge of industrial demand)

Industry/Supply Checkpoints

[ ] News of production disruptions at major copper mines

[ ] Changes in transformer lead times

[ ] Supply situation for electrical steel and windings

[ ] Pace of clearing the interconnection queue

[ ] Trends in nuclear restarts and new contracts

Theme-Risk Checkpoints

[ ] Upward/downward revisions to AI power demand forecasts

[ ] Pace of progress in data center efficiency technology

[ ] Changes in AI capex guidance (Big Tech spending)

[ ] Signs of overheating and crowded positioning in commodity prices

Comparing Characteristics by Vehicle

| Approach | Characteristics | Caveats |

| --- | --- | --- |

| Individual miners | High leverage, high volatility | Single-company risk |

| Commodity ETFs | Diversified, accessible | Need to understand roll costs and structure |

| Utilities/power stocks | Relatively defensive | Sensitive to regulation and rates |

| Transformer/equipment stocks | Possible bottleneck beneficiary | Order-cycle volatility |

Each vehicle has a different risk and return profile. None offers guaranteed returns, and reviewing diversification and your own risk tolerance should come first.

Closing: Picks and Shovels, but with Care

AI is not merely a software revolution; it is an industry that demands enormous physical infrastructure. Behind the GPU is a data center, behind the data center is a power grid, and behind the power grid are copper, transformers, and raw materials. Following this chain reveals that the benefits of the AI boom do not stay confined to semiconductors.

That said, the commodities and infrastructure theme is a high-volatility space that swings sharply with the economic cycle and supply variables. The bull case for structural demand and the bear case for overheating and economic concerns both rest on valid grounds. What matters is not getting swept up in only one narrative, but steadily monitoring macro and industry indicators while keeping a balanced view.

> To emphasize once more: this article is analysis for informational and educational purposes only and is not a recommendation to buy or sell any specific security or asset. Market outlooks and figures are based on cited reporting and institutional materials and may change going forward. Investing carries the risk of loss of principal, and all decisions and responsibility rest with the investor. Please consult a qualified financial professional if needed.

References

- Reuters, coverage of global commodities and copper markets, [https://www.reuters.com/markets/commodities/](https://www.reuters.com/markets/commodities/)

- Bloomberg, energy and commodities section, [https://www.bloomberg.com/markets/commodities](https://www.bloomberg.com/markets/commodities)

- CNBC, markets and energy news, [https://www.cnbc.com/energy/](https://www.cnbc.com/energy/)

- Financial Times, commodities coverage, [https://www.ft.com/commodities](https://www.ft.com/commodities)

- The Wall Street Journal, markets section, [https://www.wsj.com/news/markets](https://www.wsj.com/news/markets)

- International Energy Agency (IEA), analysis of data centers and power demand, [https://www.iea.org/energy-system/buildings/data-centres-and-data-transmission-networks](https://www.iea.org/energy-system/buildings/data-centres-and-data-transmission-networks)

- Constellation Energy, investor information, [https://www.constellationenergy.com/](https://www.constellationenergy.com/)

- Yahoo Finance, copper futures and commodity quotes, [https://finance.yahoo.com/commodities/](https://finance.yahoo.com/commodities/)

- Nvidia, investor relations, [https://investor.nvidia.com/](https://investor.nvidia.com/)

- U.S. Energy Information Administration, electricity data, [https://www.eia.gov/electricity/](https://www.eia.gov/electricity/)