After the storage track, copper and fiber optics may be the next markets to explode due to AI.

Charlie, a metal strategist from Citi, has been eager to make a trade in recent weeks: buying a digital call option for LME copper with a strike price of $15,250 expiring in August.

He believes that since 2022, almost all growth in copper demand has come from energy transition and AI-related sources.

As of May 14, 2026, LME three-month copper is approaching $14,000 per ton, with COMEX copper opening at $6.63 per pound. Copper has risen 41% over the past 12 months and 10% in the past four weeks. This price is near an all-time high.

Over the past two years, the entire market has framed the AI story as a chip story—Nvidia’s market cap, TSMC’s capacity, HBM yields, and CoWoS packaging bottlenecks. Almost all discussions about “AI infrastructure” have centered around those few square centimeters of silicon chips.

But from a perspective few are aware of, the demand from AI is sinking from silicon chips into copper mines and glass.

Copper is the Oil of the AI Era

Supply and demand determine prices; the notion of “strong demand for copper” can be seen more directly in the market.

LME three-month copper closed at $13,943 per ton on May 11, 2026, marking a historical high for LME close, with a single-day increase of 2.7%. COMEX copper reached a new intraday high of $6.58 per pound on May 12. Over the past 12 months, copper has risen 41% and 10% in the past four weeks.

At the beginning of 2025, copper was still around $9,000. By mid-year, it broke through $12,000, finishing the year with a 43% increase, the best year for copper since 2009. In January 2026, copper first broke $13,000 in intraday trading. Then, four months later, $14,000 is also close at hand. The shape of this curve resembles that of a rediscovered asset, being priced by a new logic.

Trafigura is the second-largest metal trader in the world. Its head of metal analysis, Graeme Train, provided a very concise breakdown of demand: one-third of the additional 10 million tons of copper consumption over the next decade will come from electric vehicles, one-third from power generation and transmission, and the remaining one-third from automation, manufacturing capital expenditures, and cooling systems for data centers.

Goldman Sachs provided a sharper assertion in a report titled “AI and Defense Place the Power Grid at the Center of Energy Security”: copper will become the oil of the AI era. Goldman Sachs estimates that by 2030, global grid and power infrastructure construction will contribute over 60% of the increase in copper demand.

This may sound exaggerated, but upon closer consideration, it is reasonable.

Copper's electrical conductivity is 100% IACS, second only to silver among all metals. However, silver is too expensive, making copper the only answer for almost all large-scale conductive scenarios in industry. The nearest alternative is aluminum, but its electrical conductivity is only 61% that of copper, meaning that for the same power delivery, aluminum wire requires a thicker cross-section, making it heavier, taking more space, and incurring more heat loss. In the confined space of data center cabinets, this difference is nearly unacceptable.

The same applies to thermal conductivity. Copper’s thermal conductivity is 401 W/(m·K), five times that of iron and eight times that of stainless steel. The power consumption of an NVIDIA GB200 single card is 1,200W, with a standard cabinet housing 72 cards, resulting in total cabinet power exceeding 130kW. At this level of thermal density, air cooling is insufficient, and liquid cooling must be implemented. Almost every component in the liquid cooling system that deals with “heat” involves copper cold plates, copper tubes, and copper water cooling heads.

In other words, copper is not the “preferred material” for AI data centers, but the “physically only choice.”

The power consumption of AI is disruptive, and delivering electricity to data centers is highly copper-intensive.

A 1GW AI data center requires nearly 27,000 tons of copper just for power distribution and cabling. The Hyperion data center that Meta is building in Louisiana has a scale of 5GW. Calculating just this one project’s copper demand approaches 135,000 tons, not including the high-voltage transmission lines, substations, and grid upgrades needed to deliver electricity to the data center.

In our past impressions, copper was a metal that was easy to obtain, but recent data suggests this impression might need some adjustments.

Since March 2026, the US-Iran conflict has disrupted the export of sulfur and sulfuric acid from the Middle East, which are key inputs for heap leaching copper refining, forcing Chilean smelters to reduce output. This has been the catalyst for the recent surge in 2026.

Moreover, structurally and macroeconomically, no super-large copper mines have been discovered globally in the past decade. John Meyer, an analyst at the British sell-side firm SP Angel, believes the breakeven price for developing the next generation of new copper mines is $13,000 per ton, which already exceeds current copper prices. The estimates from the team led by Wang Jiechao at CITIC Jiantou suggest that in 2026, global refined copper shortages will exceed 100,000 tons; Citi’s forecast is even more aggressive, at 308,000 tons.

2026 Fiber Optic "Great Famine"

The copper story thus far presents a clear bullish narrative. However, if you pull back to the internal perspective of AI data centers, you will discover something very subtle: part of the demand for copper is being substituted.

“The next generation of AI infrastructure will require a large amount of optical connections because the computational demand is growing rapidly to the extent that copper wires can no longer meet the needs.” This is the view expressed by Jensen Huang in an interview this month.

As Huang stated, the data transmission demand of AI clusters is breaking through the physical limits of copper cables.

Copper cables face two fundamental constraints for high-speed signal transmission: first, signal loss increases sharply with rising frequency; second, the size and weight of copper cables become unacceptable at high frequencies. The interconnect bandwidth between GPU clusters is advancing from 200G and 400G to 800G and 1.6T, while the distance copper cables can support has shrunk from several meters to several tens of centimeters. AI clusters operate at tens of thousands of nodes, often spanning multiple cabinets and, at times, across data centers, which is physically unmanageable with copper.

But fiber optics can do it.

This is why the wave of growth in fiber optics is stronger, purer, and more irreversible than that of copper. How exaggerated is the increase in fiber optics?

Data from CRU shows that the price of China's G.652D bare fiber increased by over 80% in three months from November 2025 to January 2026. The average price in January was 31.5 yuan per core-kilometer, with some actual transactions reaching the range of 40 to 50 yuan, accumulating an increase of 94% to 144%.

Fiber optics, an industrial product that has been stable in price for the past few years, doubled in price within three months.

By February 2026, high-end fiber optic types saw even steeper increases. For example, G.657.A bend-insensitive fiber rose from over 30 yuan per core-kilometer to over 50 yuan within a month. Sun Telecom directly declared a "Great Fiber Famine in 2026," with its G.652D priced at $2.20 per kilometer in 2024, rising to $3 in December 2025, and reaching $4.10 a month later. Overall, Asian fiber prices rose by 75%, reaching a seven-year high.

The demand from AI data centers for fiber optics represents a magnitude of disruption.

Rahul Puri, CEO of STL's optical network business, mentioned a figure that amazed me when I first saw it: the amount of fiber required for an AI data center is 36 times that of traditional CPU racks, which is a dramatic leap.

The operational mode of GPU clusters is completely different from that of CPUs. A large training cluster requires a non-blocking high-speed interconnect among all GPUs. This network structure is called a scale-out architecture, with bandwidth requirements that are unimaginable in the CPU era. Additionally, inter-data-center DCI links are needed to piece together computational power clusters distributed across different geographical locations into a supercomputer. For Meta’s Hyperion data center project, fiber optic demand has reached 8 million miles.

Returning to the economic theory that prices are determined by supply and demand, given the state of the demand side, what about the supply side?

Light Reading reports that at least one leading fiber optic manufacturer has completely sold out its inventory for the entire year of 2026. Data Center Dynamics reports that delivery times for large customers have extended to 20 weeks, and for smaller customers nearly a year.

Why can’t capacity be expanded? Because the expansion cycle for optical preforms, the core raw material for fiber optics, is 18 to 24 months and the process is extremely complex. Even if all manufacturers decided to expand production today, the earliest additional capacity wouldn't be available until the second half of 2027. During this period, demand will only continue to rise.

Anis Khemakhem, Chief Commercial Officer of Clearfield, provided an even more macro figure: by 2029, the United States alone will need an additional 213.3 million miles of fiber optics, doubling the existing 159.6 million miles to 372.9 million miles. Within six years, the national fiber stock will double.

The biggest winner in this story is Corning.

This is a glass company founded in 1851, known for making glass for Edison’s light bulbs, glass for television picture tubes, and Gorilla Glass for iPhones. Many people may not even know it is still operating. However, it is now a core fiber optic supplier for Meta, Nvidia, OpenAI, Google, AWS, and Microsoft. Its stock price has risen more than 75% over the past year. About the story of Corning, I may elaborate on it in a new article, so I won't discuss it further here.

The stories of copper and fiber optics seem to have just begun to attract public market attention, but I believe this may be the next exploding track after the storage sector.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。