Data-center power-cable demand is moving deeper into AI halls, with procurement priorities shifting toward medium-voltage feeder cables, DC distribution, termination accessories and aluminium alternatives. The market estimate in [S2] is not a substitute for project-level specification, but it provides a clear commercial context: global data-center power-cable spending is estimated at USD 1.74 billion in 2025 and projected at USD 4.72 billion by 2035, while AI campuses bring 15 kV to 35 kV feeders closer to the load. [S1] offers a separate, adjacent design signal: a solar-powered e-paper frame uses a magnetic modular perovskite panel, local-network image loading and a battery, illustrating how power-supply requirements can be reduced through architecture and energy harvesting; however, it does not specify a panel supplier, power output or efficiency and is not yet shipping.
1. Market backdrop: AI campuses are changing cable architecture
[S2] estimates the data-center power-cable market at USD 1.74 billion in 2025, USD 2.26 billion in 2026 and USD 4.72 billion by 2035, representing a 10.50% CAGR from 2025 to 2035. The 2026 estimate is stated as a 30.08% year-on-year increase, with a 2030 estimate of USD 3.67 billion. The forecast range is 8.00% in the low case and 13.00% in the high case, so buyers should treat the headline CAGR as a scenario rather than a guaranteed rate. The architectural driver is material: AI campuses are taking 15 kV to 35 kV feeders closer to the load, while copper prices, higher rack density and re-cabling of older halls increase the value of each megawatt deployed. For procurement, this shifts attention beyond rack-level cords and whips toward feeder-cable capacity, routing space, terminations, accessories and interfaces between medium- and low-voltage systems. [S2] also reports average power-cable content of about USD 115,000 per new MW in an air-cooled hall and USD 145,000 per MW in an AI hall at 2025 copper prices, providing a benchmark for evaluating whether an AI project’s cable content is materially above conventional deployment.
2. Product mix: medium voltage, DC cables and accessories
[S2] identifies low-voltage power cables as the largest product type, representing 40.4% of 2025 value, or USD 702.90 million. Medium-voltage feeder cables account for 25.4% and are projected to reach USD 1.21 billion by 2035. This is consistent with the movement of 15 kV to 35 kV distribution closer to data-hall loads, but the material does not provide medium-voltage market value for 2025 or a separate CAGR for that product type. Engineers should therefore avoid assigning a growth rate to medium voltage that is not stated in the source. DC power cables are the fastest-growing product type, with an 11.55% CAGR, attributed to battery strings, rack-level DC architectures and higher-voltage DC distribution. The segmented scope includes medium-voltage feeder cables, low-voltage power cables, power whips and rack power cords, DC power cables, and cable accessories and terminations. A complete bid package should cover these as a connected system rather than evaluating only bulk cable. As feeder voltage and distribution architecture move closer to the load, the technical and commercial importance of accessories and terminations rises with the cable itself, although [S2] does not quantify their market share or growth independently.
3. Conductor economics: copper remains dominant, aluminium is the faster-growth hedge
[S2] reports that copper conductors represented USD 1.39 billion, or 80.0% of 2025 market value. Aluminium grows faster, at a 13.43% CAGR, and is expected to reach 26.0% of spending by 2035. These figures support a dual-source strategy: copper remains the principal installed base and benchmark, while aluminium is the faster-growing alternative whose relative weight is expected to increase. For engineers, the decision cannot be made from growth alone. The source identifies copper prices as a market driver but provides no conductor-level pricing, conductor cross-section equivalence, voltage-class comparison, installation requirement or lifecycle data. Technical evaluation should therefore preserve copper as the baseline specification and require project-specific comparison of electrical performance, thermal behavior, jointing and termination methods, installation route, available supplier capacity and total installed cost. [S1] reinforces the importance of avoiding procurement assumptions where critical data are undisclosed: in the Inkanva product, the supplier, power output and efficiency of the perovskite panel are not named or disclosed. That discipline should apply to cable bids, where unverified performance or capacity claims should not substitute for documented data.
4. Geography, competition and project segmentation
[S2] places North America first with 42.0% of 2025 spending, equal to USD 729.96 million. Asia Pacific is the fastest-growing region at a 12.46% CAGR, and India leads large-country growth at 16.75%. The geographic scope also includes Europe, Latin America, the Middle East and Africa, with 12 country forecasts. This supports differentiated sourcing: global framework agreements can standardize core specifications, while regional approvals, local manufacturing content, labor conditions, logistics and currency exposure may affect award decisions; those additional commercial variables are not quantified in the material. The market is concentrated but not dominated by a single supplier: the top five suppliers account for about 38% of 2025 value. Buyers can use that structure to avoid dependence on one producer while still maintaining supplier depth. Segmentation by project type—new build, retrofit and re-cabling—and by data-center type—hyperscale and AI campuses, colocation, enterprise, and edge and telecom—matters because existing halls may require replacement and compatibility work rather than a greenfield package. [S2] notes that re-cabling of older halls is increasing cable value per megawatt, reinforcing the need to separate replacement, retrofit and new-build requirements in specifications and pricing.
5. Adjacent signal: energy-efficient distributed electronics and procurement implications
[S1] describes Inkanva’s Kickstarter color e-paper picture frame, which can charge from indoor light or sunlight through a flexible perovskite solar panel and can run without a power cable. The 7.09-inch and 13.3-inch models use a 1,200 x 1,600 display and carry the solar panel; the panel is modular and attaches magnetically. The 28.5-inch model has a 2,160 x 3,060 display but is not solar-powered and requires a power outlet. The frames include a 5,000 mAh battery, reported to last up to four months at one image change per day without the panel because e-paper draws power primarily when the image changes. Images can be loaded through a companion app or a browser on a local network without internet access, with up to 4,000 images in a slideshow, and the product supports Home Assistant and an API for AI agents. The Kickstarter campaign showed more than 1,900% of its funding goal, and shipment was planned for December 2026. The product was not yet shipping; its panel supplier, power output and efficiency were undisclosed, so it is not evidence of a proven perovskite supply chain. For power-cable buyers, the transferable lesson is architectural rather than product-specific: reducing cable demand through energy harvesting, local distribution and low-power interfaces is an alternative to adding cable and outlets, but each approach still requires verification of duty cycle, available energy, storage duration, connectivity, serviceability and failure modes.
Product encyclopedia: power cable.