AI data centers face power and cooling bottlenecks

Behind the semiconductor supercycle lies a deeper infrastructure challenge: every megawatt of hyperscale AI data center capacity requires approximately 27 tonnes of copper for transformers, substations, power distribution, cooling systems, and high-capacity cabling. This realization is driving a secondary wave of capex demand across the energy, materials, and industrial-infrastructure complex. Softbank's recent multi-billion-dollar investment in AI data center battery systems signals that institutional capital is shifting attention from pure compute to the power ecosystem. Companies like Vistra, Oklo, and various battery-storage plays are seeing renewed interest from both institutional and retail investors.

The power constraint is no longer theoretical; hyperscalers are competing fiercely for grid access and reliable power sources. This has created opportunities for geothermal developers like Fervo Energy, which recently upsized its IPOInitial Public Offering - a company's first public sale of stock. target to $1.82 billion from $1.33 billion, signaling strong demand from large infrastructure funds. Energy-integration plays like FLNC, cooling specialists, and fuel-cell companies are attracting retail attention as traders recognize the scope of the build-out required. The narrative has broadened from 'AI chips are scarce' to 'AI infrastructure requires a complete ecosystem upgrade.'

Silver and copper face secular tailwinds from this shift, as both metals are essential to power transmission and cooling systems. The China sulphuric acid ban compounds the urgency: with refining capacity potentially constrained, miners and recyclers will face higher costs, further supporting metal prices. Mining companies are actively exploring new sources and partnerships to ensure supply meets the multi-year buildout timeline. This creates a favorable backdrop for mining equities and commodities, though the outcome depends on sustained AI capex growth and no major policy reversals.

The risk is that if AI infrastructure capex disappoints or delays materialize, demand for power-adjacent materials could evaporate. Additionally, a broader equity selloff or recession would eliminate the growth premium currently supporting energy and mining stocks. For now, however, the convergence of semiconductor strength, metal supply tightness, and power-infrastructure demand is creating a rare tailwind for the materials and energy complex.