AI Capex Supercycle Extends Beyond Chips to Power and Cooling

The narrative around AI infrastructure investment is evolving beyond the obvious semiconductor plays into less-crowded segments of the value chain. As hyperscale data centers scale, the limiting factors have become clear: reliable power supply, efficient cooling, battery backup systems, and physical real estate for campus development. This recognition is driving capital rotation into companies positioned to solve these choke points.

SoftBank's recent multi-billion dollar investment in AI data center battery and power infrastructure signals institutional acknowledgment of this trend. Companies like Eos Energy Solutions (EOSE), Form Energy (FCEL), and Advanced Energy (BE) are capturing investor attention as the "next bottleneck" plays. Networking infrastructure firms like Cisco and optical component makers are also benefiting from renewed demand for high-capacity cabling and interconnect technology required to handle AI compute densities.

Real estate is another emerging vector. AI infrastructure campuses require contiguous land parcels with reliable power grid connections, creating demand for industrial and logistics properties in specific geographies. Land development and data center REITs are attracting capital as a result. Meanwhile, cooling technology providers are becoming critical; every megawatt of hyperscale infrastructure requires approximately 27 tonnes of copper for transformers, substations, and cooling systems, creating secondary demand pressures on commodity metals.

This broadening of the AI capex narrative has positive implications for small-cap and mid-cap companies with niche exposure to power, cooling, and real estate segments. However, it also increases the fragility of the overall thesis: if hyperscale capex guidanceCompany-issued forecasts of future financial performance. weakens, this second-order demand could evaporate quickly, creating a cascading selloff across the infrastructure value chain.