Ultimately, the demand for AI services will determine the scale of data center investment. As long as demand continues to grow, data center construction will eventually proceed, but rising costs and project delays may slow the expansion and create localized supply gaps, while the risk of large-scale oversupply is relatively low.
The US midterm elections have further increased uncertainty surrounding new projects. Texas Governor-elect Gina Hinojosa has pledged to suspend approvals for new data centers until the state legislature enacts community protection measures. If such policies are implemented, some new projects may face slower approvals or even halts. Against this backdrop, projects that have already obtained permits, secured power access capacity, and signed long-term power and equipment procurement contracts may further enhance their scarcity and competitive advantage, forming new industry moats.
Hyperscale cloud service providers such as Microsoft, Meta, Amazon, and Google are expected to benefit from the increased value and pricing power of their existing data center capacity. Furthermore, if capital expenditure expectations for 2028 are lowered, the free cash flow of large cloud vendors is expected to improve, and valuations previously pressured by excessive AI investment may recover.
