352
Productivity & Workflow350
Software Development246
Automation & Workflow238
Marketing & Growth203
Photography & Imaging164
AI Infrastructure & MLOps193
Writing & Content Creation192
Design & Creative171
Data & Analytics166
Customer Support155
Education & Learning141
Sales & Outreach131
Voice & Speech130
Operations & Admin119
Shares of companies that supply power, cooling, and cables for data centers have dropped sharply since June as investors worry about AI spending and rules.
In short: Investors have sold off shares in companies that help build and run data centers, wiping about $4.2 trillion in market value since June.
A group of manufacturing, utility, and mining companies tied to data center construction has lost $4.2tn in value since June, according to a Financial Times analysis. These firms are sometimes described as the AI “picks and shovels” companies, meaning they sell the tools and materials needed to build the facilities that run AI.
The drop has erased almost half of the $9.4tn these same companies gained in the prior 12 months. The companies hit include Caterpillar, which sells gas generators used to power data centers. Caterpillar’s market value fell by $214bn, or 44%, from its June 30 peak. Other examples include Oklo, a nuclear power company that fell 49% even after signing a contract with Meta in January, plus Comfort Systems (cooling and ventilation equipment) and Corning (fibre-optic cables), both down about half.
Some analysts say investors pushed prices too high by treating older industrial firms like fast-growing chip companies. Others say demand for data centers is still strong. US spending on data center construction, not counting computers and chips, reached $96bn in the 12 months to September, based on government data.
A key unknown is whether tougher local rules in the US will slow new data center projects. Some places have discussed moratoriums (a temporary pause), but researchers at Citi said widespread restrictions still look unlikely. Even if some areas say no, construction could move elsewhere, since companies still need places to run their AI systems.
Source: Financial Times