AI-Ready Infrastructure: The Billions Being Spent on a New Digital Backbone
The growing demand for artificial intelligence (AI) infrastructure is driving a massive rebuild of digital systems around the world. Every AI model needs a robust platform to operate, with specific requirements for power, cooling, and connectivity that are far beyond those of traditional data centres.
This 'infrastructure upgrade' has accelerated over the past year as companies like OpenAI and Meta have signed leases for massive data centre campuses, some of which will require as much electricity as a mid-sized city. The facilities need to be connected to the grid through upgraded transmission lines and substations, a process that can take five years or more.
A key challenge is providing power that never cuts out, with companies like Pacific Gas & Electric announcing a $73 billion capital plan to upgrade transmission and distribution infrastructure in California alone. In the UK, National Grid has committed up to £35 billion to doubling transferable power by 2029 as data centre connection requests surge.
Subsea cable networks are also being upgraded to handle the vast amounts of data moving between training clusters, cloud regions, and end users. Companies like Amazon Web Services (AWS) and Meta have announced new cables with capacities of over 320 terabits per second, while Deutsche Telekom and NVIDIA opened their Industrial AI Cloud in Munich, an industrial-grade GPU cluster built around 10,000 NVIDIA Blackwell GPUs.