What happened on 18 June: The US Federal Energy Regulatory Commission issued unanimous show-cause orders to six of America's largest regional grid operators — PJM, MISO, SPP, CAISO, ISO-NE, and NYISO — directing them to justify or revise their interconnection rules for large power consumers, specifically naming AI data centres as a "national priority." Grid operators have 30 days to report on available capacity. The move fast-tracks power connections that currently take years under standard grid interconnection queue processes.
Why FERC moved now
The FERC action stems from a rulemaking effort that Energy Secretary Chris Wright initiated in October 2025, directing FERC to prioritise large-load interconnection — defined as facilities consuming more than 20 megawatts. The standard grid interconnection queue in the US processes requests in order, with average wait times of three to five years for large facilities. AI data centres require power at a speed that the existing process cannot accommodate. The FERC ruling creates a government-endorsed expedited pathway, treating AI compute as infrastructure rather than commercial property.
All five FERC commissioners voted unanimously. In a divided US political environment, a unanimous five-commissioner vote on energy regulation is notable — it signals bipartisan agreement that AI infrastructure capacity is a strategic national concern, not a private sector issue.
Why this matters for UK businesses using cloud AI tools
Most UK businesses do not own AI data centres. They access AI through cloud platforms — AWS (Amazon), Azure (Microsoft), and Google Cloud. The power capacity of those platforms determines how much compute is available, how reliably services run, and ultimately whether AI tool pricing stays stable or rises as demand outstrips supply.
The FERC ruling accelerates the timeline for US data centre expansion, which means the large cloud providers can build more AI compute infrastructure faster. That is directionally positive for UK cloud AI users — more capacity eventually means more stable pricing and less risk of service degradation during demand spikes.
The relevant counterpoint: electricity is now a meaningful input cost for AI services, and AI data centres are competing with other industries for grid access. The ruling forces grid operators to prioritise AI over other large industrial users, which is a political choice about resource allocation. UK businesses are indirect beneficiaries of that choice when it supports the cloud providers they rely on — but the energy cost of running AI at scale will remain a structural factor in pricing for the foreseeable future.
The platform selection implication
This is a signal that favours the major hyperscalers over smaller AI API providers when making long-term platform decisions. Amazon Web Services, Microsoft Azure, and Google Cloud all have direct long-term power purchase agreements, data centre construction pipelines, and now regulatory support for grid access. They are positioned to grow compute capacity at a pace that matches AI demand. Smaller AI providers without their own infrastructure are dependent on the same hyperscaler capacity — without the long-term agreements that lock in pricing and supply. For UK businesses committing to an AI platform for a 2–3 year horizon, infrastructure stability is a real consideration alongside capability and cost.
What UK operators should take from this
This is a long-game signal, not a this-week action: The FERC ruling does not change what is available to you today. It shapes what will be available in 2027–2028. Read it as confirmation that the major US cloud platforms are building AI infrastructure for the long term, with government support.
Prefer platforms with owned infrastructure when making 2-year commitments: When choosing AI tools and platforms for workflows you intend to run for two or more years, consider whether the provider has its own data centre infrastructure (AWS Bedrock, Google Vertex, Azure OpenAI) or relies on reselling others' capacity. Infrastructure-backed providers are more likely to offer pricing stability and capacity guarantees over a multi-year horizon.
Do not expect AI pricing to drop sharply in the near term: Energy costs are a real input cost to AI services. The infrastructure investment needed to meet AI demand is substantial. Build your business case for AI tools on current pricing, not on anticipated price drops. Price reductions will come through efficiency gains in chips and models, not from energy getting cheaper.
UK energy parallel: The UK is having its own version of this conversation — data centre capacity, planning permission, and grid access are live policy debates in Westminster. The FERC move signals where the US has landed: AI infrastructure is treated like roads and airports, not like office blocks.