Summary

• Project SAIL sets out ten industry recommendations covering computing capacity, electricity connections, community benefits, skills and AI sovereignty.
• It proposes a Compute Accelerator operating by April 2027 and a community benefits pilot by October 2027.
• The recommendations have not been adopted by government, and the proposed delivery deadlines are not binding.

A technology industry taskforce convened by Rigby Group has proposed changes to Britain’s AI infrastructure programme, including a new computing access service, reforms to electricity connections and community benefits for places hosting data centres.

The Project Sovereign AI Leadership Taskforce published ten recommendations in Project SAIL: Mobilising the Nation, with an announcement issued on 22 October. Its membership includes senior representatives from Microsoft, Nvidia, Cisco, BT, Barclays and OneAdvanced, although the resulting proposals are not government decisions.

Rigby Group, which convened the initiative, argues that expanding AI adoption will require investment in infrastructure alongside practical arrangements for skills and public confidence. The report proposes a Prosperity Compact linking central government, businesses, communities and workers, with indicative milestones extending through 2027.

Several participating businesses supply the hardware, connectivity and services likely to benefit from increased investment. Their experience may inform the technical recommendations, but public authorities will still need independent evidence on cost, feasibility and competing demands for public resources.

Access to computing beyond the largest buyers

Project SAIL recommends a Compute Accelerator intended to address a gap between modest cloud requirements and the much larger computing clusters purchased by major technology companies and research institutions. Its proposed deadline for an operational service is April 2027.

Smaller businesses developing AI products may require access to specialist processors without possessing the resources to reserve substantial capacity or manage dedicated infrastructure. Pricing and availability can influence whether an experimental application progresses beyond initial demonstrations.

A shared service would still need rules determining who qualifies, how capacity is allocated and whether users receive technical support. Access to processors does not, by itself, provide the software engineering, data preparation or evaluation needed to build a reliable application.

The taskforce assigns responsibility in its recommendation to the government department concerned with business, innovation, science and trade. That is a proposed allocation, not evidence that a department has accepted responsibility or committed the money required to establish the service.

Data centres and local communities

A second recommendation calls for Community Benefit Agreements associated with data centre developments, supported by arrangements allowing host areas to retain relevant business rates. Project SAIL proposes consultation by January 2027 and a pilot by October 2027.

Developments can require major electricity connections, large sites and supporting construction. Residents and councils may question whether the local employment and economic return adequately compensate for infrastructure pressure or environmental effects.

Community agreements could create a formal route for discussing benefits with host authorities. Their value would depend on enforceable commitments, transparent assessment and a distinction between benefits offered to residents and the mitigation measures already required through planning.

Allowing an area to retain business rates would involve fiscal choices affecting central and local government. Neither developers nor the taskforce can establish such arrangements without the relevant public decisions.

The proposal also raises questions of consistency. An agreement suitable for one location may not address the needs of a community facing different electricity, water or land use constraints. A national framework would need room for local requirements without making the process unpredictable for investors.

Electricity connections remain a constraint

Project SAIL calls for reform of the grid connection queue through a framework involving the Department for Energy Security and Net Zero, the National Energy System Operator and Ofgem by April 2027.

AI computing facilities require dependable power, and a delay in obtaining a suitable grid connection can determine whether an otherwise financed development is viable. Yet housing, factories, transport and other electricity users are competing for capacity and infrastructure investment.

Changing queue management can help distinguish credible projects from speculative reservations, but administrative reform does not create physical network capacity. Transmission and distribution equipment must be available and capable of meeting the loads proposed.

Planned connection capacity should also be distinguished from actual electricity consumption. The amount used by a data centre depends on the installed equipment, its utilisation and the site’s operating arrangements.

The taskforce’s framework would need to balance those technical issues with fairness to other applicants. Preferential treatment for AI infrastructure could impose costs or delays elsewhere if the wider system is not considered.

Skills and sovereignty require more detail

The recommendations include a coordinated AI skills settlement, with the first sector programme proposed for October 2027. Employers adopting AI need staff able to examine model outputs, integrate applications and maintain appropriate information security rather than relying solely on the purchase of software.

Project SAIL also calls for a practical approach to sovereign AI. Government and industry may seek stronger control over computing and critical services, but even domestically operated facilities depend on international semiconductor manufacturing, specialist equipment and software supply chains.

The taskforce therefore identifies genuine dependencies without providing evidence that complete technological independence is affordable or achievable. Contracts, alternative suppliers and domestic investment may offer different ways to manage specific risks.

Although Project SAIL provides dates and suggested accountable bodies, it offers no authority to require delivery. The government must still decide whether to consult, fund or implement any recommendation.

Its proposals are more readily testable than a general call for AI investment because they set milestones for computing access, community agreements and grid reform. Whether those milestones become policy will establish the report’s influence, while the cost and distribution of benefits remain central questions for government and communities.