英国“最大人工智能超级计算机”的发布因电力供应问题推迟
导读｜英国劳顿人工智能超级计算机原计划于2027年投入运营，如今因无法获得足够电力，可能推迟到2030年代中期。项目延期暴露出英国人工智能基础设施建设中的现实约束：算力设备可以快速部署，电网接入和新增发电能力却需要更长时间。
要点速览
｜劳顿项目由英国初创企业Nscale建设，曾被政府称为英国最大的自主人工智能数据中心。
｜项目计划申请约90兆瓦电力，接近31.5万户家庭的能源消耗量。
｜英国目前有315个数据中心等待接入国家电网，申请总容量达到73吉瓦。
｜英国人工智能战略、数据中心建设和电力基础设施之间出现明显错位。
图1｜劳顿人工智能数据中心的规划效果图。项目曾被政府列为英国人工智能基础设施建设的代表性工程。
一、项目延期与建设现状
英国政府大力宣传的劳顿项目位于埃塞克斯郡，建设方是总部位于英国的人工智能基础设施初创企业Nscale。项目在2025年公布时被称为英国最大的人工智能超级计算机，原计划于2027年首次运行；目前由于无法获得足够电力，启动时间可能推迟到2030年代中期。
《卫报》今年3月曾报道，项目所在地仍是一处正在运营的脚手架堆场，引发外界对建设进度的质疑。Nscale表示仍然致力于项目，现已清理场地，并研究现场发电和加快并网的方案。电力接入问题让项目从建设进度争议转向基础设施可行性问题。
Nscale今年计划上市，目标融资额为350亿美元。英国政府在2025年人工智能战略中把劳顿项目称作“英国最大的自主人工智能数据中心”，并将数据中心比作聊天机器人等人工智能服务的中枢。项目能否按期运行，关系到英国扩大本土算力、吸引人工智能企业和提升数字基础设施能力的政策目标。
图2｜劳顿项目所在地的施工现场，原址仍可见脚手架堆场。电力并网延期使项目启动时间进一步推后。
二、电力接入成为数据中心建设瓶颈
人工智能超级计算机需要大量高密度芯片持续运行，电力需求远高于普通办公楼或传统数据中心。劳顿项目正在寻求最高90兆瓦的发电容量，大致相当于31.5万户家庭的能源消耗量。英国电力网络公司已经告知Nscale，现有电网可能要到2030年代初期至中期才能提供足够容量。
英国能源监管机构今年6月披露，315个数据中心正在排队等待接入国家电网，总申请需求约73吉瓦，而英国全国峰值用电需求约为45吉瓦。数据中心需要先接入地方配电网络，随后还要等待国家输电网络完成扩容，项目能否开工因此取决于更大范围的电网建设。
英国政府已经承认，数据中心扩张正在挤占有限的电网接入资源。能源监管机构随后提出改革方案，准备清理缺乏实际建设进展的项目，并通过承诺费用和阶段性里程碑提高接入申请的真实性。劳顿项目的延期，正好体现了大型人工智能设施与现有电力系统之间的时间差。
三、人工智能战略与能源基础设施的错位
英国政府的人工智能机遇行动计划提出建设人工智能增长区，通过更快的规划审批和更稳定的电力接入，加快数据中心建设。政府随后公布的算力路线图计划扩大公共人工智能计算能力，并鼓励私人企业在英国建设大规模数据中心。Nscale的劳顿项目正是在这套政策环境下获得高度关注。
人工智能企业可以在短时间内采购芯片、设计机房和筹集资金，输电线路、变电站和发电设施却需要多年规划、审批和施工。电力不足会推迟数据中心上线并抬高运营成本。劳顿项目表明，人工智能竞争已经延伸到电网、土地、冷却系统和能源供应，英国的本土算力建设取决于这些设施能否同步推进。
原文附录｜Launch of UK’s “largest AI supercomputer” delayed by power supply problems
Dan Milmo, Global technology editor｜24 September 2026
A huge datacentre project hailed by the UK government will miss its launch date next year and could be delayed into the mid-2030s.
The site in Loughton, Essex, was described as the country’s largest AI supercomputer when it was announced in 2025, but power supply problems mean it now faces a lengthy wait before coming online.
Doubts had already emerged about the scheme after the Guardian revealed in March that the planned location was still an operational scaffolding yard.
Now there are further concerns about the project due to its inability to obtain enough power before its planned 2027 debut. The Loughton supercomputer is being built by Nscale, a UK-based startup seeking a $35bn (£26bn) flotation this year.
UK Power Networks (UKPN), the company that will connect the site to the local power grid, has reportedly told Nscale the grid will not be able to supply sufficient power to the datacentre until the early to mid-2030s.
The power delay was first reported by Morning Intelligence, an AI industry newsletter.
Nscale said it was “committed to the project” but the failure to find a quick fix to the power supply issue underlines how much energy has become a hindrance to datacentre development. The scaffolding yard site has now been cleared by Nscale, however. The company is also investigating whether it can generate power onsite and whether it can speed up the connection process.
Keir Starmer’s government cited the Loughton project when announcing the UK’s AI strategy in 2025, describing the site as the “largest UK sovereign AI datacentre”. Datacentres are the central nervous system of AI technology such as chatbots.
Ofgem, the energy industry regulator, has warned of a logjam in applications for electricity connections. In June, it revealed there were 315 datacentres queueing to connect to the National Grid, representing 73GW of demand – compared with the peak energy demand for the entire country of 45GW.
The Loughton site is seeking up to 90MW of capacity, equivalent to the energy consumption of about 315,000 homes.
UKPN said it could not comment on an individual project. Schemes with large energy demand also require work to be done at a National Grid level, in order to ensure that energy demands can be met properly.
“In general, many datacentres, whilst they will connect to our distribution network, are awaiting works on the wider national electricity transmission network before they can begin operation,” said a UKPN spokesperson.
National Grid said: “We’re continuing to work closely with NESO, the government-owned National Energy System Operator, distribution network operators and customers to identify opportunities to accelerate connections where possible and support economic growth.”
参考资料
• 《卫报》：英国“最大人工智能超级计算机”的发布因电力供应问题推迟：https://www.theguardian.com/technology/2026/sep/24/construction-largest-supercomputer-delayed
• 英国能源监管机构：释放电网容量，处理投机性数据中心项目：https://www.ofgem.gov.uk/press-release/ofgem-acts-free-grid-capacity-tackling-speculative-data-centre-projects
• 英国政府：人工智能机遇行动计划：https://www.gov.uk/government/publications/ai-opportunities-action-plan/ai-opportunities-action-plan
• 英国政府：人工智能机遇行动计划回应文件：https://www.gov.uk/government/publications/ai-opportunities-action-plan-government-response/ai-opportunities-action-plan-government-response
• 英国政府：人工智能增长区实施文件：https://www.gov.uk/government/publications/delivering-ai-growth-zones/delivering-ai-growth-zones
• Nscale：在英国数据中心产业投资20亿英镑：https://www.nscale.com/press-releases/ai-hyperscaler-nscale-to-invest-gbp-2-billion-in-the-uk-data-centre-industry