A technology industry group that includes NVIDIA Corporation, Microsoft Corporation and Cisco Systems Inc. has warned the UK's Labour government that it has roughly 12 months to pre-empt public opposition to data centres before organised campaigns threaten construction, in a submission that highlights deepening friction between AI infrastructure ambitions and community resistance on multiple continents. The group, operating as the Sovereign AI Leadership (SAIL) taskforce, argued that successive dry summers and rising energy bills are increasing public scrutiny of the warehouses used to train and run AI models.

SAIL's submission urged ministers to establish a 'compact' framework that would allow councils hosting data centres to retain more local tax revenue, citing the United States as a cautionary example where organised local campaigns have halted fully permitted projects and some growth-oriented states have announced moratoriums. The taskforce pointed to 15 community campaigns already active in Scotland, as documented by Action to Protect Rural Scotland, opposing hyperscale AI data centres. Microsoft, one of the SAIL members, has pledged $30 billion in UK AI infrastructure investment and has built a data centre in Acton, London, with another under construction in Newport [Source: The Guardian].

The UK warning arrived alongside a separate disclosure from South Korean infrastructure arm SK Hyper Co., which is in talks with major technology firms and financial investors to secure funding for a $350 billion data centre buildout. SK Hyper CEO Chung Suk-geun said the company plans to share the overall cost with data centre tenants and outside investors [Source: Bloomberg]. The $350 billion covers only the first 5-gigawatt phase — including equipment — of a broader plan to build 15 gigawatts of data centre capacity in South Korea by 2035. SK Hyper aims to close its first round of external capital by the end of 2026 or early 2027, with further fundraising spread over three to four years. The company is already developing a 900-megawatt project in Ulsan alongside a previously announced 100-megawatt project for Amazon.com Inc.. SK Telecom, SK Hyper's parent, plans to invest 750 billion won (approximately $567 million) in the unit by 2030. SK Group chairman Chey Tae-won has discussed an investment path of roughly 1,000 trillion won (approximately $648 billion) into AI data centres by 2035, with 3 million GPUs eventually running through them [Source: Seoul Economic Daily].

Chung acknowledged that securing global demand and raising the required capital will be the biggest challenges, while power and sites for the first 5-gigawatt phase should be more readily available [Source: Seoul Economic Daily]. The fundraising effort is unfolding as markets are contending with fears of an AI bubble, with investors backing data centre projects uncertain whether future AI service spending will justify the trillions flowing into the global buildout. Chung said SK Hyper would adjust the pace of investment according to actual demand rather than committing to the full plan upfront. The infrastructure tensions extend beyond the UK and South Korea: more than three-quarters of data centres set to be built around Australia's biggest cities face risk from stretched power grids and local opposition, threatening that country's ambition to become an Asia-Pacific AI hub, while new and proposed data centres in Texas are generating pollution risks that could return air quality to 2000s levels. In Germany, idle nuclear plants are being examined for potential repurposing as data centres while reactor dismantling continues, even as a separate plan to funnel heat from data centres into nearby homes has stalled amid competing climate policy objectives. NVIDIA shares closed at $221.97 on 21 October, down 1.4% over the prior four weeks, according to Valdrans data. Valdrans' insider activity signal for NVIDIA is rated Negative as of 22 October, with 25 sales and no purchases recorded over the prior 90 days per SEC filings compiled by Valdrans.

Sources: The Guardian, Bloomberg, Seoul Economic Daily, The Edge Malaysia, The Verge, Wired, Valdrans Data