OpenAI files for IPO; UK commits £1.1bn to AI hardware

Europe and the US are racing simultaneously to set the terms of AI — financially, geopolitically, and environmentally.

OpenAI files for IPO, joining Anthropic on the Wall Street runway

OpenAI filed confidential paperwork with the US Securities and Exchange Commission on 8 June, beginning the process of becoming a publicly traded company. The company, valued at between $730 billion and $850 billion in private markets, follows rival Anthropic, which made a similar confidential filing one week earlier. The three simultaneous filings — OpenAI, Anthropic, and SpaceX — mark the most concentrated cluster of high-profile AI company IPO preparations on record, according to the Associated Press. OpenAI has described the move as part of a plan to "build to benefit everyone."

UK launches £1.1 billion AI hardware plan anchored by a national supercomputer

The UK government announced a £1.1 billion AI Hardware Plan at London Tech Week on 8 June, including a £750 million investment in a national AI supercomputer to be deployed at the University of Edinburgh by 2030. A further £400 million is earmarked for next-generation chip development, and £150 million will be used to purchase inference chips from British firms this summer. Technology Secretary Liz Kendall framed the programme as building sovereign computing capacity to keep the UK competitive as global AI infrastructure investment accelerates. Details are published on GOV.UK.

EU adjusts AI Act timetable — transparency and sandbox rules deferred

The European Commission has confirmed a staggered compliance calendar under the EU AI Act. Transparency obligations for AI systems that generate or manipulate synthetic content — requiring outputs to be machine-readable and labelled as AI-generated — have been deferred from 2 August 2026 to 2 December 2026. Requirements for member states to establish national AI regulatory sandboxes have been pushed back one year to 2 August 2027. Draft guidelines on classifying high-risk systems remain open for public consultation until 23 June 2026.

UN warns AI could consume as much water as 1.3 billion people annually by 2030

A United Nations University report published on 8 June found that AI expansion is placing measurable pressure on global water, land, and climate resources. Data centres consumed an estimated 448 terawatt-hours of electricity in 2025, roughly 20 percent of which was AI-related. By 2030, AI-related electricity demand could reach 945 terawatt-hours, and AI-related water consumption could equal the basic annual domestic needs of 1.3 billion people. The authors call for binding reporting standards so the environmental cost of AI can be tracked and managed.

Trump and Sanders converge on public ownership stake in AI companies

In an unusual moment of cross-partisan agreement, both President Donald Trump and Senator Bernie Sanders signalled support for the American public holding an ownership stake in AI companies. OpenAI chief executive Sam Altman met with Senator Sanders to discuss the idea, expressing openness to public equity arrangements; Broadband Breakfast reported both leaders' positions on the same day. Speaking on Air Force One, President Trump described a potential structure "where the American people can benefit from the success of AI." No specific legislative mechanism has yet been proposed.

Meta launches $115 million free workforce academy for AI infrastructure jobs

Meta announced on 8 June a $115 million initiative — "America's Workforce Academy" — to train workers for careers in AI data centre construction and maintenance. The programme is cost-free for participants, with guaranteed job offers upon completion. The first training centre opens in Baton Rouge, Louisiana, with further locations planned in Indiana, Texas, and Ohio. The academy supports Meta's multi-billion dollar data centre expansion in Richland Parish, Louisiana, and forms part of a broader industry effort to build the skilled trades base AI infrastructure requires.