Tech brief
Grid Deals, Classroom Pushback, and Court Hallucinations
Drax strikes a £20bn compute pact as generative AI runs into fierce institutional pushback across UK schools, courtrooms, and global supply chains.
Tech
Drax is transforming Britain's largest biomass plant into a £20bn AI compute stronghold — unlocking immediate grid access while tethering tech infrastructure directly to contested carbon accounting.
BackgroundHyperscale artificial intelligence models require immense volumes of steady electrical power that strain traditional national distribution grids. Drax operates Britain's largest power station, converting former coal-fired generating units to burn imported biomass pellets under heavy government subsidies.
- The initial phase commits £2.5bn to construct 100 megawatts of capacity by early 2028 before scaling to an unprecedented one-gigawatt compute footprint.
- Developers plan to tap Drax's massive high-voltage grid connections directly, bypassing standard distribution queues that currently delay greenfield UK data builds by up to a decade.
- Environmental advocacy groups raised immediate objections, warning that feeding energy-intensive AI training clusters directly from biomass combustion undermines domestic decarbonisation targets.
Tech
Washington is shutting down offshore cloud leasing loopholes — making it clear that export controls apply to the compute capacity itself, wherever the physical silicon is housed.
BackgroundUS export controls prohibit domestic technology providers from shipping cutting-edge semiconductor hardware directly into mainland China. Tech giants frequently rented access to high-performance GPUs installed in third-party data centres across Southeast Asia to bypass territorial shipment curbs.
- The terminated agreement planned to deliver compute over five years, providing Tencent with vital infrastructure to train next-generation conversational models.
- White House officials warned American cloud providers that routing compute through overseas facilities to service Chinese clients directly breaches semiconductor restrictions.
- The abrupt deal collapse threatens forward revenue forecasts for cloud infrastructure providers that spent billions aggressively procuring Nvidia accelerator clusters.
Tech
Google has made generative AI mandatory for using its core search engine — presenting schools with an ultimatum that risks locking out classrooms unwilling to be algorithmic testbeds.
BackgroundGoogle Workspace for Education supplies search, email, and document software to hundreds of thousands of educational institutions worldwide. Schools routinely filter search results and disable untested AI models to protect minors from algorithmic hallucinations and data harvesting.
- Company executives confirmed that underlying search algorithms are now inseparable from AI overviews, leaving schools with an all-or-nothing choice on platform access.
- UK academy trusts and teaching unions expressed outrage over tech platform lock-in, warning that mandatory AI integration bypasses school safeguarding reviews.
- Google clarified that standalone chatbot apps can be disabled, but confirmed student search queries will inevitably generate synthesised summaries within primary search results.
Tech
The field's leading architect is warning that voluntary safety pledges will not stop a systemic catastrophe — demanding statutory powers before autonomous systems manage critical national infrastructure.
BackgroundGeoffrey Hinton earned the Nobel Prize in Physics for pioneering artificial neural networks before resigning from Google to openly warn about existential threats. Western governments currently rely on soft-touch evaluation institutes that inspect proprietary models only with developer consent.
- Hinton highlighted high risks of systemic disruptions targeting automated financial settlement systems and energy grids as frontier models gain agentic autonomous capabilities.
- The researcher endorsed a binding legal pact signed by 33 countries that mandates independent red-teaming prior to commercial foundation model release.
- Whitehall officials are weighing whether to empower the UK AI Safety Institute with statutory powers to halt commercial deployment of risky neural architectures.
Tech
Unlock the full brief
Sign in to read every signal, takeaway, and source. Free account — Apple, Google, or email.
