The rapid integration of artificial intelligence into daily work has reached a new threshold: 76% of employees now use AI on the job, up from 30% in 2023, according to McKinsey data cited by industry practitioners. This surge is delivering measurable productivity gains—sectors with intensive AI adoption saw output grow 34% between 2018 and 2025, PwC’s 2026 AI Jobs Barometer shows—but it is also accelerating two underappreciated costs. Global data centres consumed an estimated 448 trillion watt-hours of electricity last year, a figure the UN University expects to more than double within four years, while the water required to cool them is projected to reach 9.3 trillion litres by 2030. A single 100-word AI query can consume half a litre of water, researchers at the University of California, Riverside, have calculated.

The environmental strain is reshaping local politics. In June, Monterey Park, California, became the first US city to impose a moratorium on new data centres, citing water scarcity. Imperial County, also in California, reversed approval for what would have been the state’s largest such facility after residents challenged its environmental review. Across the Atlantic, European regulators are embedding sustainability requirements into the EU’s AI Act, while in Asia-Pacific, a First Street report finds 89% of data centre capacity faces high risk from extreme heat and drought. These pressures are not slowing investment—US start-ups attracted $339 billion in 2025, two-thirds of it AI-focused—but they are forcing a reassessment of where and how infrastructure is built.

Governance gaps are widening in parallel. The Five Eyes intelligence alliance issued a joint advisory in June warning that AI is “rapidly transforming cyber risk” and that breaches will occur as models learn to exploit zero-day vulnerabilities. Within firms, the phenomenon of “Shadow AI”—employees using unsanctioned tools—is alarming security chiefs; Gartner predicts over 40% of enterprises will suffer related incidents by 2030. In response, companies are creating AI officer roles and centralising platforms, a shift that Italian AI firm Memori describes as the defining challenge of 2026. Financial institutions, meanwhile, are moving from pilots to scaled deployment, with a World Economic Forum report stressing that trust and human oversight must remain at the core.

The labour market is bifurcating. Jobs heavy in routine data entry, basic customer service, and simple translation are contracting, while roles demanding complex judgment, creativity, and interpersonal skills are proving resilient. The UAE illustrates a hybrid model: 98% of investors there use AI tools, the highest rate globally alongside India, yet 56% prefer combining AI with expert human advisers for final decisions, an HSBC survey found. Amazon Web Services’ CEO told the Platformer podcast that half of white-collar jobs “may change” rather than disappear, but workers must be willing to learn. The next factual milestone is the 2026 deadline many firms have set to move from fragmented AI experimentation to structured governance, as the EU’s AI Act begins phased enforcement and corporate boards face mounting pressure to account for both the productivity promise and the environmental and security liabilities of the technology they are racing to adopt.