Brightstar Resources has reported a 305.4m gold intercept grading 1.82g/t from 351.9m depth in hole TMHRCD26020 at the Two Mile Hill–Shillington deposit, Sandstone project, Western Australia, indicating a mineralised intrusion roughly three times wider than previous models. The intercept is unconstrained at depth, suggesting continuity beyond the current 657.3m end-of-hole and potential for a large-tonnage, bulk underground target rather than narrow-vein stoping. For resource modellers and mine planners, the result may justify revising intrusion geometry, tonnage estimates and potential underground development scenarios.

Artificial intelligence is being rolled out across mines for tasks such as mill optimisation, predictive maintenance and fleet dispatch, but its success still hinges on experienced operators, metallurgists and geotechs defining constraints and validating outputs. Vendors are pairing machine-learning models with domain-specific tools – for example, digital twins of crushers and grinding circuits, or AI-assisted short-interval control in fleet management systems – to capture tacit knowledge from senior crews. The message for sites is to treat AI as an augmentation layer on existing control, planning and reconciliation workflows, not a plug‑and‑play replacement for human judgement.

AI-driven data centres are emerging as both customers and competitors for miners, with Meta’s 1‑GW, US$9.5‑billion Sturgeon County facility and a 932‑MW dedicated gas plant in Alberta illustrating how hyperscale loads can strain grids and local support near projects like E3 Lithium’s Clearwater brine operation. Alberta has imposed a 1,200‑MW cap on large new connections to 2028, while BC Hydro is rationing up to 400 MW for tech projects but exempting mining, forestry and LNG. S&P Global warns copper demand could reach 42 Mt by 2040, yet rising power prices, grid queues and competition for electricians and engineers risk delaying mine and smelter investments.