According to Bloomberg, Brookfield Asset Management CFO Hadley Peer Marshall estimates roughly 7 trillion dollars of capital is required to finance rapid AI growth. source: Bloomberg post on X by @business dated Oct 1, 2025. Brookfield has identified data centers and renewable power as core growth areas within its infrastructure strategy that underpin AI buildouts. source: Brookfield Asset Management 2023 annual report and investor materials. For crypto markets, rising AI-driven data center load can tighten electricity supply and raise operating costs for BTC miners, pressuring margins and hashprice, as the International Energy Agency projected global data center electricity consumption could reach 620 to 1,050 TWh by 2026 from about 460 TWh in 2022. source: International Energy Agency Electricity 2024 report. Several North American miners have begun monetizing AI and high-performance computing to diversify revenue, including Core Scientific’s 12-year AI compute hosting agreement with CoreWeave in 2024 and Iris Energy’s purchase of Nvidia H100-class GPUs for AI workloads in 2023. source: Core Scientific press release June 2024 and Iris Energy company update May 2023. Trading takeaway: prioritize AI infrastructure equities and crypto mining stocks with HPC exposure, and monitor regional power prices and BTC hashprice given the capital scale cited by Bloomberg and the power-demand trajectory flagged by the IEA. source: Bloomberg post on X by @business dated Oct 1, 2025 and International Energy Agency Electricity 2024 report.

According to @JeffDean, Google has released a technical paper detailing a methodology to measure the environmental impact of Gemini inference, providing a verifiable baseline for energy-per-inference modeling for traders (source: Jeff Dean on X, Aug 21, 2025). @JeffDean reports the median Gemini Apps text prompt consumes 0.24 watt-hours, establishing a concrete per-prompt energy intensity metric (source: Jeff Dean on X). Using this figure, 1 million prompts would consume about 240 kWh and 10 million prompts about 2.4 MWh, enabling direct scaling of workload energy and cost models (source: Jeff Dean on X). For crypto markets, this per-inference benchmark can be applied when assessing data center power needs and efficiency for compute-linked infrastructure adjacent to mining and AI-related tokens (source: Jeff Dean on X).