16 charts that explain the AI boom (www.understandingai.org)

A new visual roundup — “16 charts that explain the AI boom” — frames today’s surge as one of the largest post‑war investment waves: big tech capex (Amazon, Meta, Microsoft, Alphabet, Oracle) hit $241B in 2024 (≈0.82% of US GDP) and $97B in Q2 2025 (≈1.28% of that quarter’s GDP). The piece links market momentum (Nvidia’s record highs, OpenAI’s Sora launch, and an Anthropic–Google deal for access to up to one million Google chips) to a rapid buildout of GPUs, data centers and cloud services that could eclipse past national-scale projects if sustained. Key technical takeaways: GPU and large-computer imports have spiked (over $200B annualized), and data‑center construction and rents are surging amid ~1.6% vacancy in primary markets—fueling geographic concentration (Northern Virginia, low‑cost energy regions) and local grid stress. Global data‑center electricity was ~415 TWh in 2024 and could rise to 700–1,400 TWh by 2030 depending on the model (IEA, McKinsey, Deloitte, Goldman ranges). Inference demand is already enormous: Google reports ~1.3 quadrillion tokens/month and OpenAI ~260 trillion tokens/month on its API. Water use, often cited as a concern, remains modest (~48M gallons/day on site; ~250M including power generation). The charts underscore practical constraints — supply chains, tariffs, power and cooling — that will shape where and how the next phase of AI infrastructure is deployed.