Google and Amazon reports expose AI’s soaring carbon costs

The sustainability reports from Google and Amazon, and the TechCrunch analysis around them, are a sobering reminder that scaling toward AGI is not just about parameters and GPUs—it’s about megatons of CO₂ and billions of litres of water. A 25% jump in Google’s emissions and a mid‑teens rise at Amazon, against the backdrop of net‑zero pledges, shows that AI is currently a climate regressive technology at the margin.

Strategically, this creates a three‑way tension: between performance, cost and externalities. If each new model generation demands more compute and more data center build‑out, big tech can either raise prices, absorb lower margins, or lean harder into offsets and creative accounting. None of those are stable equilibria. That’s likely to accelerate interest in efficiency‑first architectures, sparse and modular models, and more aggressive workload placement policies. It also opens competitive space for companies that can credibly claim “green AI” advantages—through custom chips, waste‑heat recovery, or siting next to abundant renewables.