Simon Willison: AI hyperscalers should buy up golf courses and convert them to public parks to offset water use

Main idea

Willison argues that hyperscalers face growing criticism for massive data center water consumption — Google used 10.9 billion gallons of water in 2025. Instead of buying carbon offsets, he proposes a concrete 'compromise': buy golf courses in dry regions (Coachella Valley has 120 courses consuming ~750,000 gallons/day) and convert them into public parks.

Context

The essay emerged amid growing public pressure on tech companies over the energy and water footprint of AI infrastructure. Willison consistently tracks AI's societal impacts — the essay appears a day after reports on Big Tech emissions (Microsoft +25%, Big Tech emissions = a third of France's total).

Why it matters

The satire combines a genuine environmental argument: AI data centers consume physical water and most offsets exist only on paper. For developers and decision makers, this is a reminder that the infrastructure debt of AI is physical — not merely regulatory.

Details / arguments

• Google 2025: 10.9 billion gallons of water for data centers
• Coachella Valley: 120 golf courses x ~6,250 gal/day = ~750,000 gal/day
• Proposal: tech firms buy golf courses, convert to parks, give golfers binoculars for birdwatching
• Tone: satirical, but the environmental argument is real and well-supported