Data-centre water consumption body with ChatGPT card

Data-centre water consumption body with ChatGPT card, editorial, light-mode, light

Preview image. Unlock full-res

Cream body slide with a serif headline, a blue iceberg-style "ChatGPT" data card with example, capability and limitation columns, and body copy on the right.

Summary

A cream slide led by a serif headline about data-centre water use, with a large blue iceberg-style panel labelled "ChatGPT" splitting into Examples, Capabilities and Limitations columns, and a column of body copy on the right.

Visual description

Warm cream ground with the section-nav bar and "13" page index up top. A three-line serif headline runs across the top-left ("Researchers estimated that Microsoft may have consumed 700,000 litres of drinkable water to power the servers for training GPT-3, enough to produce 370 BMW vehicles."). The center-left holds a large blue panel styled as an iceberg above and below a waterline: a monospace caption ("Running GPT-3 inference for 10-50 queries consumes 500 ml of water") sits at the top, a "ChatGPT" label centers it, and below the waterline three labelled columns (Examples, Capabilities, Limitations) list short items. The right column carries small dark sans-serif body copy with underlined links.

Key takeaway

Dressing a plain three-column comparison as an iceberg so the "hidden cost below the surface" metaphor carries the data-centre-water point visually. The monospace waterline caption ties the metaphor to a concrete figure.

Reuse notes

Good for a slide that needs to present a tidy three-bucket breakdown while also making a "what lies beneath" argument. The iceberg framing is reusable for hidden-cost or hidden-complexity narratives; keep it to one flat blue and let the waterline split the buckets. Pair the serif headline conclusion with a body column for sourcing.

From this deck: Data-centre water consumption body with ChatGPT card

View deck

More like this