AI datacetnteers use 10x more power than non-ai: https://techplustrends.com/power-requirements-ai-data-centers/
You spout Democrat talking pointed.
Ca. is draining Az. water supply. Nuke desalinization plants should have been operational and a massive underground pipeline with pumping stations should have been constructed and operational to transport water from the Great Lakes to the Colorado River. Instead the Ca. dems sat on their hands waiting for someone else to save them.
Grok:
False.
AI data centers (or AI-focused facilities/racks) do not use 10 times more power overall than non-AI (traditional) data centers. Aggregate global figures show the opposite or a closer balance so far, though AI facilities have much higher power density per rack or per square foot, and their share of total data center electricity is rising rapidly.Aggregate electricity useAccording to International Energy Agency (IEA) data (around 2025 figures referenced in 2026 analyses), total data center electricity consumption was roughly 415–485 TWh globally. Non-AI data centers accounted for about two-thirds, and AI-focused ones the remaining third. AI's share of world electricity was estimated at around 0.5%.
ourworldindata.org
Mainstream/traditional compute still dominates overall data center energy use in many assessments (often cited as 70–90% recently, with AI at 10–30% and growing). AI is the main driver of growth, with projections that AI-focused consumption will roughly triple or more by 2030 while total data center demand roughly doubles.
hpe.com
Power density (where the big difference is)AI racks/servers are far more power-hungry:Traditional data center racks: typically 5–15 kW (sometimes cited as 5–10 kW).
AI/GPU racks: commonly 40–80+ kW today, with advanced configurations (e.g., dense NVIDIA setups) reaching 100–130+ kW or higher (some next-gen designs hundreds of kW). This is often 5–10× (or more) the density of traditional racks.
red-eng.com
Some industry commentary refers to a “10x challenge” for AI factories in terms of energy, heat, and complexity relative to traditional data centers—mainly reflecting this higher density, sustained high utilization, and cooling needs rather than total facility power being uniformly 10× higher.
wired.com
Other notesIndividual AI hyperscale facilities can be very large (100+ MW, with some planned in the hundreds of MW to GW scale), comparable to or larger than many traditional ones, but total AI capacity is still smaller than the existing non-AI base.
Cooling is a bigger factor for AI due to heat from dense GPUs (often requiring liquid cooling instead of air).
Chips in AI servers can draw 2–4× (or more) the power of traditional counterparts under load.
In short: AI data centers are significantly more power-dense and are driving most of the growth in data center electricity demand, but the blanket claim that they use 10× more power than non-AI ones is not accurate on a total-consumption basis.
Elon is highly intelligent. Perhaps he’ll find a way to turn HT+0 into water.
Not quite sure whether to take this seriously or duck flying paperweights.