Disclaimer: This article provides general information and is not legal or technical advice. For official guidelines on the safe and responsible use of AI, please refer to the Australian Government’s Guidance for AI Adoption →
Three questions people are hammering into search and chat right now, plus the short answers you can steal.
Do "thinking" models use significantly more electricity?
Yes. GPT-5's energy consumption ranges from 2.33 Wh for minimal reasoning to 17.15 Wh for high reasoning—a more than seven-fold increase. The most energy-hungry models can exceed ~33 Wh per long prompt, which is 70x+ more than efficient deployments.
Is "Model Quality" actually just "Infrastructure Efficiency"?
This week's research suggests a model's "goodness" is an emergent property of the data center it lives in. The same model can use 70% less energy and water by switching infrastructure, meaning "quality" isn't just about weights and biases—it's about the grid, cooling, and hardware.
Can developers reduce AI's environmental impact?
Yes. Improving batch sizes from 4 to 8 can reduce energy per prompt by approximately 45%. The real question for builders: are you optimizing for capability only, or for capability per watt, per litre, per tonne of CO₂ at the scale your product is heading?
💡Quick note
This guide is part of our broader series on Weekly Deep Dive into AI and ML Advancements & Updates. Prefer to jump ahead? Browse related articles →
Read this if you are:
Founders & Teams
For leaders validating ideas, seeking funding, or managing teams.
Students & Switchers
For those building portfolios, learning new skills, or changing careers.
Community Builders
For workshop facilitators, mentors, and ecosystem supporters.
AI Bits for Techies | Issue #3 | 26 Jan 2026
Your weekly Aussie-flavoured deep dive into what changed in AI/ML, what matters, and what to do next (without living on release-note social media).
This week in one breath: A paper benchmarking the energy, water, and carbon footprint of LLM inference showing 70x+ differences between models, tools for local image generation and multilingual translation, and a shift in thinking: "model quality" might actually be "infrastructure efficiency"—the same model can use 70% less energy just by switching infrastructure.
The one paper you should pretend you read at lunch
How Hungry is AI? Benchmarking Energy, Water, and Carbon Footprint of LLM Inference
What is the setup?
Everyone argues about training being expensive, but the real day-to-day bill is inference: the prompts you send all day, every day. The paper's point is simple: we still lack clean, standardised, prompt-level numbers that factor in infrastructure, not just "the model."
What they did (yes, really)
They benchmarked 30 models using public API performance data, then layered on infrastructure multipliers like PUE and regional carbon intensity, with hardware configurations inferred statistically. They also use a probabilistic (Monte Carlo) approach and rank "eco-efficiency" with cross-efficiency DEA.
What happened
The spread is massive. The most energy-hungry models (they call out o3 and DeepSeek-R1) can exceed ~33 Wh per long prompt, which is 70x+ more than much smaller/efficient deployments. At the other end, they note a single short GPT-4o query at about 0.43 Wh, which looks tiny until you scale it.
Why it is interesting (beyond the number)
Because "same model" does not mean "same footprint." Datacentre overhead and where the workload runs can swing the outcome dramatically. In other words: model choice matters, but infrastructure choice is right behind it.
The real question
If per-query costs keep getting cheaper and faster, usage will explode anyway (Jevons vibes). So the real builder question becomes: are you optimising for capability only, or for capability per watt, per litre, per tonne of CO₂, at the scale your product is heading?
Best for: Developers and creators with consumer-grade hardware. These compressed models allow for professional-grade image generation and multi-reference editing directly on local GPUs (like the RTX 3090/4090) with sub-second response times. https://blackforestlabs.ai/
TranslateGemma (Google)
Best for: Building low-latency, multilingual applications. These open-weights models are optimized for speed and accuracy across 55 languages, fitting on everything from H100 GPUs down to mobile devices while maintaining high-fidelity translation. https://ai.google.dev/gemma
GLM-Image (Zhipu AI)
Best for: Designers requiring precise text rendering and multi-subject consistency. This hybrid model combines autoregressive and diffusion techniques to excel at complex tasks like style transfer and generating high-resolution images with legible text. https://github.com/THUDM/GLM-Image
Book recommendation (because your brain deserves more than changelogs)
The Atlas of AI – Kate Crawford
Crawford's work is the grounding counterweight to this week's paper. It zooms out and maps the physical stuff AI is actually made of.
Her core argument is blunt: AI is neither artificial nor intelligent. It is an extractive industry. One that runs on lithium mines, exploited labour, and relentless data harvesting.
Where this week's paper measures the operational hunger of LLMs, Crawford exposes the deeper, structural costs. The supply chains, the labour, the land.
The "cloud" stops looking fluffy very quickly. In her framing, it is a planetary-scale industrial system that centralises power and steadily drains natural resources.
Same story. Different layers.
Geeky thought of the day
Is "Model Quality" actually just "Infrastructure Efficiency"?
We often treat an AI's intelligence as a fixed property, but this research suggests that a model's "goodness" is actually an emergent property of the data center it lives in. If the same model uses 70% less energy and water simply by switching from a proprietary server to a highly optimized cloud provider, then a model's "quality" isn't just about weights and biases—it's about the grid, the cooling, and the hardware.
In an era of adaptive routing, we have to stop asking "How smart is this AI?" and start asking "How effectively can this infrastructure support its reasoning?"
Housekeeping (so we stay honest)
This is general information, not legal advice. If you ship user-facing AI, be transparent about where AI is used, what it cannot do, and where humans stay in the loop.
About the Authors
Dr Sam Donegan
Founder & Lead Editor
Sam leads the MLAI editorial team, combining deep research in machine learning with practical guidance for Australian teams adopting AI responsibly.
Jun Kai (Luc) Chang
AI Software Developer
Luc is an AI Software Developer at Monash AIM, building neural networks on FPGA boards. He is pursuing a Master of AI at Monash and co-founding a startup in the event space.
Julia Ponder
Technical Writer
Julia specialises in translating developer jargon into plain English. She creates clear, expertly formatted documentation and tests products before they go to market.
Shivang Shekhar
Technical Writer
Shivang is a mechanical engineer and AI masters student at Monash University with a diverse science background. He is the main author for AI Bits for Techies each week.
AI-assisted drafting, human-edited and reviewed.
Frequently Asked Questions
Does a "Thinking" model use more electricity?
Yes, significantly. For a medium-length query, GPT-5's average energy consumption ranges from 2.33 Wh for minimal reasoning to 17.15 Wh for high reasoning—a more than seven-fold increase.
How does one AI query compare to a Google search?
A single short GPT-4o query consumes 0.42 Wh, which exceeds the footprint of a traditional Google search (0.30 Wh) by approximately 40%.
Why is water usage a factor in AI?
Data centers require massive amounts of water for cooling and off-site electricity generation. GPT-4o alone is projected to evaporate enough freshwater to fill over 500 Olympic-sized pools annually.
Can developers reduce this impact?
Yes. Improving batch sizes is one of the most effective levers; moving from a batch size of 4 to 8 can reduce the energy per prompt by approximately 45%.
Do I need to change my privacy notices for new AI features?
If you introduce new AI features that process personal or sensitive information, update your privacy notice and consent flows. Reference the OAIC APPs and include a short, plain-English description of what the model does, inputs needed, retention, and human oversight.
What is the safest way to start a pilot?
Begin with low-risk internal content (policies, FAQs), apply rate limits, log prompts/outputs, and perform red-team style testing. Use feature flags and role-based access. Run a DPIA/PIA if personal data is involved.
How should teams validate model performance?
Create a small, labeled evaluation set that mirrors your domain. Track accuracy, hallucination rate, latency, and cost per request. Re-test after any model switch or prompt change, and record changes in a decision log.
Are there grants or programs in Australia for AI experiments?
Check current state-based innovation vouchers, CSIRO Kick-Start, and university accelerator programs. Funding cycles shift, so confirm eligibility windows and co-contribution rules before committing spend.
