What is “intelligence per watt”? Can it helps Local AI replace the Cloud?

Share it with your friends and colleagues

Reading Time: 2 minutes

Can Local AI Replace the Cloud?

For the last 2-3 years, the AI conversation has been dominated by one assumption:

AI = massive cloud data centres.

But a new study suggests something interesting:

Your laptop may soon handle a large portion of AI workloads.

And the key metric behind this shift is something called ā€œintelligence per watt.ā€

The Core Idea – intelligence per watt

Think of intelligence per watt as:

How much useful intelligence do you get for each unit of energy consumed.

If two systems produce similar answers, the one that uses less power is the better system.

Historically, cloud systems have dominated here.

But that advantage is shrinking.

What the Research Found

Researchers tested multiple open-weight models including:

– Qwen3

– Gemma3

– IBM Granite 4.0

– Mixtral-8x7B

– Llama 3.1

They ran them on:

• Laptop chips like the Apple M4 Max

• Data-center GPUs like the Nvidia H100

Then they evaluated 1 million real-world queries and measured both accuracy and power consumption.

The Results Are Surprising

Local AI efficiency is rising extremely fast

– Intelligence per watt improved 5.3Ɨ

– Smaller local models achieved ~88.7% of cloud accuracy

– Hybrid routing (local + cloud) could reduce energy usage by over 80%

Cloud GPUs still lead today.

But the gap is closing fast.

One More Interesting Angle

Most discussions about local AI focus on privacy.

But this research highlights something equally powerful: Economics and energy efficiency

If devices can answer most queries locally, the pressure on AI data centers could reduce dramatically.

What do you think?

Learn AI Agents through entertaining web series, and not lecture-style video

Like us, if you also hate learning through lectures then we invite you to watch our engaging educational web series.

You can explore the courses here: https://www.tisdoms.com/

If you have questions, feedback, or disagree with something in this article, I’d love to hear your perspective. Connect with me on LinkedIn:
https://www.linkedin.com/in/nikhileshtayal/

Common questions about the programs are answered here:
https://www.tisdoms.com/faqs-tisdoms-an-edu-tain-tech-platform-to-learn-ai/

Share it with your friends and colleagues

Nikhilesh Tayal

Nikhilesh Tayal

Articles: 17