Right now, as you read this sentence...

Your browser is burning energy
and no one will tell you how much.

Scroll to discover
↓

Every click. Every scroll. Every page load.

Energy flows from power plants, through cables, into data centers, across fiber optic lines, through your router, into your deviceβ€”

Just so you can see these words.

The internet consumes more electricity than

✈️
The entire aviation industry
πŸ‡¬πŸ‡§
The United Kingdom
🏭
Most countries on Earth
416
Terawatt-hours per year
And growing 10% annually

But here's what no one talks about:

You have no idea how much energy your browsing uses.
Is this news article using 10 millijoules?
Or 1,000?
Is that JavaScript animation draining your battery?
Are those autoplay videos wasting watts?

You can't know.

Because browsers won't tell you.

The Invisible Wall

What exists
CPU Power Sensors
GPU Power Rails
Battery Controllers
Energy Profilers
πŸ”’
Browser
Sandbox
What you get
Nothing

Your computer knows exactly how much power it's using. Your operating system can measure it precisely.

But the browser? It's locked away. Sandboxed. Silent.

This isn't a bug. It's a feature.

πŸ”

Security researchers discovered something terrifying:

By watching power fluctuations, attackers could extract encryption keys. Your power consumption was leaking your secrets.

πŸ‘οΈ

Advertisers found a new way to track you:

Your battery's unique drain pattern became a fingerprint. They could identify you without cookies.

πŸ›‘οΈ

So browsers did the responsible thing:

They locked it all down. Removed the Battery API. Blocked all hardware access. Problem solved.

Except now, no one can measure web energy.

And what you can't measure, you can't improve.

So we asked ourselves:

What if we could measure it anyway?

We can't measure power directly.

But we can measure everything else.

βœ“ How long things take
βœ“ How much data flows
βœ“ How many operations run
βœ“ What hardware you have

And here's the key insight:

If we run the same code on a server with real power measurement,
and then run it in your browser...
we can learn your device's energy signature.

Introducing: Energy Estimation

πŸ–₯️
Your Device
We detect your hardware through clues the browser does shareβ€” GPU model, CPU cores, memory size.
β†’
⚑
Calibration
We run standardized benchmarks and compare your timing to our server's real power measurements.
β†’
πŸ“Š
Estimation
Now we can translate your browser's activity into energy estimatesβ€”typically within 20-40% accuracy.
"But that's not perfect!"

It doesn't need to be.

You don't need millijoule precision to know that a 500 mJ page is ten times worse than a 50 mJ page.

You just need to see the difference.

What becomes possible

For You

See which sites drain your battery. Discover the hidden cost of your daily browsing. Make choices that matter.

For Developers

Get real-time feedback on energy impact. Optimize for efficiency, not just speed. Build websites that respect resources.

For the Planet

If every website reduced energy by 20%, we'd save 80 TWh annuallyβ€” equivalent to 56 million tons of COβ‚‚.

The web has been optimized for speed, engagement, and conversion.

Never for energy efficiency.

Because we couldn't measure it.

Now we can.

See the difference for yourself

A+
text.npr.org
8 mJ
A
lite.cnn.com
12 mJ
B
Wikipedia
45 mJ
C
GitHub
78 mJ
D
Medium
234 mJ
F
News sites (avg)
890 mJ

The difference between the best and worst? Over 100x.

Same information. Same purpose. Vastly different impact.

You can't optimize
what you can't measure.

Now you can measure it.

Every page you visit. Graded A+ to F.

Real-time energy monitoring.

The invisible, finally visible.

For the curious: How it actually works

Technical details for developers and skeptics

πŸ”§ Browser APIs we use
  • Performance API β€” High-resolution timing and resource measurement
  • PerformanceObserver β€” Real-time monitoring of long tasks
  • Navigator API β€” Hardware concurrency and device memory
  • WebGL β€” GPU identification via renderer string
  • requestAnimationFrame β€” Frame rate and rendering load
πŸ“ The energy model

We combine three estimation methods:

  • Time-based β€” Power correlates with CPU active time
  • Activity-based β€” Different operations have different costs
  • Calibrated β€” Adjusted by your device's measured performance

The final estimate weights each method by confidence.

🎯 Calibration process

We run standard benchmarks (Fibonacci, Mandelbrot, memory access) and compare:

  1. Your browser's execution time
  2. Our server's execution time (measured with powermetrics)
  3. The ratio becomes your calibration factor

Cached for 7 days, automatically refreshed.

πŸ”’ Privacy & security
  • 100% local β€” All analysis in your browser
  • No tracking β€” We don't know what sites you visit
  • No accounts β€” Install and use, nothing else
  • Open source β€” Audit every line of code
  • Minimal permissions β€” Only activeTab and storage