How Much Electricity Does a Website Use? Éclair and WordPress in Numbers

How do you estimate a website’s energy consumption?

The most widely used method is the Sustainable Web Design model, version 4. It assigns each gigabyte transferred an operational energy of 0.055 kWh in data centers, 0.059 kWh in networks and 0.080 kWh on visitors’ devices, for a total of 0.194 kWh per GB. Adding the energy used to manufacture the hardware brings this to 0.300 kWh per GB^[1].

The model has limits, and its authors spell them out: it tends to overestimate actual emissions, and it is a top-down model that cannot finely estimate any single part of the system. It is also the method used by the Website Carbon Calculator, which describes it as giving “a rough idea”^[2]. So I use it for an order of magnitude, not as a measurement.

How much less data does Éclair transfer than WordPress?

In the public benchmark, Éclair 2.0.0’s article page makes the browser download 10.8 KB, versus 97.6 KB for WordPress 7.1.4 with Cache Enabler: 86.8 KB less, a ratio of 9 to 1^[3]. These are two fresh installs with their default themes, on a text page.

So this gap is a floor that only measures the overhead of the CMS and theme. A real WordPress site is usually much heavier: according to the HTTP Archive 2025 Web Almanac, the median WordPress page weighs 2,894 KB on mobile, including 638 KB of JavaScript^[4]. Most of that weight comes from images, scripts and plugins, which no CMS makes disappear on its own.

How many kWh, euros and kilograms of CO2 does that add up to?

Here is the formula: kWh per year = page views per month × 12 × 86.8 KB ÷ 1,000,000 × 0.194. For euros, I use the Eurostat household electricity price for the first half of 2026, all taxes included: €0.2485 per kWh in France and €0.2509 in Spain (€0.2832 for the European Union)^[5]. For CO2, the 2025 lifecycle carbon intensity: 41.44 g per kWh in France and 153.6 g in Spain^[7].

Those euros do not come out of your pocket: they are spread across the hosting provider, network operators and visitors. With the model’s broadest assumptions (manufacturing included, 0.300 kWh per GB, and the default global intensity of 494 g per kWh), you get 1.5 kg of CO2 a year at 10,000 page views a month and 15.4 kg at 100,000. With the direct intensity of French electricity generation, 19.6 g per kWh in 2025 according to RTE^[6], it drops to 0.04 and 0.40 kg.

The choice of carbon intensity alone changes the result by a factor of 25. That is why a CO2 figure without its method means nothing: the same site hosted in France looks almost neutral, and the same site assessed with the global average looks heavy.

What about the server’s consumption?

The benchmark measures neither watts nor joules, so I have no direct measurement of the energy Éclair uses. The best public reference is a Green Coding case study: about 30 joules per request for WordPress, versus 1.68 joules for a static site (Hugo), with measurements updated in May 2024 using WordPress 6.5.3^[8]. A cached Éclair page is served by a single PHP file reading a ready-made HTML page, which is close to static; that is an assumption, not a measurement.

For an uncached WordPress page, the formula becomes: kWh per year = page views per month × 12 × (30 − 1.68) J ÷ 3,600,000. Result: 0.94 kWh a year at 10,000 page views a month, 9.4 kWh at 100,000, which is €1.93 in France and €1.54 in Spain at the Eurostat non-household price, excluding VAT^[5]. Against WordPress with a caching plugin, the gap becomes negligible in absolute terms. As Green Coding points out, a server that is switched on already uses a lot of energy at idle, so the real saving is often smaller.

Where is the real gain?

The solid numbers are not kWh but ratios measured on the same machine. Éclair makes the browser download 9 times less data per page. Without a cache, it builds a page in 3.37 ms versus 75.72 ms, about 22 times less computing time. With a cache, it serves 1,997 requests per second versus 1,060: the same server handles about 1.9 times more visitors, with no database server to run^[3].

The most tangible benefit is for your mobile visitors. With 86.8 KB less per page, one gigabyte of data plan is saved roughly every 11,500 page views, and simulated LCP on slow 4G drops from 1.50 s to 0.95 s. The case studies collected by Google on web.dev show that speed translates into revenue: at Vodafone, a 31% improvement in LCP increased sales by 8%, and the BBC lost an additional 10% of users for every additional second of loading time^[9]. Those are their results, not Éclair’s.

And across 1,000 or 10,000 sites?

This is where the small numbers add up. A projection, not a measurement: if 1,000 sites with 100,000 page views a month switched from a cached WordPress to Éclair, the transfer saving would be about 20 MWh a year; with 10,000 sites, about 200 MWh, roughly one petabyte less data a year and some €50,000 at the French household price. To put this in perspective, the household consumption band used here in Eurostat statistics runs from 2,500 to 5,000 kWh a year: 200 MWh is the annual consumption of 40 to 80 such households.

This projection relies on a minimal weight gap, measured on fresh installs: on real sites, the gaps can be much larger, or smaller if the starting WordPress site is already highly optimized. It is only valid as an order of magnitude.

How much does a WordPress or Éclair site cost per year?

On shared hosting, the price is the same for both. The Cloud plan from the French host o2switch costs €1.86 excl. VAT per month in the first year (€22.32 excl. VAT for 12 months), then €16 excl. VAT per month, or €192 excl. VAT a year; MySQL databases are included, so WordPress runs there just as well as Éclair^[10]. Éclair’s advantage on this kind of plan is not price but having no database to manage, and more headroom. At the other end, managed WordPress hosting such as Kinsta starts at $35 a month ($280 a year when billed annually) for 35,000 monthly visits^[11]; Éclair does not need it.

The real difference lies in premium plugins, if you buy them. Prices checked on November 9, 2026 on the official pages, for one site: WP Rocket at $5 a month regular price, or $60 a year ($3 a month for the first year during a promotion running until November 28)^[12]; Yoast SEO Premium at $118.80 a year excluding VAT^[13]; Wordfence Premium at $149 a year^[14]; UpdraftPlus Premium at $70 a year^[15]. Total: $397.80 a year. A trade site reports that Wordfence Premium will move to $199 on December 3, 2026, which I could not confirm on Wordfence’s own site^[16].

Two honest caveats. All these plugins have a free version, and many WordPress sites make do with it. And Éclair’s built-in features are not identical, feature for feature, to these premium products: they cover the essentials for a small site, not everything each plugin offers.

My take

If someone asks me “Does Éclair save electricity?”, my answer is: yes, but for a small-business site the saving is counted in kWh and euros per year, not in hundreds. I advise against selling a CMS switch on that argument alone. The solid reasons are the speed perceived by mobile visitors, server capacity, simpler maintenance and, if you were buying premium plugins, a few hundred dollars a year.

The energy argument becomes serious at the scale of a hosting provider, an agency or a portfolio of sites: that is where 9 times less data and 1.9 times more visitors per server turn into megawatt-hours and fewer machines. I cover the other measured gaps in my Éclair CMS vs WordPress comparison.

What should you do in practice?

1. Measure the real weight of your main pages in PageSpeed Insights, in mobile mode.
2. Apply the formula: page views per year × MB per page ÷ 1,000 × 0.194 kWh. Each MB saved per page at 10,000 page views a month represents about 23 kWh a year.
3. Tackle images and third-party scripts first: that is where most of a real site’s weight lies.
4. Before migrating, measure before and after on your own site rather than relying on an average.
5. If you publish a CO2 figure, always state the method and the carbon intensity used.

Sources

1. Sustainable Web Design, Estimating Digital Emissions (v4 model)
2. Website Carbon Calculator, How does it work?
3. Eclair Benchmarks, results of November 8, 2026 (GitHub)
4. HTTP Archive, 2025 Web Almanac, CMS chapter
5. Eurostat, household electricity prices (nrg_pc_204), H1 2026 and non-household prices (nrg_pc_205), H1 2026
6. RTE, 2025 Electricity Report, generation emissions
7. Our World in Data, Carbon intensity of electricity (2025)
8. Green Coding, WordPress vs Hugo: energy per request (updated May 2024)
9. web.dev, Why speed matters
10. o2switch, hosting plans
11. Kinsta, pricing
12. WP Rocket, pricing
13. Yoast SEO Premium, pricing
14. Wordfence, plans and pricing
15. UpdraftPlus, pricing
16. WebHosting.Today, reported Wordfence price increase (November 4, 2026)