Can digitalization be climate-friendly?
In 2023, the world’s total electricity consumption was 29,471 terawatt-hours (TWh), and consumption is expected to increase by 3.3% annually. The internet consumes around 800 terawatt-hours in total, which is equivalent to running an air fryer non-stop for over 60 million years.
The internet accounts for approximately 2.7% of global electricity use, but there are alarming forecasts suggesting that this could double by 2030, being primarily driven by AI and crypto. And it’s also worth noting that any consumption looks small when measured as a percentage of total global usage. For example, air travel only accounts for 2-3% of global CO2 emissions. That’s why it's relevant to consider resource consumption and climate impact across the board – and reduce wherever possible.
When it comes to making digitalization more climate-friendly, Dwarf helps our clients in two ways. We advise on how the digital products we develop can have the lowest possible indirect energy and resource use, and we factor in resource consumption when prioritizing which digital projects create the most value. We’re not a sustainable or green agency, but like many others, we are concerned about climate change and therefore committed to doing what we can to drive the necessary changes.
How can we and our clients minimize the climate footprint of and through our digital solutions?
It's a big, important, and complex question, so we've tried to break it down schematically. There’s no one-size-fits-all solution, but there are many levers you can pull to significantly reduce the climate impact of your digital presence – and even more levers to ensure that digitalization actively helps reduce your company’s overall climate footprint.
We're not experts ourselves, so we've done our homework, cross-checked our points and data with experts, and included sources for everything – because we want to be 100% clear on what we, as a digital agency, can contribute.
As mentioned, there are many nuances and variables, but we’re starting from the top to make sure we cover everything. The menu on the right is ranked from low to high impact.
It might sound like an urban myth, but it’s true: colors affect energy consumption on screens. Dark colors use less power than bright ones on OLED screens, where dark mode can save up to 39–47% at full brightness, though only about 10% with normal use. On LCD/LED screens, the difference is minimal or even negative due to contrast needs. So, in short: keep your logo and corporate colors – the climate will survive.
There’s much more to gain in terms of energy savings by optimizing data handling than tweaking color schemes. At Dwarf, we use Cloudflare and other CDNs so users fetch data locally rather than from a central server, which saves traffic and energy, especially on data-heavy and rich content-heavy sites. Most hosting providers offer similar solutions. In addition, browser-based caching and local storage can reduce unnecessary data requests, especially in digital tools and self-service solutions where the software often weighs more than the data being collected.
The way a user interface is designed and coded also impacts energy consumption. Fortunately, good UX often aligns with energy efficiency. A streamlined user experience means fewer clicks and thus fewer unnecessary frontend and backend calls. How the frontend is coded also plays a big role: auto-scaling rich content, animating only what’s in the viewport, pagination, lazy loading, and minimizing server calls all contribute to making a solution faster and more energy efficient. The smarter the data handling, the less wasted computing power, benefitting both user experience and energy use.
There have been plenty of stories claiming our Netflix habits are a massive climate problem, but that’s (fortunately) exaggerated. Netflix uses around 1 TWh annually for its systems and cloud services, far from the wild estimates from the French think tank The Shift Project, which overshot by a factor of 100. That flawed estimate has, however, lived a long life online. In reality, the biggest sources of electricity use at home come from appliances, game consoles, and gaming PCs, being another story entirely.
Hosting choice may be the single most important factor in energy consumption and climate impact. Modern green cloud infrastructure uses far less energy than old-school on-premises hosting and local data centers. Major cloud providers are investing heavily in renewable energy and are approaching 100% self-supplied green power. They also have significantly better capacity utilization because they can scale globally and balance demand across time zones. A 2020 study even showed that 25% of all local servers could be turned off without any consequences. They’re simply left running because no one ever questioned it.
AI frenzy
There are concerning predictions circulating that AI will double the internet’s electricity consumption by 2030. That may well be true. AI is undeniably a game changer, but there are at least two strong counterforces.
First, processors, AI models and data centers are becoming significantly more energy efficient. In the year 2000, transferring 1 GB of data required 136 kWh. Today, it takes just 0.194 kWh - an improvement of more than 700 times. Earlier doomsday forecasts have also missed the mark: in 2007, the U.S. Environmental Protection Agency predicted that energy use would double within six years. Yet despite a 3,600 percent increase in data traffic, power consumption rose by only 10 percent.
Similarly, initial estimates showed that a ChatGPT request consumed about 2.9 Wh which is ten times more than a Google search. But new data, reported by TechCrunch in February 2025, indicates that newer AI models like ChatGPT 4.0 now consume roughly the same amount of energy as a Google search, while DeepSeek, depending on the complexity of the query, uses significantly less.
Second, there is a chance that overall energy consumption might actually decrease. Google's model is built around sending users on to advertisers and heavy, rich content, which consumes far more power than the search itself. In contrast, tools like ChatGPT and Perplexity aim to deliver a useful answer immediately. That saves clicks, navigation and time and therefore also energy. On the other hand, this creates new challenges around copyright and brand control, as users increasingly bypass the original sources.
AI will no doubt increase energy consumption. But the potential for optimization remains enormous. AI can be used to make better use of existing infrastructure and in the longer term, quantum computing could change the equation entirely. So no, the future is not pitch black - but we still need to optimize with everything we've got.
A little demonstration
We’ve built a small Dwarf-career chatbot where potential employees can get answers to questions about Dwarf as a workplace. It’s been trained in Dwarf’s tone of voice based on our 133-page values book and always references the relevant pages in the book. The experience is meant to be compared to regular search, a good FAQ, or perhaps an animated book-like interaction. We firmly believe there’s a strong connection between usability, “snappiness,” and energy efficiency. In this regard, AI clearly comes out on top, offering concrete answers, a seamless user experience, and high efficiency. Give it a try!
Our solutions
Depending on your current technical setup, there may be significant gains to be made by adjusting the levers we've already discussed. But the real potential often lies in a different perspective: how digital solutions can reduce your company’s overall climate footprint. It's not just about the internet, it's very much about your business, about better utilization of assets and capacity.
How can we minimize waste and eliminate repetitive tasks so time is spent on value-creating work? How can we optimize capacity utilization, maintain asset value, and ensure the maximum return on invested resources? With the right approach, digital solutions can not only streamline operations but also help build a more sustainable business.
Energy efficiency with BEMO
BEMO optimizes energy consumption in buildings using real-time data and intelligent algorithms. The system identifies and eliminates unnecessary energy waste while adjusting heating, ventilation, and lighting based on actual usage patterns. It reduces the carbon footprint, lowers operating costs, and supports more sustainable property management. BEMO is a clear example of how digitalization can improve energy efficiency and contribute to a greener future.
Resource optimization with Ejendom.com
Ejendom.com digitizes and streamlines property maintenance to ensure optimal use of resources. The platform provides a clear overview of both scheduled and urgent maintenance, which reduces waste, extends the lifespan of buildings, and minimizes unnecessary call-outs and material consumption. The result is both lower operating costs and a significantly reduced climate footprint. A great example of how digitalization can help preserve asset value and promote sustainable operations.
Efficient operations with Domea.dk
Domea.dk’s solution optimizes the planning and coordination of maintenance tasks across properties. By digitizing processes such as ordering, assigning, and executing contractor work, it reduces wasted time, unnecessary travel, and duplicate efforts. This leads to better capacity utilization, a lower CO₂ footprint, and more time for value-creating tasks, which is all powered by a digital setup that supports both operational efficiency and sustainability.
Sustainable Business Optimization with Comwell
Comwell’s digital platform ensures optimal utilization of capacity across the hotel chain’s rooms, meeting spaces, and facilities. By integrating data and systems across the chain, it minimizes idle time, resource consumption, and waste – for example, by only washing towels when necessary and ensuring all areas are actively in use. The solution creates business value while reducing the climate footprint. A strong example of how digitalization can support a for-profit and for-planet approach.
