Note
NOTE: gCO2eq/kWh stands for grams of carbon dioxide equivalent per kilowatt-hour. It is a unit of measurement used to express the carbon intensity or carbon emissions associated with electricity generation.
20 minutes read
Green IT encompasses various strategies and initiatives to foster sustainability within the technology sector. It involves adopting sustainable practices throughout the lifecycle of IT products and services.
Big players such as Microsoft Azure, AWS, and Google Cloud have the financial background and the responsibility to minimize their carbon footprint. They engage in various ways to do so.
So, how can you influence your organization’s carbon footprint? The weight of Green IT initiatives lay not only on the shoulders of cloud providers when it comes to software. There are some things you can do that benefit the environment and your business.
Green Information Technology (Green IT) adopts environmentally responsible technologies and practices to mitigate the environmental impact of IT products and services throughout their lifecycle. Green IT strives to optimize resource efficiency, minimize energy consumption, and mitigate electronic waste while maintaining or improving overall operational performance.
At its core, Green IT significantly emphasizes sustainability, aligning technological advancements with environmental stewardship. It highlights the importance of adopting environmentally conscious practices across the entire IT lifecycle. This means you can help the environment and your business as well.
As we want to look at modern Green IT initiatives, it's instructive to look back at the origins of these initiatives. In the 1990s, Europe and the United States began pioneering efforts to weave environmental responsibility into the very fabric of the technology sector. However, despite these initiatives, there has been a concerning uptick in CO2 emissions since the 1990s, a trend that predates even the turn of the century.
Global CO2 emissions from energy combustion and industrial processes, 1900 - 2022 (Source)
The historical initiatives laid a foundation for today's Green IT strategies, yet they also bring into sharp focus our current challenges – especially the power use within the IT sector.
The significance of electricity consumption lies in the availability of more or less sustainable methods of power production. In 2020, IT represented up to 7% of global electricity usage, measured using diverse methods. Projections indicate an expected increase of 13% by the year 2030.
Despite available energy-saving technologies, IT systems are at a critical point concerning current and future energy consumption in telecommunication networks, data centers, and user-related devices.
As we've seen, power use in IT is a significant issue, now and for future projections. Researchers like Andrae and Edler’s trend analysis offer valuable insights with trend analysis, shedding light on the potential path of IT sector power consumption through to 2030.
Most researchers agree that data traffic, defined in various ways, will continue to increase exponentially, mirroring the past decade's trends.
So let’s have a closer look at their predictions and numbers.
Although it shows that the power consumption will still rise, the best-case scenario is lower than expected.
Electricity usage of data centres 2020 - 2030 (Source)
Andrae and Edler compared these results to a previous study they have conducted.
Although energy consumption rises, the values for 2020 exhibit a decrease for most entities in general. This trend continues into 2030. The overall electric power predictions for the IT sector have experienced a reduction of 31% in 2020 and 61% in 2030, compared to the expected scenario.
Predicting global IT electric power use is inherently challenging due to uncertainties and unknowns. While understanding data center power usage is crucial, we must examine another burgeoning area with a substantial energy footprint – artificial intelligence and machine learning models.
Machine learning (ML) advancements bring environmental implications in Green IT, given the substantial computational resources, energy, and materials required for training ML models. Among the largest ML models are Large Language Models (LLMs), boasting hundreds of billions of parameters and demanding millions of GPU hours for training, contributing to carbon emissions.
For instance, let’s have a look at BLOOM's training statistics, according to the article estimating the carbon footprint of Bloom, a 176b parameter language model.
Total training time | 118 days, 5 hours, 41 min |
Total number of GPU hours | 1,082,990 hours |
Total energy used | 433,196 kWh |
GPU models used | Nvidia A100 80GB |
Carbon intensity of the energy grid | 57 gCO2eq/kWh |
If we use the formula [Total Emissions = 433,196 kWh × 57 gCO2eq/kWh], we find that Bloom emitted 25 tons of CO2 over 118 days when being trained. We see a big difference when we compare it to the average annual CO2 emissions of a U.S. household with 16 tons or the European average of 8 to 10 tons.
NOTE: gCO2eq/kWh stands for grams of carbon dioxide equivalent per kilowatt-hour. It is a unit of measurement used to express the carbon intensity or carbon emissions associated with electricity generation.
Since we have more or less some figures for comparison, let’s have a look at other LLM’s emissions and if they favor or disrupt Green IT initiatives. Bear in mind that it is hard to compare the numbers since the industry lacks standardized accounting methodologies for reporting carbon emissions. We are depicting the numbers collected in a study by Cornell University.
LLM | Carbon Intensity | Power Consumption | CO2eq emissions |
BLOOM | 57 gCO2eq/kWh | 433 MWh | 30 tons |
GPT3 | 429 gCO2eq/kWh | 1287 MWh | 552 tons |
Gopher | 330 gCO2eq/kWh | 1066 MWh | 380 tons |
OPT | 231 gCO2eq/kWh | 324 MWh | 76 tons |
What we can see is that LLMs contribute massive amounts of CO2 emissions. This means we must find ways to lower our carbon footprint while our productivity grows.
The environmental impact of machine learning models is straightforward, prompting us to the impact of existing technologies. Understanding the environmental implications of these technologies is crucial for fostering a sustainable future with Green IT.
We already know how CO2 emissions have grown since 1990 and the impact of AI models on the environment. Now, let's delve into how software, data centers, and algorithmic efficiency contribute to the global footprint of technology.
Software lies at the heart of our digital experiences, influencing the energy consumption of devices running applications. The design, coding practices, and functionality of software directly impact the power requirements of our devices.
However, there is hope in optimization – by refining code and implementing efficient algorithms, we can significantly diminish energy demands, paving the way for Green IT initiatives and a more energy-conscious digital era.
Software design plays a critical role in energy consumption, which leads us naturally to a related topic: the interdependent relationship between software and data centers, which collectively form the backbone of our technological infrastructure.
As a side note, measuring the real number of power uses, including both software and cloud, becomes hard. But we will delve into the topic later in more detail.
Algorithms within software applications emerge as silent architects of resource utilization, significantly impacting energy consumption.
Ultimately, allowing your developers to create suboptimal applications results in an increased burden on environmental costs further along the process.
Given the intertwined roles of software, data centers, and algorithms in energy consumption, you might wonder about practical steps that you can take. Indeed, small actions are steps in the right direction, but broader strategies are needed for substantial Green IT implementation.
A study from Cornell University suggests four approaches to reducing the carbon footprint of ML models and contributing to Green IT initiatives. Referred to as the “4Ms” they can significantly reduce the energy usage and carbon emissions associated with training machine learning (ML) models:
Initiatives like the Web Sustainability Guidelines (WSG) 1.0 emerge as tools to help organizations improve their green IT efforts. The WSG guidelines offer a comprehensive framework to enhance the sustainability of websites and digital products. Built on the principles of environment, social, and governance (ESG), these guidelines integrate seamlessly into decision-making processes throughout the development lifecycle.
Developers and businesses can effectively reduce their environmental impact by adhering to these guidelines. The guidelines encompass a holistic approach, integrating user-centered design, performant web development, renewable infrastructure, and sustainable business strategy. Metrics play a crucial role, allowing for measuring progress in adopting these principles.
It's essential to recognize that while these guidelines offer valuable insights for web sustainability, they may only address some possible mechanisms or strategies for achieving sustainability in the web domain. Adhering to the Web Sustainability Guidelines often results in additional benefits beyond sustainability. Web content becomes more accessible, usable, and performant, contributing to an overall improved user experience.
WSG 1.0 recommends a methodical approach to implement these guidelines effectively. Still, the WSG suggests picking and choosing the guidelines aligning with the developer’s skill set or choosing the ones feasible based on impact and effort considerations. Although it gives organizations some ground rules, it is an unofficial document and offers organizations the possibility to self-regulate.
Having explored various Green IT strategies, let's pivot to consider the advantages they bring. Green IT offers benefits extending across economic and social domains, reflecting its comprehensive impact.
Cost savings through sustainable software solutions—Green IT initiatives significantly reduce operational costs by optimizing resource consumption. By increasing the performance of your application or cloud solutions, you can significantly reduce energy consumption, and extend the lifespan of IT equipment. Such measures lead to substantial long-term savings while promoting environmental responsibility.
Scaling up with efficiency—Implementing Green IT practices facilitates scalability without proportional energy consumption or resource utilization increases. This scalability enables you to expand your organization’s operations seamlessly while maintaining operational efficiency. One potential solution is serverless computing, which empowers developers to expedite application development by eliminating the requirement to oversee infrastructure management. Whether it involves upgrading infrastructure or optimizing applications, the efficiency-driven approach allows your organization to grow sustainably.
Improved brand reputation—Green IT initiatives enhance your organization’s brand image and market competitiveness. Consumers are increasingly inclined towards eco-conscious businesses that prioritize sustainability. The green strategy also garners positive attention from stakeholders, investors, and potential partners. This, in turn, fosters loyalty and trust, ultimately benefiting the business's bottom line.
Contribution to an eco-friendly environment—Green IT practices directly contribute to minimizing the environmental impact of technology on ecosystems. Software developers play a crucial part in sustainable practices within the tech industry. One of the most direct ways they can influence this is through the optimization of the performance of their applications. This means in the broader picture, more efficient code equals less energy consumption.
Encouraging a culture of sustainability—Organisations typically fulfill the basic requirements to comply with legal frameworks. While companies may declare reductions in emissions or claim carbon neutrality, these assertions often hinge on selective data interpretation or offsets that might not effectively reduce the company's actual carbon footprint. Instilling a culture of sustainability among your employees and stakeholders can inspire positive changes in their personal lives, fostering a ripple effect.
Need for experts in their field—As organizations prioritize energy efficiency, renewable energy integration, and eco-friendly practices, there is a growing demand for professionals skilled in sustainable technology, green data management, and environmental compliance. If you seek Green IT expertise, you might want to think about collaborating with a capable service partner.
While the benefits of Green IT are extensive, they come with certain challenges. The path toward sustainable IT practices is paved with complex obstacles that organizations must navigate.
Software significantly contributes to CO2 emissions, impacting the climate and increasing the likelihood of extreme weather events. The carbon footprint of the cloud, growing by 9% annually, exceeds that of the airline industry.
Digital transformation holds the potential for mitigating climate impacts, but it comes with challenges. Every digital transaction requires energy, physical infrastructure, and a device, impacting raw material manufacturing, transportation, maintenance, and end-of-life processes.
Every organization faces several challenges.
As organizations grapple with the challenges of Green IT, they also operate within a broader context of legal frameworks and compliance requirements. The legal framework for green efficiency is the right step towards a sustainable future. These guidelines set a standard for organizations to meet in advancing their energy efficiency. However, relying solely on these standards may hinder achieving a sustainable future.
The volume of legislation and agencies has skyrocketed in recent years in America and Europe. Germany, for instance, submitted a proposal for an amendment of the Energy Efficiency Directive. EU's Corporate Sustainability Reporting Directive (CSRD) published in December 2022, applies to around 50,000 companies listed in an EU operating in Europe and demands the inclusion of sustainability reports.
The goal is to reduce energy consumption by 50 terawatt hours by 2030. Some describe it as an ambitious project, comparable to the goals of the German Climate Protection Act, which mandates a reduction in final energy consumption of over 550 terawatt hours by 2030 compared to 2008 levels.
However, the rise in regulations sometimes means more effective or actionable directives. Instead, these laws may set broad objectives without detailing the necessary actions and criteria for achieving them. This vagueness forces companies to independently interpret and apply these rules, leading to a self-regulatory approach.
The pitfall of self-regulation is the potential for a wide range of compliance levels. Companies might opt for the easiest or cheapest route of Green IT initiatives, potentially compromising the legislation's purpose. With definitive standards or benchmarks, assessing progress or holding organizations accountable for their environmental impact becomes easier.
For instance, it took legal actions initiated by Austrian activist and lawyer Max Schrems for data protection authorities (DPA) to start imposing fines and enforcing regulations under GDPR. This raises a question: Could the enforcement of Green IT regulations unfold similarly? However, the industry still needs to consider this scenario.
To truly encourage impactful change, legal frameworks should include specific, enforceable guidelines that businesses can practically follow. For now, it's crucial to bridge the gap between the ambitions of green regulations and their real-world implementation. This gap can be filled by software engineering partners who have the expertise in performance enhancement of IT solutions.
Adopting eco-friendly practices might become a checklist for compliance rather than driving genuine change, as we can see with the enforcement of GDPR. Data Protection Authorities (DPAs) impose fines when organizations fail to comply with GDPR. This leads organizations often to focus on penalty avoidance rather than embracing the regulation's true intent.
Such a checkbox mentality may lead companies to fulfill only the minimum compliance requirements without fully integrating the regulation's principles into their business model. Similarly, businesses might claim to be eco-friendly by meeting certain standards or regulations.
However, if the compliance is superficial, it can result in "greenwashing" rather than authentic sustainability. Companies may reduce emissions in minor areas while neglecting larger, more consequential parts of their operations.
Beyond greenwashing concerns are the broader issues related to the transparency and accountability of environmental reporting in the IT sector. Greenwashing affects the measuring of emissions as the corporate world increasingly promotes its commitment to sustainability. Companies often hype figures about their Green IT initiatives in their sustainability reports and marketing materials.
Claims of reduced emissions or carbon neutrality often rest on a selective interpretation of data or on offsets that only indirectly impact their reduction in the company's actual carbon output. The credibility of sources providing these figures comes into question when there's a lack of independent verification or when the methodologies used are not openly shared to be looked at by third parties. As for now, the European Commission mandates to report on energy efficiency investments. This means no palpable measurements are mandatory.
Therefore, a healthy skepticism about the reliability of the data and figures reported by the industry is more than understandable. This distrust stems from a need for more transparency and standardized methodologies in how these figures are calculated and reported. Subsequently, it may lead to inconsistencies and potential manipulation of data to present a more environmentally friendly image than is accurate.
These concerns underline the need for standardization in reporting practices, drawing on robust and transparent metrics that can be universally applied. They also call for third-party verification and a critical assessment of emissions data. You may find guides on energy efficiency evaluation and measurement, as provided by the EPA in the US, yet need to be revised, leave room for interpretation, and need to control the numbers produced by an organization.
Guidelines such as WSG provide a roadmap for current practices, yet looking ahead is just as important. What does the future hold for Green IT, and what trends will define the next generation of sustainable technology?
Exploring the trajectory of innovation and its implications for sustainability becomes imperative. The fusion of technology and environmental consciousness is giving rise to a new era in IT.
Emerging technologies promoting Green IT—The technology landscape is evolving, with a clear emphasis on sustainability. Innovations are emerging that not only advance the capabilities of IT but also contribute to environmental well-being. These technologies aim to strike a delicate balance between progress and ecological responsibility.
Centralization grows even bigger—Major players, such as Microsoft, explore centralizing their data centers and AI operations. This approach optimizes resource utilization and aligns with the overarching goal of reducing the environmental footprint of IT operations. On the flip side, it also creates a monopolism for some big players.
Microsoft is exploring groundbreaking solutions for powering its data centers and AI through next-generation nuclear reactors. The company's commitment to sustainability is reflected in this strategic move, as nuclear energy is renowned for its minimal greenhouse gas emissions. However, this endeavor also raises critical questions about managing radioactive waste and developing a sustainable uranium supply chain.
A noteworthy anticipation within the IT realm is the return to larger high-performance solutions. As technology continues to evolve, there is a growing expectation that larger, more powerful systems will play a pivotal role in driving efficiency and performance. Balancing this with ecological considerations will be paramount in shaping the future of IT infrastructure.
Looking forward, the IT industry anticipates significant advancements in eco-friendly practices. From energy-efficient hardware to sustainable software solutions, the collective effort is geared towards minimising environmental impact. The integration of renewable energy sources and the adoption of circular economy principles are expected to further enhance the sustainability quotient of IT operations.
We acknowledge Green IT’s vital role in marrying technological progress with environmental stewardship. At the heart of Green IT lies the dual promise of economic efficiency and social responsibility, presenting a compelling case for businesses to adopt eco-friendly practices. The benefits are clear, from the significant reduction in operational costs to fostering a culture of sustainability. Yet, challenges such as initial investments, infrastructure compatibility, and organizational resistance call for strategic approaches and incremental improvements.
Still, when creating software solutions, you can positively and actively impact the environment and your business. Better algorithms and code are efficiency drivers that will determine the future since cloud providers will react to the needs of their customers. This means technological know-how is paramount in decreasing the carbon footprint to enable a continuous commitment to a sustainable future for everybody.
Curated by
Sebastian Synowiec