Addressing the Carbon Footprint of Video Processing and Streaming
Dan Patton, Vice President of the Content Delivery Network at Ateme, delves into technological innovations underway to advance decarbonisation efforts.
In the digital age, video processing and streaming have become integral components of everyday life. From entertainment to education, the demand for high-quality video content is skyrocketing. However, video streaming is a data-intensive activity that requires large amounts of information to be transmitted over the internet.
Internet activity accounts for around 4% of global greenhouse gas emissions, with streaming media contributing 60-80% of total internet traffic – and this continues to grow. The surge in demand has led to an increase in data centres, network infrastructure, and end-user devices, all of which add to the industry’s carbon footprint. According to research by the International Energy Agency (IEA) on average video streaming habits, viewing devices account for most energy use (72%), followed by data transmission (23%) and data centres (5%).
There are many investors who are invested in or are considering investing in companies in the media and entertainment, technology, and online gaming sectors when decarbonising their portfolios. A high percentage of these companies’ carbon emissions – particularly Scope 3 – will come from video processing and streaming. So understanding how technology is being used to reduce the carbon footprint is essential when assessing the long-term sustainability and risk profile of these current and potential investment opportunities.
The following sections outline some key areas being researched and implemented to address video processing and streaming emissions.
Utilising codecs
Codecs like H.264 (AVC), H.265 (HEVC), and the newer AV1 and Versatile Video Coding (VVC) are designed to compress video files by removing redundancies and unnecessary data. HEVC can reduce file sizes by up to 50% compared to its predecessor, H.264, while maintaining similar quality. The development and adoption of newer codecs like AV1 and VVC are an important piece of the puzzle to reduce the carbon footprint of video streaming, as both offer 30-50% better efficiency than HEVC.
This increase in video processing efficiency has a cascading effect on energy consumption. By compressing video files more effectively, advanced codecs reduce the amount of storage space needed on servers and by data centres. By reducing the overall throughput to deliver streams to a wide audience, advanced codecs decrease the amount of CDN capacity and bandwidth required for the network. Although this will not directly reduce the power consumption of static network infrastructure components, the adoption of advanced codecs will reduce the growth rates of those network components as compared to overall growth of the streaming video market.
Despite the advantages, the adoption of newer codecs faces challenges, including the need for updated hardware and software support. Devices and platforms must be capable of encoding and decoding new codec formats, with a desire to leverage hardware video decoding – as it requires far fewer resources (4-10x) than software decoding.
For instance, Apple’s iPhone 15 added support for AVC hardware decoding, and Qualcomm has indicated that future versions of the Snapdragon processor releases will support VVC decode. While the adoption of these new devices will increase over time, streaming services will continue to provide content in H264 and HEVC formats for legacy devices.
Leveraging viewer data and AI
In the legacy broadcast scenario, linear content was broadcast over-the-air, and determining how many people watched the content was an after-the-fact estimate. Compare that to the streaming world – there is data about every single viewer: when they started watching, how long they watched, what device they used, and what profiles were delivered to them.
Given all this data about the streaming audience, what can be done with it? It can be used to determine if there are enough viewers using devices with advanced codecs. If the trade-off of energy savings from these devices using a more efficient codec is greater than the energy cost to process the video in an additional format, this demonstrates a reduction in the carbon footprint of the audience watching the content. Leveraging other data sources and training AI models to make these dynamic decisions provides a view into the future.
Data can also be used to make better decisions during the encoding process. By leveraging a vast history of video quality test encodings – which have been analysed by highly skilled video quality engineers – machine-learning techniques can be used to train algorithms to match outputs to the visual quality evaluations of the experts. When combined with other video processing algorithms, not only is the quality of experience improved, but, more importantly from a carbon footprint perspective, it can recommend the use of fewer variants and/or lower bitrates for each variant. That means lower storage space (up to 50%) and less network bandwidth to deliver the video experience.
Reducing cloud computing requirements
Data centres are the backbone of video streaming services. They store and process massive amounts of data, ensuring that video content is readily available to users. But they consume vast amounts of electricity, primarily for powering servers and cooling systems.
By optimising the efficiency of video processing software and leveraging elastic CDN algorithms for video delivery, companies can reduce the computational resources needed for video encoding, decoding, and transcoding, thereby lowering energy consumption and carbon emissions.
Cloud providers and processor vendors are also investing to reduce their energy footprint. AWS markets that their Graviton-based compute instances use up to 60% less energy than comparable configurations for the same performance. Processor vendors are also investing to reduce power consumption when the server is in idle state, including machine-learning algorithms to predict workload patterns and optimise power usage dynamically.
The industry is also reducing emissions by transitioning to renewable energy sources. Companies like Google and Amazon, which own and operate large data centres, have committed to using 100% renewable energy. In 2020, Google announced that it had achieved carbon neutrality and planned to run all its data centres and offices on carbon-free energy by 2030.
Additionally, techniques such as advanced cooling systems, server virtualisation, and AI-driven energy management can help lower energy consumption.
Making changes
The video processing and streaming industry, while indispensable in the modern world, has a significant carbon footprint.
Through the adoption of more efficient codecs, leveraging viewer data and AI, and optimising data centres and network growth, it is possible to reduce the environmental impact of video processing and streaming. But these changes require industry-wide cooperation, advancements in technology, and a commitment to sustainability.
As companies increasingly include Corporate Social Responsibility (CSR) and commitments to reduce their carbon footprint as integral components of their business strategies, investors also have the power to engage with them, making it clear that investing is dependent on the company consistently demonstrating their commitment to sustainability.
The post Addressing the Carbon Footprint of Video Processing and Streaming appeared first on ESG Investor.