By Dr. Leslie Kanthan, CEO and Co-founder, TurinTech
Enterprise technology is a big contributor to carbon emissions in most major sectors, and financial services is no exception. For example, according to McKinsey’s 2022 report The green IT revolution: A blueprint for CIOs to combat climate change, financial services’ IT technology emissions are especially high for insurance (45 percent of total scope 2 emissions) and for banking and investment services (36 percent). Second only to the media and communications sector in its emissions, financial services’ IT has a carbon footprint conundrum.
Finance is only one industry, and the sad fact is that enterprises globally are killing the planet, with new technologies such as AI introducing significant computational and energy demands. The MIT Technology Review reported that training just one AI model can emit more than 626,00 pounds of carbon dioxide equivalent – which is nearly five times the lifetime emissions of an average American car.
So, where should financial services companies focus? The answer lies in starting from the very beginning to optimize code to reduce the computational load of servers and data centers. If companies can lower the carbon emissions associated with data processing, storage, training AI models, and running software, they can ultimately improve their carbon footprint metrics.
What’s Causing the Financial Services Carbon Footprint Conundrum?
The finance industry is renowned for its heavy reliance on data and computational power. Critical tasks such as fraud detection, credit approval, and real-time payment processing are high-stakes and time-sensitive, and also consume vast amounts of energy.
Many headlines will have us believe that the solution to meeting energy efficiency goals lies in swapping out legacy infrastructure and adopting greener IT architecture. It is true that financial services institutions often struggle with outdated software systems that are difficult to maintain and susceptible to security vulnerabilities and inefficiencies. However, upgrading these systems is often both time-consuming and expensive, and the reality is that optimizing code is a better starting point.
Substandard code is a significant burden on both human and computational resources in software development. For example, in 2020, the Cost of Poor Software Quality (CPSQ) in the United States was an astonishing $2.08 trillion. This takes into account the costs of rework, lost productivity, and customer dissatisfaction due to inadequate code quality.
Despite the stats surrounding CPSQ, code optimization is often still overlooked in financial services. Many banks and insurers are yet to begin tackling the redundancies within codebases, and fewer still recognize the correlation between efficient code and computing energy savings.
By employing code optimization techniques, financial services institutions can seamlessly update their code to the latest language versions, enhancing performance and addressing vulnerabilities to make the code more resilient. Additionally, post-upgrade, these tools can analyze error logs and fix bugs that arise during the transition, ensuring a smooth upgrade.
Automating upgrades not only saves time and resources but also allows institutions to focus more on their core business activities.
Read More : Future-Proofing Your Fintech: Essential Strategies for Mitigating Risk
The Need for Banks to Move to Greener AI
Financial services is undoubtedly a leader in the application of Generative AI solutions: it’s driving efficiency in everything from risk management, sentiment analysis, document generation, and other operational processes.
However, it’s no secret that Gen AI requires a lot of energy for training and even more to produce answers to queries. Training OpenAI’s GPT-3 consumed 1,287 MWh of electricity, resulting in emissions exceeding 550 tons of carbon dioxide equivalent. This amount of emissions is comparable to one person making 550 round trips between New York and San Francisco. It has also been estimated that ChatGPT bills an eye-watering $700,000 a day to operate – a highly unsustainable model – and submitting just one GPT query will consume 15 times more energy than a Google search query.
The latest code optimization platforms can help by using advanced techniques and pre-trained LLMs to automatically scan entire codebases and identify inefficient code. These platforms recommend and benchmark optimal code changes to enhance performance, allowing developers to compare the improvements with the original code. This accelerates application execution time and significantly reduces compute usage and costs, achieving the same tasks with much greater efficiency.
AI code optimization can result in an average 46% reduction in memory and energy consumption and deliver cost savings of nearly $2 million annually for AI model production and deployment. This optimization also has the potential to save 256 kg of CO2 emissions per year, helping to meet sustainability targets more quickly.
Getting Proactive About IT-focused Emissions Reduction
When it comes to IT-focused emissions reduction, financial services leaders cannot afford to bury their heads in the sand – just like regulation surrounding the ethics of AI, we’ll soon see regulation governing the use of computing power surrounding AI. In fact, according to a 2024 report from the University of Cambridge, which includes an assessment of the current state of AI governance, computing power is the most likely for future policymaking.
As financial software challenges evolve, advanced code optimization techniques can empower banks and other financial services organizations to streamline code management, and enhance performance while also meeting ESG goals.
Read More : Global Fintech Series Interview with Daniel Muller, CEO of Aeropay
[To share your insights with us, please write to psen@itechseries.com ]
