When we use ChatGPT, there is a hidden environmental cost. As early as 2019, MIT Technology Review detailed how the process of training AI models like those powering ChatGPT emits more than 626,000 pounds of carbon dioxide equivalent, an amount almost five times the lifetime emissions of the average American car.
So, with the effects of climate change intensifying around the world, companies are under pressure to harness emerging technologies without sacrificing the environment. This pressure is markedly felt by Southeast Asian (SEA) organizations, which operate in one of the region’s most vulnerable to climate change.
In a virtual press briefing on last July 31, though, international cloud services provider Amazon Web Services (AWS) proposed a solution to SEA organizations’ sustainability dilemma.
AWS holds that migrating workloads onto the cloud and maximizing cloud solutions can empower SEA organizations to reduce carbon emissions, while also enabling them to capitalize on emerging technologies.
“There’s just a ton of data, analytics, and innovation, and organizations are going to need more compute and more storage,” said Ken Haig, AWS head of energy and sustainability policy in Asia Pacific, during the briefing.
“As we continue to embrace this digital transformation, we need to do so in a way that is sustainable. AWS believes that the Cloud is the most sustainable digital infrastructure available today,” Haig asserted.
The environmental benefits of cloud infrastructure are partly attributable to its scale. Compared to on-premises (on-prem) set-ups that require companies to manage their own servers, shifting workloads onto the cloud enables companies to access more efficient data center infrastructures and built-in sustainability tools.
In fact, a study by analyst firm 451 Research found that cloud data centers are five times more energy efficient than on-prem in Asia Pacific.
Haig described how the AWS Cloud enables SEA companies to improve their energy efficiency by giving them the option to choose serverless when possible, integrate AWS Instance Scheduler to shut down and terminate workloads when not in use, and use AWS Compute Optimizer for right-sizing recommendations of workloads.
Companies on the AWS Cloud also have access to databases and technologies such as AI, machine learning, and Internet of Things that can be used to build sustainability solutions for purposes like carbon tracking, energy conservation, and waste reduction.
However, out of sight should not mean out of mind. Just because computing is no longer occurring on-prem, does not mean that the process is not producing emissions or consuming resources. MIT reported that one data center can monopolize electricity equivalent to 50,000 homes and most data centers run on coal or petroleum-fuelled electricity grids.
Besides electricity, data centers also consume massive amounts of water to cool their servers, with a large data center capable of going through 1 million to 5 million gallons a day. Furthermore, data centers produce computer waste that take thousands of years to degrade.
Having said that, AWS has been taking steps to reduce their data centers’ carbon footprint, resource consumption, and waste while still meeting the surging demand for computing power driven by technologies like AI, machine learning, and the Internet of Things.
In terms of energy, AWS data centers are on the way to being 100 percent powered by renewable energy by 2025. The corporation is also streamlining its energy consumption by upgrading its chips.
For example, the AWS’ Inferentia machine learning chip optimized for large scale generative AI apps is up to 54 percent more energy efficient than previous models.
“Rapid innovation and the use of digital technology in data centers is already allowing the industry to offset huge increases in demand with improved infrastructure efficiency. We are building more and more digital capacity, modernizing legacy infrastructure and meeting the growth, but in real terms we’re not using much more energy than we were five years ago,” an AWS representative reported.
On the other hand, there is more work to be done to minimize water usage and waste. AWS has already started using recycled water at 20 of its data centers around the world, but it plans to expand its use to others.
AWS has also pledged to become water positive by 2030, which means it will be returning more water to communities than it uses in its direct operations.
Regarding waste, AWS’s two Reverse Logistic Hubs evaluate electronic equipment such as memory modules, networking switches, optical modules, and Nitro cards for reuse. Once deemed functional, the equipment is redistributed to AWS data centers or to third parties.
“Cloud computing really helps to democratize access to cutting-edge technologies,” Haig maintained.
“Whether its scaling proven solutions or accelerating existing solutions — like energy efficiency is a great one — or whether it centers around driving longer term, innovative solutions, like decentralizing, digitalizing, or thinking about how we move to a smarter grid and smarter energy deployment, these are all opportunities that we find in the cloud,” he affirmed.