At a hospital down the street, a physician turns to an AI-powered full-body MRI to not only capture images of the patient but also analyze the scans to detect cancer earlier, which increases survival rates. A few counties over, a farmer uses AI to determine whether they are using the best soil combination, spot potential disease outbreaks and predict how many ears of corn this year’s crop will yield. Inside people’s homes, it’s increasingly likely that AI-powered devices — from a video doorbell to a smart thermostat to a voice assistant — help families efficiently manage their household needs.
Vention reported that more than 80% of businesses use AI as a core technology within their organization, with 20% using generative AI regularly within their processes. While AI provides many benefits, such as saving money through automation and smarter decisions with better insights from data, the usage also comes at a cost: more required computing power.
According to David Porter, a vice president at the Electric Power Research Institute, a ChatGPT query requires 10 times more energy than a typical Google search. Porter told Time magazine that 10%-20% of data center energy in the U.S. is currently consumed by AI, and that percentage will likely “increase significantly” going forward. Additionally, Time reported that the cost of computational power required to train the most powerful AI models has doubled every nine months.
While cloud computing can help reduce energy consumption, many companies are continuing to build more on-premises data centers. Decode39 states that building a data center sometimes requires more power consumption than that of an entire African country. However, Microsoft is currently partnering with BlackRock to jointly invest $30 billion in improving AI and cloud capabilities in Italy, which includes building more data centers.
When it comes to building new data centers, access to ample energy sources has grown increasingly limited, especially for smaller companies, Harmail Chatha, senior director, global cloud operation at Nutanix, told The Forecast.
“From a data center perspective, it’s a sellers’ market, not a buyers’ market,” Harmail said.
“There just isn’t any capacity available. Hyperscalers are taking up all the data center capacity, which can take up to three to five years to invest.”
But the ferocious need for power is impacting the big cloud computing providers, too.
“I've been doing this for a really long time now, and I've never seen it like this where it's a seller's market, but even the sellers are constrained by the power companies’ capacity.”
Potential negative impacts of cloud computing
While technology can provide solutions for managing or limiting environmental impact, it also negatively impacts the environment. In a report titled AI & Energy, The World Economic Forum reported that the increasing energy demand, primarily from building and running the data centers used to train and operate AI models, is contributing to global greenhouse gas (GHG) emissions.
“Microsoft, which has invested in ChatGPT maker OpenAI and has positioned generative AI tools at the heart of its product offering, recently announced its CO2 emissions had risen nearly 30% since 2020 due to data centre expansion,” stated the report. “Google’s GHG emissions in 2023 were almost 50% higher than in 2019, largely due to the energy demand tied to data centers.”
This is having a significant impact on the environment and cost of doing business.
“I feel like sustainability is going to get blown out of the water with just the amount of power consumption required by GPUs and CPUs,” says Chatha. “We are at the point that our power bill is more than the rent in some of our data centers. While some data centers offer renewable energy, some don’t have it available.”
The growth of cloud computing in recent years is driving data centers to consume about 2% of global electricity today. Concerningly, that number could rise to nearly 8% by 2030, according to research by Huawei Technologies Sweden. Additional concerns about cloud computing include data security, compliance issues and less control by the organization.
But The World Economic Forum reported that while the computational power needed for sustaining AI’s growth is doubling roughly every 100 days, there are predictions that AI has the potential to help mitigate 5-10% of global GHG emissions by 2030.
Sustainability an increasing priority at organizations
Many companies are building AI into their organizational processes, which requires more energy during a time of greater focus on improving sustainability. According to KPMG, 79% of the top 100 companies in various countries (N100) report on sustainability. Organizations must report all environmental impacts of their processes, which includes those of third-party data centers used for cloud computing. Many organizations also turn to technology to measure environmental impact, which is required for sustainability reporting.
“Organizations developed tools and dashboards to gain visibility on their emissions and cloud costs, which I expect to continue over the next couple of years,” Philipp Vannara, senior research journalist based in London, IDC Europe, told The Forecast.
“Most organizations currently only focus on energy consumption and cloud cost emissions, but over the next couple of years, I think organizations will demand clarity on water consumption and waste.”
Simultaneously, policy has reached a similar tipping point, the latest sign of which is the United States’ decision to rejoin the ambitious Paris Agreement, an international climate treaty whose goal is achieving a climate-neutral world by reducing greenhouse gas emissions across myriad industries and activities, including computing.
Additionally, 90 countries, which include the largest emitters — China, the United States and India — have pledged to reach a net-zero carbon emissions target, reported the World Resources Institute. Net-zero means that all emissions released by human activities are offset by using carbon removal to eliminate the carbon from the atmosphere.
Server virtualization reduces IT’s impact
However, even factoring in the positive impact of technology on sustainability, organizations must proactively work to reduce their computing impact on the environment and their sustainability goals.
One key factor is server virtualization, which allows for consolidated data centers that require less physical equipment to run large workloads, thereby leading to a smaller carbon footprint and less end-of-life IT waste. Additionally, services like pay-per-use and self-service cloud-based infrastructures allow customers to consume computing resources only when they need them.
“The idea of ‘green IT’ has been around now for years, but the direct impact hyperscale computing can have on carbon dioxide emissions is getting increased notice from customers, regulators and investors, and it’s starting to factor into buying decisions,” Cushing Anderson, program vice president at IDC, said in a press release.
“For some, going ‘carbon neutral’ will be achieved using carbon offsets, but designing data centers from the ground up to be carbon neutral will be the real measure of contribution. And for advanced cloud providers, matching workloads with renewable energy availability will further accelerate their sustainability goals.”
Wherever it’s deployed, technology can do things humans can’t. By doing good for the planet and being good for the planet cloud computing might be able to make one of the biggest dents yet in the global climate crisis.
