Impact Computing: Planet-friendly AI and data platform | HCLTech

Impact Computing builds a planet-friendly AI and data platform

Through impact computing, organizations can build a sustainable data and AI platform in the cloud to realize never-before-seen benefirs
9 min. read read
Jordan Smith
Jordan Smith
US Reporter, HCLTech
9 min. read read
Impact Computing

Building a data and AI platform that is sustainable and good for the planet begins with impact computing. Impact computing helps to reduce energy consumptions as well as carbon emissions through converging data intelligence and intelligent apps, enabling cloud workloads with hyperscalers and utilizing hyperscalers to strike a balance between digitalization and decarbonization.

When considering impact computing, organizations need to determine where they can reduce emissions and energy consumptions in every component of their organization including infrastructures. For example, business operations for most organizations contribute a greater volume of carbon emissions than from its core IT services. A single AI model can emit as much carbon as five cars in one’s lifetime.

“Impact computing defines how we impactfully compute using less energy; that's the key,” Soumen Chatterjee, Associate VP, Data and AI Modernization Principle for Europe at HCLTech. “How have your computation and data platform impact the environment and the planet. Can you consume less energy and achieve net zero carbon as well as less energy?”

Additionally, data storage in recent years has made up about 4% of all global carbon emissions. At the same time, compute consumption of high-compute AI systems is doubling every 3.4 months.

The Introduction of Impact Computing

When considering impact computing, one must first consider environmental, social and governance (ESG) aspects and the two key focuses that brings: The Pledge Commitments and Science-Based Targets.

The Pledge Commitments refers to different organizations committing to net-zero carbon by 2040. These organizations implement decarbonization strategies aligned with the 2015 Paris Agreement through business changes and innovations, such as efficiency improvements, renewable energy and materials reduction.

Science-based targets refers to a zero-carbon economy. Through the Paris Agreement, governments worldwide have committed to curbing global temperature rise to well-below two degrees Celsius above pre-industrial levels and pursuing efforts to limit warming to 1.5 degrees Celsius.

“In the current environment, there are two concerns,” explained Chatterjee. “One is the temperature increase and the other one is the carbon. When organizations are focusing on ESG, they are talking about all the sustainability from the corporate side.”

Chatterjee explained that there are three different scopes from the corporate side, with scope one and two referring to the direct contribution by any organization to carbon emissions and scope three referring to the indirect carbon emissions, such as employees commuting to the office or home network features.

Accelerating impact computing

Choosing the correct Hyerscaler is critical to accelerating impact computing. A recent report on the carbon benefits of cloud computing by Microsoft found that its cloud offering is between 22 and 93% more energy efficient than traditional enterprise datacenters. AWS says that running business applications on their infrastructure could reduce energy use in excess of 80% for many businesses.

Meanwhile, Google Cloud states that its datacenters are twice as efficient as a typical enterprise datacenter.

In addition to selecting the right hyperscaler, it’s important when accelerating impact computing to determine what goals your organization is focused on. The UN outlined sustainable development goals that organizations can follow including eliminating poverty and hunger, improving health and well-being, and reducing inequalities.

Impact computing with AI and ML

Strategies for establishing green data and AI begins with choosing carbon intensity and sustainable regions. Regulations in different countries and regions vary, so organizations must choose regions near renewable energy projects and regions where the grid has low published carbon intensity to host data and workloads. Avoiding datasets and processing duplication, along with evaluating if you can avoid data processing through existing publicly available datasets.

Strategizing sustainable AI systems is a four-pronged approach: algorithm selection, smaller models, deployment strategy and deployment region.

“Most of the organizations these days are focusing on green AI because a tiny bit of additional accuracy can attribute to a significant amount of energy as well as carbon,” said Chatterjee.

For sustainable machine learning, organizations must consider AI services and pre-trained models, while also carefully considering if a workload needs to be developed as a custom model.


Act, Pact, Impact - HCLTech publishes its 2022 Sustainability Report

Download the report


Other strategies for computation with green AI and ML include optimizing the compute layer through an adaptive resource strategy and using containerization where possible as containers reduce power consumption and provide sustainable scheduling strategies. Further, optimizing traversing through the network layer can reduce the network travelled per request and reduce consumption.

Looking forward to the next generation of green AI and data platforms, Chatterjee says we need to think about the planet and how we save more energy. That starts with every product needing to have a green angle, sustainable nutrition facts stating that “every architecture we build, we have to assess against leading practices, whether they are green, how green they are and whether they're energy and carbon efficient or not.”

How HCLTech and other organizations embed responsible innovation

The 2022 UN Climate Change Conference, or COP27, and other organizations have highlighted how dire the consequences of climate change in the future are. With this in mind, it’s important for companies like HCLTech to establish responsible innovation as part of it’s business critical strategies.

Chatterjee suggests that HCLTech must continue to practice corporate responsibility, but at the same time educate or evangelize the customer.

“There are a lot of ways to help advise our customers, so we should be detailing all of our strategy to our customers to make sure that they are always using the right set up technology,” he said.

HCLTech’s CloudSMART offering architecting sustainability on cloud offerings to achieve ESG goals and create a sustainable future. As Chatterjee alludes to, technologies and offering, like CloudSMART, are permeating the marketplace to ensure that energy consumption and yearly budgets are affordable.

“There are a lot of technologies coming that will be taking over in the market,” said Chatterjee. “These technologies will make sure organizations will be adopting technologies that can make sure that their energy consumptions for their data centers and their yearly budget are allocated impactfully.”

Before adopting technologies, organizations can also look inward to begin developing green AI and data platforms by looking at all factors impacting their carbon footprints through well-architected sustainability assessments, being careful with what material is used for offices, other business structures, technologies adopted in the organization as well as for our customers.

Share On