Global energy consumption from computers is skyrocketing, electricity usage is spiking, and the carbon footprint of cloud computing is now higher than that of the airline industry’s. The adoption of artificial intelligence and machine learning further contributes to this massive amount of energy usage. Large language models (LLM) come with their own set of energy explosions since LLM training data is at the core of an effective AI model, especially in real time. More data means more energy used in data operations — real-time data ingestion, analysis and processing. In fact, training several AI models can unleash 626,000 pounds of carbon dioxide, which is nearly five times the emissions of a car over its entire life cycle. It’s clear that energy-guzzling products, processes and services are multiplying exponentially, and both the public and private sectors hold a responsibility to ensure the sustainability of software systems to curb emissions that threaten the environment.
Adding rules to reduce emissions
A significant step forward occurred at the opportune time, when AI’s adoption is climbing. In April of this year, the administration updated its sustainability mandate, requiring federal agencies to procure sustainable products and services “to the maximum extent practicable,” which went into effect at the end of May. This stems from the 2021 executive order focused on federal sustainability and is part of the 2021 Federal Sustainability Plan, which includes the goal to achieve net-zero emissions from federal procurement by 2050. Further down the line, federal suppliers also may be bound to publicly disclose their emissions and set their own goals for reducing them. For a host of reasons, these new mandates will drive an increase in the quality of solutions for agencies.
Sustainability push transforms the federal procurement process
Overall, the updated federal procurement mandate changes both the buying and selling sides. This may look very basic, like buyers and sellers simply shifting course, or look complex, such as shaking up the federal supply chain completely. IT decision-makers will need to factor this new layer of sustainability into the procurement process, taking more time and effort, potentially phasing out previous vendors who don’t fit the criteria and taking on new ones who already demonstrate sustainable practices or who have pivoted quickly.
The federal push for sustainability may also spark increased transparency in the procurement process. Though the rule to require vendors to disclose greenhouse emissions isn’t yet final, those who electively do so may offer agencies more value, as they’re likely to have strong, tested sustainability practices, which may demonstrate a commitment to energy conservation.
More broadly, with tech and AI vendors needing to refocus on sustainability, we may witness a technological renaissance — a significant increase in product quality and innovation offered to agencies. The LLM training data process demands a constant stream of feeding and tuning. When refining LLM training data, the hardware required to sustain intricate algorithms throughout their life spans only adds to the carbon load. It is imperative for vendors in this space to acknowledge their impact and act responsibly, mindful of the present and future. It’s an ethical obligation.
If vendors haven’t already been building and adapting their technology to reduce emissions, we can expect to see them adopt a thoughtful, strategic, scientifically-driven approach in light of the new measures. Increasing operational efficiency is no small task because sustainable practices can’t be bolted onto a tool. It requires organizational cooperation, efficient coding practices, optimized technology and groundbreaking innovations. Incorporating emissions-reducing capabilities into technology or building an entirely new technology with sustainability at the heart can encourage vendors to rethink their offerings and, put simply, do it right.
What federal IT decision-makers should look for in AI software
There are a few elements IT decision-makers should consider when speaking with potential vendors to choose sustainable software options. Asking questions and weighing competitive options is crucial to selecting the platform with the most value. Here are a few examples:
Is the software efficient? Look for vendors who offer high performance, increased profit margin and reduced maintenance costs. Efficient software delivers results with fewer CPU cycles, demands less hardware and energy, and contributes to a significantly lower total cost of ownership (TCO).
Is the software purpose-built for sustainability? Some vendors have made it their mission to design their products from the ground up with sustainability at the heart of the build, while others commit to sustainability later in their journeys. It is worth asking vendors about their sustainability histories as you evaluate options.
Will a vendor disclose its emissions? Though vendors are not required to do so, gleaning the answer from each vendor and evaluating those answers side-by-side can offer one of the most basic quantitative comparisons. If a vendor doesn’t disclose its emissions, ask how the solution reduces infrastructure, as this translates to a lower carbon footprint.
Low CO doesn’t mean high TCO
Solutions that offer reduced emissions without sacrificing performance and while delivering a low total cost of ownership are available. While that combination may seem to be a unicorn, increased sustainability does not and should not constitute higher costs. If it does, it’s likely not efficient infrastructure. However, the updated sustainability procurement rules, and those to come, will likely introduce many more viable solutions.
Our reliance on technology has led to a demand for real-time data, which has sparked a proliferation of AI tools. If AI solutions remain unchecked, emissions will continually be leaked needlessly. The government’s drive toward reducing emissions is a step in the right direction for responsible technology deployment and a sustainable future.
Cuong Nguyen is vice president for public sector at Aerospike.
Federal sustainability focus to transform AI procurement
More data means more energy used in data operations — real-time data ingestion, analysis and processing.
Global energy consumption from computers is skyrocketing, electricity usage is spiking, and the carbon footprint of cloud computing is now higher than that of the airline industry’s. The adoption of artificial intelligence and machine learning further contributes to this massive amount of energy usage. Large language models (LLM) come with their own set of energy explosions since LLM training data is at the core of an effective AI model, especially in real time. More data means more energy used in data operations — real-time data ingestion, analysis and processing. In fact, training several AI models can unleash 626,000 pounds of carbon dioxide, which is nearly five times the emissions of a car over its entire life cycle. It’s clear that energy-guzzling products, processes and services are multiplying exponentially, and both the public and private sectors hold a responsibility to ensure the sustainability of software systems to curb emissions that threaten the environment.
Adding rules to reduce emissions
A significant step forward occurred at the opportune time, when AI’s adoption is climbing. In April of this year, the administration updated its sustainability mandate, requiring federal agencies to procure sustainable products and services “to the maximum extent practicable,” which went into effect at the end of May. This stems from the 2021 executive order focused on federal sustainability and is part of the 2021 Federal Sustainability Plan, which includes the goal to achieve net-zero emissions from federal procurement by 2050. Further down the line, federal suppliers also may be bound to publicly disclose their emissions and set their own goals for reducing them. For a host of reasons, these new mandates will drive an increase in the quality of solutions for agencies.
Sustainability push transforms the federal procurement process
Overall, the updated federal procurement mandate changes both the buying and selling sides. This may look very basic, like buyers and sellers simply shifting course, or look complex, such as shaking up the federal supply chain completely. IT decision-makers will need to factor this new layer of sustainability into the procurement process, taking more time and effort, potentially phasing out previous vendors who don’t fit the criteria and taking on new ones who already demonstrate sustainable practices or who have pivoted quickly.
The federal push for sustainability may also spark increased transparency in the procurement process. Though the rule to require vendors to disclose greenhouse emissions isn’t yet final, those who electively do so may offer agencies more value, as they’re likely to have strong, tested sustainability practices, which may demonstrate a commitment to energy conservation.
Get tips on how your agency should tackle the data pillar of zero trust in our latest Executive Briefing, sponsored by Varonis.
More broadly, with tech and AI vendors needing to refocus on sustainability, we may witness a technological renaissance — a significant increase in product quality and innovation offered to agencies. The LLM training data process demands a constant stream of feeding and tuning. When refining LLM training data, the hardware required to sustain intricate algorithms throughout their life spans only adds to the carbon load. It is imperative for vendors in this space to acknowledge their impact and act responsibly, mindful of the present and future. It’s an ethical obligation.
If vendors haven’t already been building and adapting their technology to reduce emissions, we can expect to see them adopt a thoughtful, strategic, scientifically-driven approach in light of the new measures. Increasing operational efficiency is no small task because sustainable practices can’t be bolted onto a tool. It requires organizational cooperation, efficient coding practices, optimized technology and groundbreaking innovations. Incorporating emissions-reducing capabilities into technology or building an entirely new technology with sustainability at the heart can encourage vendors to rethink their offerings and, put simply, do it right.
What federal IT decision-makers should look for in AI software
There are a few elements IT decision-makers should consider when speaking with potential vendors to choose sustainable software options. Asking questions and weighing competitive options is crucial to selecting the platform with the most value. Here are a few examples:
Low CO doesn’t mean high TCO
Solutions that offer reduced emissions without sacrificing performance and while delivering a low total cost of ownership are available. While that combination may seem to be a unicorn, increased sustainability does not and should not constitute higher costs. If it does, it’s likely not efficient infrastructure. However, the updated sustainability procurement rules, and those to come, will likely introduce many more viable solutions.
Our reliance on technology has led to a demand for real-time data, which has sparked a proliferation of AI tools. If AI solutions remain unchecked, emissions will continually be leaked needlessly. The government’s drive toward reducing emissions is a step in the right direction for responsible technology deployment and a sustainable future.
Cuong Nguyen is vice president for public sector at Aerospike.
Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.
Related Stories
AI effort to improve contractor performance ratings stalls
Is an earthquake ahead for a mainstay of federal procurement?
Driving agency AI literacy utilizing guardrails and frameworks