AI vendor due diligence: questions to ask before you buy
A practical checklist for evaluating AI software vendors on data protection, security, accuracy, compliance, costs and long-term risk before signing contracts.
Why AI uses significant energy, what that means for businesses with sustainability goals, and practical ways to reduce the environmental impact of AI use.
AI runs in data centres that use significant electricity and, in many cases, water for cooling. Training large models is energy-intensive, and everyday use adds up across millions of requests. Businesses can reduce their impact by using AI where it adds real value, choosing smaller efficient models for simple tasks, avoiding wasteful requests, caching results and asking providers about energy sources and efficiency.
AI's environmental footprint has become a growing topic for businesses with sustainability commitments. Understanding the basics helps you use AI responsibly without abandoning its benefits.
Training large models uses huge computing resources over weeks or months. This is mostly done by AI providers.
Every request to an AI model uses computing power. Individually small, these add up across billions of requests globally.
AI runs in data centres that need electricity for computing and cooling, and sometimes water. Their impact depends heavily on efficiency and energy sources.
Avoid AI for tasks simple software handles well, such as basic calculations or formatting.
Smaller models use less computing power and often handle simple tasks well. Route only complex tasks to larger models.
Well-designed AI features make fewer, better-targeted requests.
Ask about data centre efficiency, renewable energy use and reporting. Our guide to AI vendor due diligence covers questions to include.
AI can also reduce environmental impact elsewhere: optimising logistics, reducing waste, improving energy management and replacing travel with better remote collaboration. Consider net effects.
Precise measurement is difficult because providers vary in transparency. Track usage volumes, choose providers that publish environmental information and include AI in broader sustainability reviews.
When testing new AI uses, include efficiency in evaluation alongside accuracy and cost. Our guide to running an AI pilot project explains how to evaluate projects.
A customer support team used a large AI model for every message, including simple order status questions. After reviewing usage, it routed straightforward questions to a smaller model, cached answers to common questions and shortened system instructions. Accuracy stayed the same in testing, costs fell noticeably and computing use dropped with them. The team added these efficiency checks to its standard review process for new AI features.
If your business reports on emissions, include AI and cloud services within your digital footprint where data allows, and explain assumptions where it doesn't.
If you promote sustainability, be accurate about AI's impact and your efforts. Avoid claims you can't support.
For your own applications, choose hosting regions and providers with strong efficiency and renewable energy commitments where practical.
You don't need a data centre to make a difference. Choosing efficient tools, turning off unused features, avoiding unnecessary AI generation and favouring providers with clear environmental reporting all reduce impact. These choices often lower costs too.
Asking these questions early avoids building AI features that use energy and budget without delivering real value. It also encourages teams to design efficient features from the start rather than optimising later.
Our AI solutions team designs efficient AI features using appropriately sized models, our web app developers implement caching and batching and our web development team builds fast, efficient websites that reduce overall digital footprint.
AI uses energy and resources, especially at scale. Efficient use and good provider choices reduce impact.
Generally they use less computing power per request, making them more efficient for suitable tasks.
Precise measurement is hard. Track usage and ask providers for environmental information.
This guide is general information, not legal, tax or financial advice, and details such as prices, features and rules can change. Product and company names mentioned are trademarks of their respective owners and are used only to identify those products; UK Technology Hub is not affiliated with or endorsed by them.
Tell us about your project and get a free, fixed quote.
Get a free quoteA practical checklist for evaluating AI software vendors on data protection, security, accuracy, compliance, costs and long-term risk before signing contracts.
A step-by-step approach to choosing, scoping, running and evaluating an AI pilot, so you can decide with evidence whether to scale, change or stop.
How large language models like the ones behind ChatGPT, Claude and Gemini actually work, what they're good at, where they fall short and how businesses use them.