I wrote previously about the tension between AI’s energy consumption and the sustainability commitments of the companies deploying it. That tension has not resolved. If anything, it has intensified as AI workloads have scaled. But there is a parallel story that deserves equal attention: the growing body of AI applications that are actively reducing carbon emissions, optimising energy systems, and enabling the transition to a sustainable economy.
At NexaTech Ventures, we are increasingly seeing these two themes converge. The most interesting companies in our pipeline are not choosing between AI capability and sustainability. They are building businesses where AI capability is sustainability — where the technology itself is the mechanism for environmental impact.
AI as an Accelerant for Decarbonisation
The climate challenge is fundamentally an optimisation problem, and optimisation is what AI does best. The applications are numerous and growing.
In energy grid management, AI systems are being deployed to balance supply and demand across electricity networks with increasing penetration of variable renewable sources. A grid powered by fifty percent wind and solar requires real-time demand prediction, storage management, and load balancing that exceeds human operational capability. The AI systems managing these grids are not incremental improvements — they are enabling a grid architecture that would be technically impossible without them.
In building energy management, AI-driven HVAC and lighting systems are achieving energy reductions of twenty to thirty percent in commercial buildings by learning occupancy patterns, weather forecasts, and energy pricing signals. Given that buildings account for approximately forty percent of European energy consumption, the aggregate decarbonisation potential is enormous.
In industrial process optimisation, AI is reducing energy consumption in manufacturing by identifying inefficiencies in production schedules, material usage, and equipment operation that human operators cannot detect at scale. DeepMind’s well-publicised achievement of reducing Google’s data centre cooling energy by forty percent demonstrated the principle. The companies applying similar approaches to steel production, chemical processing, and food manufacturing are building substantial businesses.
The Investment Landscape for Green AI
The investment opportunity in green AI has several characteristics that make it particularly attractive in the current market environment.
First, regulatory tailwinds. The European Green Deal, the Corporate Sustainability Reporting Directive, and the EU Taxonomy Regulation are creating both the obligation and the economic incentive for enterprises to measure, report, and reduce their environmental impact. AI-powered sustainability tools — carbon accounting platforms, supply chain emissions tracking, environmental risk modelling — are the infrastructure that makes compliance possible. This is a market driven by regulation, which means the demand curve is predictable and growing.
Second, customer willingness to pay. Unlike many enterprise software categories where budget allocation is discretionary, sustainability spending is increasingly mandatory. Companies face regulatory penalties for non-compliance, investor pressure through ESG frameworks, and customer demand for verifiable sustainability credentials. The AI tools that demonstrate measurable emissions reductions or compliance capabilities command premium pricing.
Third, data advantage. European companies operating within the EU’s regulatory framework generate sustainability data at a scale and standardisation that provides a training advantage for AI models. The EU Taxonomy’s classification system and the CSRD’s reporting requirements create structured datasets that AI systems can learn from. This is a data moat that compounds over time.
The Hypocrisy Problem AI Must Address
Any honest discussion of green AI must also address the energy consumption of AI itself. The International Energy Agency has documented how training a large language model can consume as much energy as a small town uses in a year. Running inference at scale across millions of users generates a significant and growing carbon footprint. The technology industry cannot credibly promote AI as a sustainability solution while simultaneously building AI infrastructure that increases global energy consumption.
The companies that will win in the green AI space are those that are honest about this tension and actively working to resolve it. That means investing in energy-efficient model architectures, deploying inference on renewable-powered infrastructure, and measuring and reporting the net environmental impact of their AI systems — not just the benefits they deliver to customers, but the energy costs of delivering those benefits.
At NexaTech, we assess green AI companies on their net environmental impact. A carbon accounting platform that runs on fossil-fuel-powered cloud infrastructure and has never measured its own emissions does not pass our due diligence. We want companies that practise what they sell.
Where We Are Deploying Capital
NexaTech Ventures is actively investing in several green AI subsectors. Energy grid optimisation and management systems, particularly those focused on the integration of distributed renewable generation. Industrial AI applications that deliver measurable energy and emissions reductions in hard-to-abate sectors. And sustainability data infrastructure — the platforms that collect, process, and analyse the environmental data that enterprises need for regulatory compliance and operational improvement.
The green AI market is maturing from a niche concern into a mainstream investment category. The companies that establish themselves now, while the market is still taking shape, will define the infrastructure of the sustainable economy. The returns, both financial and environmental, will be substantial.
Scott Dylan is the Founder of NexaTech Ventures. He writes on AI, sustainability, and technology investment. Read more at scottdylan.com.
