The rapid deployment of AI models comes with significant environmental consequences, from increased energy demand to water consumption.
The International Energy Agency (IEA) estimates that the electricity consumption of AI data centres could double from 2022 levels to 1,000TWh (terawatt hours) in 2026. That’s approximately the same amount of electricity used by Japan each year.
Now there’s a crop of young companies aiming to help reduce the negative environmental impact of the tech, be it through developing more efficient models or improving the hardware needed to train and run them.
We asked investors which up-and-coming companies they have on their radar. There’s just one caveat: their picks can’t be in their portfolio.
Sebastian Hunte, investment director at AlbionVC
AlbionVC is a London-based VC backing companies from seed to Series A.
Nexos.ai – Lithuania
“Nexos helps businesses to reduce wasteful compute by intelligently routing across over 200 models based on context and task. Smarter orchestration means less unnecessary inference, and far better energy efficiency at scale.”
Flower – Sweden
“Much of the talk around greener AI focuses on efficiency gains, but I think that is the wrong way to think of it. Efficiency often leads to more usage, not less energy consumption. The real levers for sustainable AI are upstream: how we orchestrate workloads, where we run them and how green the energy powering them is.
That’s why Flower is such a compelling company: by combining grid-scale batteries with AI-powered trading and forecasting, they’re enabling a more stable, dispatchable supply of clean energy at scale. Their ‘green baseload’ platform could play a pivotal role in aligning AI’s rising energy demands with genuinely sustainable infrastructure.”
Terralayr – Switzerland
“Terralayr is building a flexible energy infrastructure layer by aggregating and virtualising battery storage assets. Their platform brings cloud-like flexibility to energy, helping accelerate the shift to renewables which is ultimately the only scalable way to decarbonise AI.”
Gabriele Papievyte, head of XTX Ventures
XTX Ventures is the venture capital arm of algorithmic trading firm XTX Markets. It invests in AI and machine learning startups from seed to Series B.
Enlightra — Switzerland
“Enlightra develops next-generation photonic hardware that enables superior performance in AI computing systems and data communication. The company uses optical transceiver systems (a device that transmits data by converting electrical signals into light signals) with chip-scale multi-colour laser engines for improved communication speeds.”
Nexalus — Ireland
“Irish startup Nexalus develops systems that integrate with excessive-heat-producing electronics to cool, capture and reuse this thermal energy, while also increasing efficiency and reducing costs. Such liquid cooling technology is targeting improvements in data centre energy efficiency and reduction of emissions.”
Literal Labs — UK
“Literal Labs leverages logic-based techniques to build AI models that are more energy efficient. The company is building towards ultra-low power use while optimising for inference, model size and explainability.”
Deepgate — UK
“Deepgate is on a mission to drastically reduce the energy demands of AI and accelerate inference. The company aims to achieve this by embedding intelligence directly into the units of computation.”
Read the orginal article: https://sifted.eu/articles/greener-ai-startups-to-watch/
