The swift expansion of artificial intelligence has triggered an environmental crisis, with data centers accounting for 1-2% of global electricity consumption and AI’s computational requirements doubling every three to four months.
Europe has led green computing innovation in this context, as outlined in the 2025 white paper “Green Computing in the AI Era: Innovations Tackling Computing’s Carbon Impact.” This report, created by World Fund, Ignite (Intel’s deep-tech accelerator), and Dealroom, indicates that 54 European startups in this space have raised $970 million since 2019, with $305 million acquired in 2024 alone.
Over half of these startups emerged in the last five years, with 12 established in the past year, underscoring Europe’s rapid advancements. Germany tops the list with 22 startups, followed closely by France with 15, while growing hubs like Estonia and Portugal are paving the way for breakthroughs in quantum and photonic technologies.
Addressing AI’s environmental effects and economic prospects necessitates unprecedented collaboration among policymakers, investors, and technologists to scale effective solutions. Tech Funding News looked at the report closely — here’s what we discovered.
AI’s power paradox: innovation vs. environmental impact
The rapid rise of AI creates a dilemma: it offers potential solutions for climate issues, but its substantial carbon emissions jeopardise worldwide decarbonisation efforts. Data centers account for 1.5% to 2% of global electricity usage, projected to triple by 2030 as AI becomes more widespread. For example, training large language models, such as GPT-3, releases roughly 552 tonnes of CO₂ equivalent.
By 2030, U.S. data centers alone might use 9% of the country’s electricity, indicating similar trends worldwide. Cloudflare’s data reveals that generative AI traffic skyrocketed by 251% from 2024 to 2025, increasing computational power demand. Cryptocurrencies significantly impact energy resources; for instance, Bitcoin mining consumes electricity that is comparable to Poland’s annual usage.
Europe is taking action through various initiatives. The EU’s Digital Twin Earth Challenge and the Machine Learning for Earth and Climate Sciences program highlight efforts to align AI development with environmental objectives. The demand for AI computing capacity doubles every three to four months, which far exceeds the rate of efficiency improvements.
Even though funding for European AI startups increased by 32% to $13 billion in 2024, this growth incurs considerable environmental implications. Recent notable funding includes Mistral AI’s $640 million round at a valuation of $6 billion and Poolside AI’s $500 million round at a valuation of $3 billion.
The white paper underscores that merely transitioning data centers to renewable energy will not resolve the issue. As Daria Saharova of World Fund noted, “We must fundamentally redesign how we compute,” highlighting the need for innovations that decouple performance enhancements from energy consumption. Escalating energy prices, carbon taxes, and emission targets may make existing computing models untenable without significant transformation.
Key investment areas: semiconductors, computing innovations, and software
First — innovations in semiconductor materials. In the hardware sector, European startups are revolutionising semiconductor efficiency. GaN Systems in Germany, now owned by Infineon for $830 million, uses gallium nitride (GaN) to create chips that achieve 40% more energy efficiency than traditional silicon chips, a significant advancement currently being scaled for GPUs in data centers.
Space Forge uses microgravity manufacturing techniques in the UK to fabricate silicon carbide (SiC) wafers with 70% lower emissions than conventional methods, thereby addressing production and operational sustainability.
Second — advancements in computing. Europe’s quantum computing landscape highlights the simultaneous focus on performance and sustainability. Finland’s IQM, which secured €128 million in 2022, is developing superconducting quantum processors that use 99% less energy than classical supercomputers when tackling logistics and materials sciences optimisation challenges.
Similarly, remarkable progress is occurring in photonics: Black Semiconductor’s light-based chips handle data processing 1,000 times faster than their electronic counterparts while significantly reducing power consumption, a breakthrough made possible by integrating III-V compound semiconductors with silicon photonics.
Third — software optimisation. Although hardware innovations often dominate the headlines, software optimisation offers immediate opportunities for emissions reduction. The Green Algorithms tool, implemented by 23% of European AI startups, allows developers to measure the CO₂ impact of code changes, covering aspects like memory management and parallelisation techniques.
How Europe pioneers sustainable AI computing solutions
The emergence of startups and collaborative public-private partnerships marks Europe’s green computing sector. Of the 65 global startups in this field, 54 are based in Europe, led by Germany (22 startups) and France (15), and new hubs like Estonia and Portugal are making significant strides.
Established in 2023, Germany’s €1 billion DeepTech & Climate Fonds (DTCF) is crucial in fostering growth-stage funding, supporting companies such as the quantum computing firm IQM and photonic chip developer Black Semiconductor. This initiative complements France’s €500 million deeptech project and the European Commission’s goal to mobilise €30 billion for startups by 2030.
“Europe is not just a player; it’s shaping the future of sustainable computing,” stated Lorenzo Chiavarini, Head of Research at Dealroom. The white paper also points out that partnerships between startups and accelerators, like Intel’s Ignite program, have been vital in scaling these innovations.
Collaboration with accelerators has been crucial. Intel’s Ignite program offers technical guidance to startups like Deep Render (which focuses on AI-driven data compression) and Literal Labs (which concentrates on energy-efficient machine learning). These companies utilise these networks to refine AI workloads via model compression and dynamic resource allocation, resulting in energy savings of 30–40% in initial implementations.
Simultaneously, the Green Software Foundation’s carbon-aware computing approach, backed by Microsoft and Intel, transforms how algorithms interact with energy grids. However, some critics suggest that this must progress to “grid-aware computing,” which focuses on reducing net demand rather than optimising for time.
Europe’s regulatory framework acts as both an enabler and a limitation. The EU AI Act requires environmental impact assessments for high-risk AI systems, while the Carbon Border Adjustment Mechanism enforces tariffs on imported semiconductors with high embedded emissions. According to the white paper, suggested solutions include implementing carbon-aware procurement policies, sandbox regulations, and tax incentives.
The future? Europe can lead in sustainable computing
With the global AI chip market projected to reach $1.8 trillion by 2030, Europe’s early lead in sustainable computing could capture 12–15% of this value chain, contingent on maintaining annual R&D growth rates above 8%. Success hinges on translating pilot innovations like Space Forge’s orbital manufacturing and IQM’s quantum annealing into industrial-grade solutions, a transition requiring another $4–6 billion in venture funding by 2027.
Europe’s green computing trajectory demonstrates that environmental sustainability and technological leadership are not mutually exclusive. The region is charting a path for AI development that aligns with planetary boundaries through synergistic advancements in materials science, photonic engineering, and algorithmic efficiency.
However, as the white paper cautions, current progress remains fragile. Scaling from laboratory breakthroughs to gigaton-scale emissions reductions demands policy frameworks that reward carbon-negative compute, investment vehicles patient enough for deep-tech timelines, and a cultural shift prioritising long-term ecological viability over short-term computational benchmarks.
