Sustainable architecture
    Environmental Responsibility
    Environmental Statement

    Our Environmental
    Statement

    Transparency about our current actions, known limitations, and long-term sustainability ambition.

    At Gendo, we care deeply about the environmental impact of our technology.

    We have already made deliberate choices, such as running as much of our platform as possible on cloud infrastructure powered by 100% renewable energy. However, we recognise there is more to do.

    As a growing start-up, we are actively monitoring our energy consumption and will continue refining our approach as we scale.

    Section 1

    Our Infrastructure

    Cloud infrastructure services

    Portions of our platform run on infrastructure powered by 100% renewable energy, ensuring part of our operations are already supported by sustainable energy sources.

    Cloud GPU compute (multiple providers)

    For low-latency image generation, we rely on GPU cloud infrastructure provisioned from several providers. At present, these providers do not publish consistent environmental reporting on renewable sourcing, embodied carbon, or hardware lifecycle impact.

    We are committed to transparency about this gap and will work toward greater visibility and accountability as industry standards improve.

    Section 2

    Model Production and Data Centres

    We recognise that the environmental impact of AI extends beyond direct platform usage.

    502

    metric tons of CO₂ estimated for training GPT-3—equivalent to the lifetime emissions of 27 electric vehicles.

    1.5%

    of global electricity demand in 2024 is attributed to data centres, driven by cloud, storage, streaming, and AI workloads.

    Beyond energy

    AI contributes to embodied emissions through GPU manufacturing and water usage for cooling.

    Section 3

    Gendo's Sustainability Plan

    We are currently developing a proactive environmental responsibility strategy.

    Future Initiatives

    While this work remains in the research phase, potential future initiatives include:

    Per-generation measurement

    Tracking the energy footprint of each generation so users can better understand the impact of their workflows.

    Offsetting commitments

    Exploring mechanisms to offset measurable energy use, potentially beginning at the per-image level.

    Already Implemented

    Efficiency by design

    Deploying efficient models and controlling default resolution to reduce unnecessary energy use.

    Research indicates that doubling image resolution can increase energy consumption by 1.3–4.7×. We operate at lower resolutions by default and upscale only when necessary.

    We welcome suggestions from users, partners, and stakeholders on how we can improve this approach.

    Section 4

    Perspective for Architecture Studios

    AI-assisted workflows can represent a significant reduction in energy use compared with traditional rendering.

    Traditional CGI render

    ~1.3 kWh

    A 3-hour render on an RTX 3090 (~350W) for a single image.

    Gendo generation

    ~0.003 kWh

    A typical 30-second AI image generation on the same hardware.

    Efficiency gain

    ~350×

    Less energy per image, with compounding savings across iterative workflows.

    Traditional workflows also require additional hours of material editing, asset placement, landscaping, lighting adjustments, and re-rendering, further increasing energy consumption.

    Section 5

    Context Within the Built Environment

    It is important to contextualise these impacts against the overall carbon footprint of architecture.

    RIBA 2030 Climate Challenge Target

    New-build offices at approximately 130 kWh/m²·yr, meaning a 5,000m² office could consume roughly 39 million kWh over 60 years of operation.

    AI Design Footprint

    Producing 1,000 AI images during design would require only around 3 kWh.

    The design-phase AI footprint represents just 0.0000077% of the building's lifetime operational energy use.

    Even minor improvements, such as sealing a single window correctly, save orders of magnitude more energy than an entire design team's AI image production. If AI-supported design improves building efficiency by even 1%, the real-world savings are vastly greater than the energy cost of producing design imagery.

    Section 6

    Looking Ahead

    We believe AI will increasingly support not only image generation, but the full design process, including:

    Optimising form, material selection, and building systems for efficiency

    Simulating lifecycle carbon impacts in real time

    Enabling architects to design buildings with lower operational and embodied emissions

    Gendo's ambition is to move beyond faster visualisation and toward tools that empower architects to deliver measurably greener buildings. While this is not something we directly influence today, our long-term vision is to ensure that the small energy cost of AI-assisted design is repaid many times over through meaningful environmental gains in the built world.

    References

    1. 1.AWS Sustainability Report – The Cloud and Renewable Energy. Amazon. View source
    2. 2.Renée Cho. (2023). AI's Growing Carbon Footprint. Columbia Climate School. View source
    3. 3.IEA (2024). Energy demand from AI. International Energy Agency. View source
    4. 4.Guo, Y. et al. (2025). Energy Footprints of Image Generation Models. arXiv:2506.17016. View source
    5. 5.RIBA (2021). 2030 Climate Challenge Version 2. Royal Institution of British Architects. View source

    Calculate your studio's impact

    Use our Environmental Calculator to understand how Gendo compares to traditional rendering workflows.

    Try the calculator