With help of Sameera Mudgal and other collaborators, we have shared a proposal to extend the ideas of Sun Shines Bright to the community.
Embracing grounding as a sharing and learning experience for the community over making and hacking.
Sun Shines Bright reimagines our relationship with AI and the cloud by demystifying and DIY-ing local and self-hosted AI servers, and cultivating community-led AI systems. The project proposes that grounding technology is as much about building agency and accessibility of usage for diverse creator communities as it is about the materiality of systems to build it. We propose that to “ground the cloud”, we need to embody the principles of care and camaraderie in our technological interventions by:
- Creating community, agency and accessibility to use complex AI systems
- Making and hacking simple physical AI infrastructures with solar energy
- Nurturing home-grown, p2p AI servers
- Collecting and training our own Small Language Models that reflect our values and voices
Rather than naively offering alternative prototypes and end products for one-time use and showcase, we propose simple, radical ways of community building and facilitation that respond with deep, situated and diverse knowledge. We create access, agency, control and understanding of complex AI systems which, when combined with inexpensive and easy-to-use hardware, create potent critical and long-term alternatives to invasive, biased and blackbox systems. We see this effort aligned with the idea of the Solar Mamas of the Barefoot College, India, where rural women were trained in solar electrification of their villages, empowering themselves and their extended community. We envision our project having a similar ripple effect - a joyous space for knowledge building and care-forward tinkering with technology.

Representative image
Premise and Urgent Context
Contemporary life is inextricably linked to massively scaled AI systems trained on biased datasets and hosted in energy-intensive data centers. As an alternative, can community-learned, solar-powered small language models create new forms of technological intimacy that resist extraction while fostering care?
Inspired by Solar Protocol, which they call ‘A Naturally Intelligent Network’, Sun Shines Bright’s proof of concept simultaneously explores low-energy and sustainable model hosting and opportunities and efficacies of Small Language Models. The early experiments (Jan-Mar 2025) and subsequent workshop at NYU Shanghai (April 2025) proved that building a simple solar-powered server for hosting AI models is not only possible but can be easily shared and explored with students and creatives from non-technology backgrounds. This was followed up by an exhibit and zine about Sun Shines Bright at AIxDESIGN Festival: On Slow AI, May 2025 and Chaos Feminist Convention July 2025. Several feminist non-technologist creators have expressed their interest in building their own solar servers. This reflects an urgent need for:
- agency over the mainstream high-power black-boxed AI systems
- dissociation and simple solutions from current cloud-based models,
These models are climatically problematic and also do not reflect critical feminist non-white data. Now more than ever, we need camaraderie, friendship and intimacy as pillars of building agency and ease around fast-paced complex AI systems.

My son’s version of the Representative image