When we talk about AI, we’re usually talking about scale: larger models, more data, and ever‑expanding data centres. Companies like OpenAI and Anthropic have come to define progress in AI as something centralised, energy‑intensive and overwhelmingly shaped by the priorities of the Global North.

But a different AI story is being built further South, and further East.

Two emerging startups – Amini and Thaura AI – offer a powerful counter‑narrative. Not just alternative tools, but alternative values: climate responsibility over extraction, representation over erasure, and sovereignty over unchecked scale.

AI from the margins – by design, not by accident

Amini, a Nairobi‑based startup, is explicit about its focus on the Global South, challenging a tech ecosystem that has long relied on labour, data and natural resources from low‑ and middle‑income countries while excluding them from ownership and value creation¹. Its mission is to build the foundational data and compute infrastructure needed for communities to develop AI on their own terms, rather than importing models and systems designed elsewhere².

This matters because today’s dominant AI models are disproportionately trained on English‑language, Western‑centric datasets – reinforcing narrow worldviews while rendering billions of people invisible, misrepresented or statistically irrelevant³.  Take into account something like ChatGPT’s pro-Israel bias, you can see how just one example should make you ask, what other biases are being prioritised?

Thaura AI takes a similar stance as Amini, but from a West Asian perspective. Founded by Syrian engineers with lived experience of displacement and surveillance, Thaura openly positions itself not as a “neutral” technology, but as an ethical and political alternative to Big Tech AI⁴ ⁵. It rejects military and surveillance contracts, refuses to train on user data, and frames AI development as inseparable from questions of justice, power and historical context.

In both cases, the point is not to “catch up” with Silicon Valley – but to reject its assumptions altogether.

Smaller, local, and mobile – infrastructure as resistance

One of the sharpest distinctions between these startups and large AI labs is infrastructure design.

Large‑scale AI development is driving a rapid expansion of hyperscale data centres, with enormous energy and water demands that often fall on regions already under climate stress⁶. These facilities are frequently built on or near Indigenous land, with minimal community benefit and long‑term environmental cost.

Amini has taken a different approach, developing local‑first, modular compute infrastructure, including “Amini Pods” – containerised micro‑data centres that can be deployed closer to where data is generated⁷. Rather than forcing communities to rely on distant cloud regions, this enables data to be processed locally, supports sovereignty over infrastructure, and allows services to reach smaller and more remote communities across Africa⁷ ⁸.

This is not just a technical choice; it is a political one. Africa currently hosts almost 19% of the global population but only around 1% of global data‑centre capacity, with the vast majority of its data processed elsewhere⁹. Amini’s model directly challenges this imbalance.

Thaura’s approach is similarly frugal by design. Its architecture is intentionally smaller and more energy‑efficient than the trillion‑parameter models pursued by Big Tech, with renewable energy commitments and explicit rejection of “bigger is better” thinking¹⁰.

Making climate impact visible – not hidden

A particularly striking feature of Thaura AI is how it makes environmental impact visible.

Each user prompt displays an estimate of carbon saved compared to larger commercial AI models, translating abstract compute emissions into something tangible at the point of use¹⁰. In an industry where environmental costs are usually hidden behind “cloud” metaphors and offset promises, this level of transparency is rare – and powerful.

It reframes AI use as a choice with material consequences, not an inevitable or neutral transaction.

Why this matters for Indigenous, local and diaspora communities

AI infrastructure is not abstract – it is physical, land‑based and resource‑intensive. Research increasingly shows how the expansion of data centres and digital infrastructure clashes with Indigenous stewardship principles and exacerbates environmental injustice, particularly in the Global South⁶ ¹¹.

When AI systems are built without Indigenous governance, data sovereignty or consent, they risk reproducing colonial patterns under a digital guise.

Models emerging from West Asia, Africa and the broader Global South point toward a different future – one where AI supports local decision‑making, cultural context, and community control rather than extraction and surveillance¹².

For diaspora communities in more affluent countries, these alternatives also matter deeply. They shape which histories are recognised, which languages are supported, and whether political realities – particularly those relating to occupation, displacement or climate vulnerability – are acknowledged rather than erased or censored.

Part of a wider movement

Amini and Thaura are not anomalies. They sit within a growing ecosystem of initiatives reimagining AI:

• Lelapa AI building language models for African languages historically excluded from mainstream AI³
• Masakhane, embedding community‑centred values like Ubuntu into AI development across Africa³
• Karya and other South Asian initiatives ensuring data labour directly benefits local workers rather than being quietly extracted¹³
• Indigenous‑led governance frameworks arguing for AI grounded in stewardship, reciprocity and sovereignty¹¹ ¹²

Together, these efforts challenge the assumption that innovation must be centralised, extractive and environmentally costly.

A closing Reflection

If AI is reshaping labour, culture and planetary resources, then who builds it – and how – is a racial justice issue, a climate justice issue, and a power issue.

Amini and Thaura remind us that AI does not have to be louder, larger or more extractive to be transformative.

Sometimes, technology is at its most radical when it rewrites the rules in ways that feel more natural, more humane, and closer to our shared community values. In serving those who are too often overlooked, it ultimately serves humanity itself. Sources:

¹ Amini – The Global South Needs to Own Its AI Revolution

² Africa Briefing – Nairobi’s Amini AI eyes data revolution

³ World Economic Forum – How the Global South is reimagining the future of AI

⁴ L’Orient Today – Thaura: a Syrian AI designed to ‘decolonize’ tech

⁵ Barakah Insider – Meet Thaura: An Ethical, Privacy‑First Alternative to Big Tech AI

⁶ Brookings – AI infrastructure’s environmental costs clash with Pacific Island nations’ needs

⁷ Amini – Official site (Amini Pods, modular micro‑data centres)

⁸ Fast Company – This Nairobi startup brings agriculture data to African farmers

⁹ Xalam Analytics (via AFP reporting on African data capacity)

¹⁰ Thaura AI – Official site (energy efficiency and carbon‑savings disclosures)

¹¹ Forbes – Tech With Respect: AI and Indigenous Community Power

¹² CARE International – AI and the Global South

¹³ Asia Climate Hub – Asian startups using AI to reduce emissions at scale