SpazioiTech
AI

Google backs 16 green AI projects in Asia-Pacific

Google is supporting 16 organisations in Asia-Pacific with frontier AI models for climate, agriculture and biodiversity. What it means for Italy and green AI.

By Daniele Messi 7 September 2026 5 min read Read in Italian →
Google backs 16 green AI projects in Asia-Pacific

Google has announced its support for 16 organisations in the Asia-Pacific region that are using frontier artificial intelligence models to tackle three areas: climate, agriculture and biodiversity. It is not news that will fill the front pages of the mainstream press, but for anyone following the relationship between green AI and environmental policy it is a concrete signal: a Big Tech company is putting resources where, until a few years ago, the public sector alone struggled to close the technology gap.

Google backs 16 green AI projects in Asia-Pacific
Image credit: Google Blog

This is an initiative that has no brand name yet, no public budget figure, and no delivery timetable. Google presents it on its blog as a programme of support for local organisations, and therein lie both its informational limits and its merit: it is not a product to be launched, but an ecosystem of collaborations. For the Italian reader, the question is simple: why should we care about what is happening thousands of kilometres from here? The answer is that the chains of climate, agricultural supply and biodiversity do not recognise customs borders.

What green AI is and why Asia-Pacific matters

Green AI, strictly speaking, means the use of advanced artificial intelligence systems to measure, predict and mitigate environmental impact. It is not a subset of the generative AI that produces text or images: here we are talking about models trained on satellite data, field sensors and historical weather records. Asia-Pacific, with its density of small island states, its tropical monocultures and its fragmented habitats, is an open-air laboratory for these applications. A frontier model, in this context, means a system capable of processing volumes of data and patterns that a traditional analysis would miss: the spread of a drought in South-East Asia, for example, cannot be read from a single weather station but from the convergence of dozens of variables.

The fact that Google is making computing capacity and access to frontier models available to 16 different organisations suggests a logic of scale: not a single pilot project, but a network. That is the point of interest for anyone in Italy working on agricultural supply chains or coastal monitoring. If an approach works in Manila or Jakarta, transposing it to the Mediterranean is a matter of adaptation, not of inventing from scratch.

Frontier AI: what changes compared with yesterday's tools

The term frontier AI entered the technical vocabulary in 2024-2025 and refers to the largest and most capable models, those that push the limits of training at scales of parameters and data out of reach of a single academic laboratory. For the 16 organisations Google mentions, access to this level of computing power means being able to process, for example, high-resolution satellite images in near real time to monitor deforestation, or to cross-reference agronomic data with fine-resolution weather forecasts to reduce the use of pesticides.

That is no small detail. Until a couple of years ago, a small NGO in Vietnam or Indonesia would have had to ask for access to computing clusters costing millions of dollars. Now, if Google structures its support in this direction, the barrier to entry is lowered. For the Italian reader involved in agricultural research, the practical question is: how much of this infrastructure will be replicable or accessible in Europe too, where the European AI framework (the so-called AI Act) imposes transparency and audit requirements that do not exist in the same way in other jurisdictions?

Why it should matter to the Italian reader

The biodiversity of the Mediterranean and that of Asia's rainforests are not watertight compartments. Migratory species, trade routes for agricultural commodities, heatwaves spreading on atmospheric currents: everything is connected. A model trained to predict the resilience of a rice crop in Thailand can offer reasoning architectures that transfer to vines in Piedmont or olive trees in Puglia, even if the parameters change.

That said, there is one aspect that should not be underestimated. Google is a company with its own interests: every support programme has a logic of branding, data collection and competitive positioning. The 16 organisations benefit from resources, but the relationship is not symmetrical. Anyone reading this announcement should ask who holds the data generated, who can reuse it, and for how long. These are legitimate questions, not conspiratorial sighs. Green AI works when transparency about the flow of data is at least on a par with transparency about the model's output.

A step, not a solution

Fifteen or sixteen projects in a region as vast as Asia-Pacific remain a drop in the ocean. I do not say this to belittle them: I say it because context is needed. The United Nations Environment Programme and regional agencies have been pointing out for years that the gap between measurement capacity and capacity to intervene remains wide, especially in middle-income countries. With this move, Google closes a piece of that gap. But only a piece.

For the Italian reader, the takeaway is twofold. On one hand, access to frontier models for sustainability is being democratised: what three years ago was the preserve of a few research centres is now available to wider networks. On the other, sustainability is not solved with a more powerful chip: it is solved with agricultural policies, monitoring infrastructure and public funding that keeps pace. Green AI is a tool, not a strategy. And in Italy, where the PNRR (the National Recovery and Resilience Plan) has already earmarked resources for the digital and environmental transition, the question remains open: how much of this international push will translate into concrete projects, with calls for proposals, local partners and measurable results by 2026 and 2027?

For now, 16 organisations and three sectors. A starting point that, if followed through consistently, will have something to say. If it remains one more announcement on a corporate blog, it will say little. What makes the difference will be time, not the tone of the communications.

Frequently asked questions

What is green AI in practice?

Green AI refers to the use of artificial intelligence systems, in particular frontier models, to analyse environmental data, predict climate events and support decisions in agriculture and biodiversity protection. It is not a single product, but a family of applications ranging from satellite monitoring to forecasting crop yields.

Which sectors do Google's 16 projects in Asia-Pacific cover?

According to Google's announcement, the 16 projects focus on three areas: climate, agriculture and biodiversity. The company has not, for the time being, released the names of the individual organisations or the technical details of each initiative.

Can green AI be applied in Italy too?

Yes, models trained on Asian data can be adapted to Mediterranean contexts, even if the parameters change. In Italy, the PNRR provides resources for the digital and environmental transition: the question is how far these international programmes will translate into calls for proposals and concrete collaborations on the ground.

Via: Google Blog

Translated from the Italian original. Read in Italian

AuthorDaniele Messi