During the AI impact Summit event, held in February 2026, Spain and 88 other countries and international organizations adopted the New Delhi declaration on AI.
This declaration is much more than a mere diplomatic communiqué. It is an explicit attempt to rebalance the global governance of artificial intelligence at a time of technological fragmentation, geopolitical tensions, and increasing concentration of power around foundational models, computing infrastructure, and data.
Under the motto “AI for All” and anchored in the maxim “Sarvajan Hitaya, Sarvajan Sukhaya” (well-being and happiness for all), the Indian initiative proposes a framework for voluntary cooperation structured on seven pillars:
- democratization of resources
- economic growth and social good
- safe and reliable AI
- AI for science
- access for social empowerment
- human capital development
- resilient and efficient systems
The relevant question is not whether the statement is well-intentioned —it is— but whether it is strategic, executable, and capable of altering the current asymmetries.
As is known, the current board of AI governance is structured into three clearly differentiated spaces: the regulatory approach typical of the European Union (focused on the protection of people's fundamental rights and the defense of innovation), the emphasis on innovation and national security of the United States and China's state-technological model.
In reality, the achievement of the objective known as the “democratization of AI resources”—including access to computing capacity, to models, data, and capabilities—continues to be hindered by the traditional concentration of infrastructure in a handful of technologically advanced private actors, and States that use that technology to achieve their geostrategic policy objectives.
Voluntariness versus normativity
However, all the instruments proposed in this meeting have been voluntary and non-binding. This has diplomatic advantages —it facilitates broad adherence and avoids regulatory blockages— but also evident limits.
While the European Union advances with mandatory legal frameworks on risks, transparency, and algorithmic accountability, and the United States consolidates sectoral standards and strategic export controls, the Indian proposal opts for flexible principles, technical cooperation, and soft law.
Is this approach sufficient in an environment where AI impacts so intensely on -to cite just a few- fundamental rights, labor markets, electoral processes and international security?
This reality can lead us to conclude that while the architecture of the proposal is collaborative, its enforceability seems diffuse.
In fact, the Declaration does not address critical issues such as civil liability for algorithmic damages, the requirement to carry out mandatory audits, to accredit the traceability of models, or to have cross-border redress mechanisms.
Energy, infrastructure and technological realism
One of the most interesting contributions derived from said declaration is the one that has to do with the explicit recognition of the energy and material footprint of AI, an aspect that has been an inherent risk since the generalization of this technology.
Many countries of the “Global South” cannot sustain energy-intensive models
At a time when large-scale models consume massive amounts of energy and water, linking democratization with efficiency is not only an environmental issue, but once again becomes a geopolitical aspiration: many countries of what has come to be called the “Global South” (nations in Africa, Latin America, and parts of Asia and Oceania, which seek to develop and apply AI adapted to their own cultural, linguistic, and socioeconomic contexts) cannot sustain energy-intensive models.
Such being the case, if the “AI for all” requires having data centers that turn out to be unfeasible due to being located in contexts of energy scarcity, the declaration remains a mere hope.
Human capital and sovereignty
The Declaration dedicates a complete pillar to the development of a recurring issue in any debate on AI and human capital: AI literacy, re-skilling, training of public officials and adapted ecosystems. In this sense, it can be concluded that the growing and structural need for human talent eclipses the debate on the current infrastructure gap.
Nevertheless, an inevitable tension appears, which takes the form of respect for national sovereignty.
In this sense, in a world of fragmented supply chains and export and foreign investment controls, the technological sovereignty of territories does not depend solely on their political will or on attracting talent. Such autonomy seems to depend on access to strategic hardware and software (in particular, to latest-generation chips), on the adequate protection of intellectual property and on the formalization of solid industrial alliances.
Between idealism and strategy
The Declaration of New Delhi is, in addition to a moral ambition, a strategic bet for India, which wants to position itself on the new global playing field as a bridge between the blocks of technological dominance, and as an architect of equitable access to AI.
It remains to be seen if the application of the Declaration determines if “AI for All” becomes a new axis of global governance or, simply, remains as one more declaration in the diplomatic archive of the technological revolution. It will depend on its capacity to put into practice the essential elements in an international environment where access to technology and the resources to develop it, will continue to be conditioned by the enormous strategic rivalries between countries, which AI seems to have exacerbated.
about the firm:
Francisco Pérez Bes is deputy of the Spanish Data Protection Agency. In addition, he was a partner in the Digital Law area of Ecix Group and is former Secretary General of the National Cybersecurity Institute (INCIBE).