The artificial intelligence debate may be entering a new phase. For several years, the dominant question was how fast AI could advance. Increasingly, the more pressing question is how to ensure that AI advances no faster than our collective ability to understand, coordinate and responsibly deploy it. That distinction matters a great deal.
The recent call by Anthropic CEO Dario Amodei for the AI industry to "pace the frontier" reflects a growing concern that the development of increasingly powerful AI systems may require stronger mechanisms for evaluation, safety and coordination. Yet there is an important question that extends beyond frontier-model development: as AI spreads through countries, industries, professions and economic sectors, who will help coordinate the ecosystems developing around it?
This is where the concept of AI Ecosystem Custodianship becomes relevant, and it is also where the Country + AI Platform model being developed through the AI First Gates initiative may have a potentially important role to play. Importantly, however, custodianship should not mean central control of AI innovation. It should instead mean creating an environment in which innovation can develop safely, visibly, responsibly and with appropriate connections to the wider ecosystem.
— The distinction between AI development and AI deployment
The AI ecosystem contains two related but distinct challenges. The first is the development of increasingly capable AI systems, which is where companies such as Anthropic, OpenAI, Google and other frontier developers operate. The second is what happens once these capabilities spread into the real economy, as AI is progressively incorporated into healthcare, finance, government, law, education, agriculture, manufacturing, construction, logistics, energy, cybersecurity, transportation, professional services and countless other activities.
This second challenge is far more distributed than the first. It involves thousands, potentially millions, of organisations, professionals, entrepreneurs, researchers and users, and it cannot realistically be solved by a handful of AI companies alone, nor should it necessarily be solved through a single centralised institution. This is where the concept of ecosystem custodianship becomes useful.
— What is AI Ecosystem Custodianship?
AI Ecosystem Custodianship is the principle that an AI ecosystem can benefit from a trusted, neutral and continuously maintained layer that helps participants connect, understand, coordinate and navigate the ecosystem around them. The custodian does not need to own the AI, develop foundation models, control every application, determine which innovations succeed, or prevent experimentation. Its role is fundamentally different: to provide structure around innovation without controlling the innovation itself. That distinction is central to the concept.
— Custodianship is not control
The word "custodian" can easily create the wrong impression. A custodian is not necessarily a regulator, nor an AI gatekeeper deciding who may innovate, nor an institution that dictates which applications are permissible simply because they are unconventional. Custodianship should never become a justification for suppressing competition or entrepreneurial experimentation. The concept is closer to stewardship: a custodian helps maintain the environment in which an ecosystem operates, offering reliable information, ecosystem mapping, common principles, standards information, connections between stakeholders, responsible-AI resources, transparency, interoperability, infrastructure information, opportunities for collaboration and connections to other ecosystems, while the innovation itself remains distributed.
Custodianship provides the environment for responsible innovation; it does not prescribe the innovation itself.
— Coordination without centralisation
This distinction may become increasingly important as AI develops, because there are two undesirable extremes to avoid. The first is fragmentation, in which every organisation develops AI independently, with little coordination, limited interoperability and inconsistent approaches to safety and responsibility, producing duplication, confusion, distrust and potentially dangerous gaps. The second is centralisation, in which a small number of institutions attempt to control the direction of AI development and determine which innovations are acceptable, thereby constraining competition, experimentation, creativity and local adaptation.
A third model is possible: distributed innovation with coordinated custodianship. Under this model, innovation remains decentralised while coordination becomes organised, standards can be shared, information can be connected, responsible practices can be promoted, and experimentation remains open. This may offer a better balance between technological freedom and institutional responsibility than either extreme.
— The Country + AI Platform as a national custodian
This is the rationale behind the Country + AI Platform model. Such a platform can potentially serve as a neutral digital gateway into a country's AI ecosystem, not to control the country's AI development, but to help make that ecosystem more visible, connected and navigable. A mature national platform, for example, could connect information relating to government AI policies, programmes, institutions and public-sector initiatives; to industry, meaning the companies developing or deploying AI; to infrastructure, including the datacentres, cloud, telecommunications, energy and compute that support AI; to research conducted through universities, laboratories and researchers; to investment, spanning investors, startups and funding opportunities; to workforce development through education and training; to responsible AI, including standards, principles, risk-management resources and relevant regulatory information; and to international cooperation, through connections to AI ecosystems in other countries. The platform thus becomes an ecosystem interface rather than a centralised command centre.
— Innovation can flourish beneath the custodian layer
This is perhaps the most important clarification of all. The existence of a Country + AI Platform does not mean that all AI activity within the country needs to occur through that platform; quite the opposite. The platform can provide a common ecosystem layer while innovation occurs everywhere else — a technology entrepreneur building an AI healthcare company, a lawyer developing a legal-AI application, an agricultural company creating AI-powered farming systems, an engineer designing an AI industrial solution, a university establishing an AI research laboratory, a startup pursuing a new business model, a professional association developing AI standards for its members, or a researcher experimenting with new applications. None of these activities needs to be controlled by the custodian. The custodian's role is simply to help create a trusted ecosystem in which such activities can be discovered, connected and responsibly developed.
— The model can extend far beyond Country + AI
This is where the concept becomes particularly interesting, because the same architecture can operate at multiple levels. At the national level, one can imagine Nigeria + AI, Vietnam + AI, the Philippines + AI, India + AI, Saudi Arabia + AI and potentially many other national ecosystems. But the model does not stop at geography. It can extend vertically into industry and profession, encompassing Healthcare + AI, Finance + AI, Agriculture + AI, Energy + AI, Manufacturing + AI, Construction + AI, Logistics + AI, Insurance + AI, Legal + AI, Medical + AI, Accounting + AI, Engineering + AI, Architecture + AI and Education + AI, and increasingly into highly specialised applications such as Property + AI, Mining + AI, Supply Chain + AI, Cybersecurity + AI, Robotics + AI and Semiconductor + AI. This creates an enormous landscape for innovation, and the custodian does not need to build all of these applications; it can simply provide the ecosystem layer within which they emerge.
— From Country + AI to Industry + AI
This creates an important evolution in the AI First Gates thesis. The first question was which geographic AI gateways would become strategically important. The next question is what ecosystems will develop behind those gateways, and then what industries, professions and sectors will develop specialised AI ecosystems of their own. This suggests the AI economy may develop simultaneously along two dimensions: a horizontal dimension running from country to region to global network, and a vertical dimension running from industry to profession to sector to specialised application. These dimensions can intersect, giving rise to combinations such as Nigeria + AI + Healthcare, India + AI + Manufacturing, Saudi Arabia + AI + Energy, Vietnam + AI + Electronics, and the Philippines + AI + Business Services. The resulting ecosystem is much larger than a mere collection of domain names; it becomes a framework for understanding how AI is being embedded into the global economy.
— The global network
The national platforms can potentially become interconnected. One can imagine a future network in which NigeriaAIPlatform connects to other national platforms, national platforms connect to regional networks, industry platforms connect across countries, professional AI ecosystems exchange knowledge, research institutions collaborate internationally, and companies discover opportunities across multiple markets. This does not require every platform to be centrally controlled; it could instead operate as a federated network, in which each national or sectoral platform retains its own identity, governance and local priorities while participating in common principles of interoperability, transparency and responsible development. The architecture would therefore be one of local innovation, local custodianship and global connectivity.
— Why this matters for developing economies
This model may be particularly relevant to countries that are not currently producing frontier AI models. A country does not need to build the world's most advanced foundation model to participate meaningfully in the AI economy; it can instead develop strengths in AI applications, data, telecommunications, financial technology, healthcare, agriculture, industrial automation, research, education, AI infrastructure, AI-enabled services and specialised professional applications. The challenge lies in making these capabilities visible and connected, and a Country + AI Platform can potentially provide part of that missing connective infrastructure, helping transform an otherwise fragmented collection of companies and institutions into something that can genuinely be understood as an AI ecosystem.
— The relationship with AI safety
The recent discussion about AI safety makes this concept particularly timely. The argument for stronger evaluation and coordination at the frontier is fundamentally about creating institutional mechanisms that can keep pace with technological development, and the same principle can be considered at the deployment level. As AI enters more parts of the economy, a number of ecosystem questions arise: who provides reliable information; who connects relevant stakeholders; who helps identify emerging infrastructure; who makes responsible-AI resources accessible; who connects national ecosystems internationally; and who helps an ecosystem understand where it is heading. These questions cannot be answered solely by frontier AI developers, nor solely by governments; they require participation from the broader AI ecosystem. This is where custodianship can complement, rather than replace, regulation and governance.
— Custodianship should enable creativity
There is an even deeper implication here. A well-designed custodian layer could actually increase innovation rather than reduce it, because innovators benefit from visibility and connectivity. A startup can find potential partners; a researcher can find another research group; an investor can discover an emerging company; a government can identify relevant infrastructure providers; a professional can discover new AI tools; a company can find opportunities in another country; and a university can connect with international programmes. The custodian therefore becomes a connector of possibilities. Its success should not be measured by how many innovations it controls, but by how effectively it helps an ecosystem discover, connect and develop innovation.
— A new architecture for the AI economy
The traditional model of AI can be represented simply as AI developer, AI product, AI user. The emerging economy is considerably more complex, and a broader architecture might be described across six layers.
- Frontier AIFoundation models, agents, advanced AI systems and robotics.
- AI infrastructureCompute, datacentres, semiconductors, cloud, telecommunications and energy.
- National and regional ecosystemsGovernments, companies, universities, investors, startups and users.
- Custodian platformsTrusted national, regional, industry and professional ecosystem gateways.
- Specialised innovationIndustry AI, Profession AI, Sector AI and countless specialised applications.
- Global connectivityNetworks connecting these ecosystems across borders.
This architecture does not centralise AI; rather, it organises the environment around a highly decentralised innovation economy.
— From AI land grab to AI ecosystem buildout
The initial AI First Gates thesis focused attention on the emergence of Country + AI digital territories. The strategic significance of these assets, however, should not be reduced to domain ownership. The more important question is what legitimate, useful and trusted infrastructure can eventually exist behind these digital addresses. That is where the concept evolves: the domain can become the address, the platform can become the operating layer, the custodian can become the stewardship layer, the network can become the connectivity layer, and thousands of independent innovators can build above and around them. This represents the transition from digital real estate to digital ecosystem infrastructure.
— A potential framework for AI Ecosystem Custodianship
The concept can be summarised through five principles.
- Custodianship, not controlThe custodian coordinates and facilitates without owning or directing every innovation.
- Standards, not uniformityCommon principles can coexist with different approaches, applications and business models.
- Connectivity, not centralisationThe objective is to connect participants rather than bring all activity under one institution.
- Responsible deployment, not restricted innovationSafety should build confidence for innovation, not serve as a blanket justification for preventing experimentation.
- Local ecosystems, global networkCountries, industries and professions can develop their own ecosystems while remaining connected internationally.
— The future may require both acceleration and restraint
There is no contradiction between wanting AI innovation to continue and recognising the need for stronger safeguards; the real challenge lies in finding the right balance. AI needs entrepreneurs who experiment, researchers who explore, companies that compete, professionals who adapt, governments that govern, infrastructure providers that build, and institutions capable of helping all these participants operate within an increasingly complex ecosystem. The objective should therefore be neither to slow innovation for its own sake nor to accelerate blindly, but to build the institutional capacity necessary for innovation to proceed responsibly at scale. This is where AI Ecosystem Custodianship becomes relevant.
— From concept to experiment
The Country + AI Platform model is not being developed solely as a theoretical proposition; it is already being tested through the limited activation of selected platforms. NigeriaAIPlatform.com, VietnamAIPlatform.com and PhilippinesAIPlatform.com currently carry monthly ecosystem updates covering June, July and August. The purpose of this limited activation is not to suggest that these platforms have already become mature national AI institutions, but rather to experiment with how a Country + AI digital gateway can be maintained, updated and presented as an active ecosystem platform.
The first experimental lease is still planned for NigeriaAIPlatform.com, which will provide an important real-world test of a further proposition within the AI First Gates model: whether a strategically positioned Country + AI digital asset can transition from ownership and limited activation into an independently operated platform. This distinction matters. The experiment is not about controlling AI activity in Nigeria; it is about testing whether a trusted digital gateway can provide useful infrastructure around an increasingly decentralised AI ecosystem. If successful, the model could eventually extend beyond geography into Industry + AI, Profession + AI and Sector + AI platforms, creating multiple layers of specialised innovation connected through a broader network. The objective, in short, is not to build a centralised AI system, but to test whether distributed innovation can be supported by a network of trusted digital ecosystem gateways.
— UKAI: an early example of the industry-custodian concept
The custodian concept does not have to begin with a country; it can also emerge around an industry, profession or sector. A useful real-world example is UKAI, the UK trade association for the AI sector. UKAI describes its role as championing UK businesses that develop, use and sell AI products and services, representing AI businesses of different sizes, providing a voice to government and policymakers, helping businesses access global markets, and creating a community in which members can share knowledge and skills.
This is not identical to the proposed Country + AI custodian model, since UKAI is a membership and trade association rather than a national digital platform. Nevertheless, its role illustrates an important principle: an AI ecosystem can have an organising and representative layer without that layer controlling the innovation taking place within the ecosystem. UKAI's activities extend across policy, business representation, networking, knowledge sharing and sector development, and its published work also covers areas including healthcare, life sciences, energy, infrastructure, creative industries, skills, regulation and responsible AI. This is close to what an industry-level custodian could look like: the businesses remain independent, the innovators remain independent, and the technologies remain independently developed, but the ecosystem benefits from an organisation that helps participants connect, communicate, represent their interests and engage with government and other stakeholders. That is precisely the distinction between custodianship and control.
— A potential future role for UKAIPlatform.com
The existence of UKAIPlatform.com creates an interesting possibility within the broader AI First Gates framework. Rather than immediately assuming that the domain should become another generic AI website, it could eventually be given a limited activation similar to the experimental approach currently being used with NigeriaAIPlatform.com, VietnamAIPlatform.com and PhilippinesAIPlatform.com. The purpose would initially be demonstrative rather than to claim that UKAIPlatform.com is itself an established industry institution.
A future limited activation could explore how an industry-focused AI platform might organise information around UK AI businesses, AI industry developments, AI policy, AI infrastructure, research, investment, skills, responsible AI, industry events, sector-specific AI developments and international opportunities. The platform could then potentially provide a digital environment that complements existing institutions rather than attempting to replace them, creating an interesting possibility for collaboration with established organisations such as UKAI. The proposition would not be that another AI trade association has been built, but rather that a digital asset and platform concept has been developed that could provide an additional ecosystem layer for organisations already working to develop the AI industry — a materially different proposition.
— Collaboration rather than competition
This distinction would be particularly important if MyDomainPlan were eventually to approach UKAI or similar organisations. The objective would not be to compete with an established trade association, but to explore whether UKAIPlatform.com could complement the work of organisations already serving the UK AI ecosystem. An established trade association, for instance, could provide membership and representation while the platform provides digital ecosystem infrastructure; an association could provide policy advocacy while the platform provides information and discovery infrastructure; and an association could provide industry events and working groups while the platform provides a persistent digital gateway through which the ecosystem can be presented and connected. This distinction could make collaboration considerably more natural. The platform does not have to replace the custodian; it can potentially extend the custodian's reach.
— The larger implication
This suggests that the AI custodianship model could eventually develop at several levels: geographic custodians, in the form of Country + AI platforms such as NigeriaAIPlatform.com; regional custodians, in the form of Region + AI platforms such as AfricaAIPlatform.com; industry custodians, in the form of Industry + AI platforms such as a potential UKAIPlatform.com ecosystem; professional custodians, in the form of Profession + AI platforms serving legal, medical, accounting, engineering and other professional communities; and sectoral custodians, in the form of Sector + AI platforms serving healthcare, energy, construction, agriculture, manufacturing, logistics and other sectors. These do not have to operate as competing institutions; they can instead become interconnected layers of the same broader AI ecosystem, in which a national AI platform connects to an industry platform, an industry platform connects to professional platforms, professional platforms connect to research institutions, and all potentially connect to an international network.
— From platforms to an AI ecosystem network
This is where the concept becomes particularly interesting, because the future may not be one giant AI platform but rather a network of specialised platforms and custodians, each serving a particular community, retaining its own identity, and supporting innovation within its area, while remaining interoperable and connected to the whole. The architecture could therefore be described as a progression running from Country + AI, through Industry + AI, Profession + AI and Sector + AI, into specialised AI innovation and, ultimately, a global AI ecosystem. In such a structure, organisations such as trade associations, professional bodies, universities, companies and governments can all play meaningful roles: the platforms become connective tissue, the custodians provide stewardship, and innovators remain free to experiment.
— A potential experimental path
The current experimental work with NigeriaAIPlatform.com, alongside the limited activation of VietnamAIPlatform.com and PhilippinesAIPlatform.com, provides an initial test of the geographic model. A future limited activation of UKAIPlatform.com could provide a different experiment, testing whether the same digital-gateway concept can work around an industry ecosystem rather than a national one. If eventually approached, organisations such as UKAI could therefore be viewed not simply as potential users or lessees, but as potential ecosystem partners whose existing industry role could complement the digital infrastructure represented by the platform. No particular collaboration should be assumed in advance; the value of the experiment lies in exploring the possibility. This is consistent with the broader AI First Gates philosophy: secure strategically relevant digital territories first, test their practical utility through limited activation, and allow real-world partnerships and use cases to determine what develops behind them. The objective is not to predict exactly what every AI platform will become, but to create the digital infrastructure from which useful ecosystems can emerge.
— The custodian does not have to own the innovation
This may ultimately be the most important lesson of all. The custodian provides the gateway; the ecosystem provides the participants; the innovators provide the ideas; the businesses provide the applications; the researchers provide the knowledge; the government provides the policy environment; and the global network provides the connections. That is why the custodian model should not be interpreted as centralised control — its purpose is almost the opposite.
A good custodian can create the conditions for a much larger number of independent innovators to participate.
UKAI's existing role as a trade association illustrates one form of this ecosystem coordination in practice; a future UKAIPlatform.com experiment could explore another, namely how a dedicated digital gateway might complement an existing industry institution and connect that ecosystem to the wider global AI network.
— Conclusion: building the environment around intelligence
The AI debate has largely focused on the intelligence itself: who will build the most capable models, who will achieve the next breakthrough, who will dominate AI infrastructure, and who will capture the economic value. These questions remain important, but another question is emerging alongside them — what infrastructure will surround increasingly powerful AI? The answer will not consist only of datacentres, chips, cloud systems and networks; it will also require human and institutional infrastructure, including independent evaluation, responsible deployment, industry coordination, professional adaptation, national ecosystem development, international cooperation, trusted information, and mechanisms for connecting the many participants in the AI economy.
This does not require centralised control. It may instead require something more flexible: distributed innovation with coordinated custodianship. That is the principle behind the emerging AI Ecosystem Custodianship Model, under which Country + AI Platforms can potentially serve as national digital ecosystem gateways, while Industry + AI, Profession + AI and Sector + AI ecosystems provide enormous space for specialised innovation and creativity. The custodian does not have to decide what the future looks like; it helps create the conditions in which different participants can build the future themselves.
The architecture could ultimately be expressed simply: custodians coordinate, innovators create, ecosystems connect, countries participate, and the global network grows. That may be a more sustainable model for the next phase of the AI economy than either unrestricted acceleration or excessive centralisation. The future of AI may therefore depend not only on how intelligent our machines become, but on how intelligently we organise the ecosystems around them.
The AI First Gate may ultimately be more than a digital address. It could become an entry point into an ecosystem — and, potentially, part of the institutional infrastructure that helps that ecosystem develop responsibly.