Regional Leaders Discuss AI-Ready Digital Infrastructure

20 Feb 2026 17:00h - 18:00h

Session at a glance

Summary

This discussion focused on AI infrastructure development and opportunities for the Global South, featuring perspectives from government officials, international organizations, and development banks. Dr. Saurabh Garg opened by emphasizing four key elements for AI-ready data: discoverability through proper metadata structures, trustworthiness via quality assessment frameworks, interoperability through unique identifiers, and usability across systems through common standards and classifications. He also questioned whether current AI models are too infrastructure-heavy, noting that while AI systems require gigawatts of power, humans operate on just 100 watts.

The panel discussion revealed diverse national strategies for AI development. Uzbekistan is investing $300 million in AI development, including $200 million for government data centers with NVIDIA GPUs and a $5 billion energy-efficient data center project with Saudi partners, aiming for $1.5 billion in AI-related exports by 2030. Indonesia faces a “triple deficit” in data infrastructure, compute capacity, and AI talent, with ambitious plans to train 12 million AI professionals by 2030 through their Korika Academy and pentahelix platform involving government, industry, academia, civil society, and media.

The World Trade Organization highlighted that AI could increase global trade by 40% by 2040, but emphasized the need for digital infrastructure, skills, and policy readiness to realize these opportunities. Regional cooperation emerged as crucial for smaller economies that cannot achieve the scale needed for major AI investments alone. The discussion concluded that while AI presents significant opportunities for economic growth and development, success requires balanced approaches addressing infrastructure, skills development, and appropriate regulatory frameworks tailored to each country’s specific context and needs.

Keypoints

Major Discussion Points:

– AI-Ready Data Infrastructure: Dr. Garg emphasized four critical elements for making data AI-ready: discoverability through proper metadata structures, trustworthiness through quality assessment frameworks, interoperability with unique identifiers for data communication, and usability across systems through common standards and classifications.

– Digital Infrastructure Gaps and Investment Strategies: Multiple panelists discussed the “triple deficit” facing developing nations – inadequate data/compute infrastructure, shortage of AI-skilled talent, and limited connectivity. Countries like Uzbekistan are investing $300 million USD in AI development, while Indonesia targets training 12 million AI talents by 2030.

– Regional Cooperation vs. National Sovereignty: The discussion explored balancing collaborative approaches to share infrastructure costs and expertise while respecting data sovereignty. The WTO representative highlighted how regional frameworks like ASEAN’s AI policies and trade agreements can provide economies of scale for smaller nations.

– Private-Public Partnership Models: Panelists shared various approaches to mobilizing capital, from Uzbekistan’s partnerships with Chinese companies like Huawei, to Indonesia’s collaboration with hyperscalers like Microsoft, demonstrating different models for attracting international investment and expertise.

– Contextual AI Solutions and Skills Development: The conversation emphasized that AI applications must address specific local needs rather than adopting one-size-fits-all approaches. Examples included Indonesia’s focus on climate-health nexus for disaster-prone areas and the recognition that employment concerns may outweigh automation benefits in some contexts.

Overall Purpose:

The discussion aimed to examine digital infrastructure challenges and opportunities for the Global South in AI adoption, covering the full spectrum from foundational compute infrastructure to skills development, policy frameworks, and practical implementation strategies across different national contexts.

Overall Tone:

The discussion maintained a consistently optimistic yet pragmatic tone throughout. Panelists were enthusiastic about AI’s potential while being realistic about implementation challenges. The conversation was collaborative and solution-oriented, with participants sharing specific strategies and investment figures. The tone remained constructive even when addressing significant gaps and constraints, focusing on actionable approaches rather than dwelling on obstacles.

Speakers

Speakers from the provided list:

– Dr. Saurabh Garg – Secretary (specific ministry/department not mentioned), focuses on AI-ready data, works with ministries and governments across the country

– Arndt Husar – Moderator/Host of the fireside chat discussion on digital infrastructure

– Johanna Hill – World Trade Organization (WTO) representative, works on AI and trade policy

– Zuhriddin Shadmanov – Ministry of Digital Technology, Uzbekistan, works at the center of development of AI and digital economy

– Hamam Riza – Professor, Co-chair of the National AI Roadmap Indonesia 2030, President of the Collaborative Research and Industrial Innovation in Artificial Intelligence

– Mio Oka – Asian Development Bank (ADB) Country Director for India

Additional speakers:

None – all speakers mentioned in the transcript are included in the provided speakers names list.

Full session report

This panel discussion at an AI summit in India brought together government officials and development finance representatives to examine AI infrastructure challenges and opportunities for developing economies. Moderated by Arndt Husar, the conversation explored practical approaches to building AI capabilities while addressing real-world constraints facing the Global South.

Foundational Infrastructure Requirements

Dr. Saurabh Garg opened the discussion by outlining four essential elements for AI-ready data infrastructure. First, discoverability requires well-defined metadata structures that enable data to be easily found and understood across systems. Second, trustworthiness depends on comprehensive quality assessment frameworks that ensure data credibility. Third, interoperability relies on unique identifiers that allow different datasets to communicate effectively. Finally, usability across systems requires common definitions and standards to prevent confusion and ensure consistency.

Dr. Garg also raised concerns about the energy efficiency of current AI models, noting the stark contrast between AI infrastructure requiring gigawatts of power while human intelligence operates on merely 100 watts. This comparison questions whether the industry is pursuing the right technological path and highlights the need for more efficient approaches.

National Strategies: Indonesia’s Comprehensive Approach

Hamam Riza, co-chair of Indonesia’s National AI Roadmap 2030 and president of the Collaborative Research and Industrial Innovation in Artificial Intelligence, outlined Indonesia’s multi-faceted strategy for AI development. He referenced a “triple deficit” facing developing nations, though technical issues with the audio made his detailed explanation unclear.

Indonesia’s approach emphasizes culturally aligned AI development, recognizing the need for large language models that reflect local contexts rather than relying solely on foreign-developed systems. The country is focusing on climate-health applications, particularly predicting climate-sensitive infectious diseases like malaria and dengue, which are critical challenges for disaster-prone regions.

The Indonesian strategy involves collaboration with international partners, including Microsoft and other major technology companies, to build domestic capabilities while leveraging global expertise. Their approach includes comprehensive skills development programs aimed at creating sustainable capacity building across the country’s vast population.

Uzbekistan’s Investment and Partnership Model

Zuhriddin Shadmanov from Uzbekistan’s Ministry of Digital Technology outlined his country’s significant financial commitment to AI development. The government has allocated $300 million for AI advancement, with $200 million specifically designated for government data centers equipped with NVIDIA GPUs. The country aims to attract $1 billion in AI-related infrastructure investment by 2030.

Uzbekistan’s strategy demonstrates comprehensive ecosystem thinking, combining domestic investment with strategic international partnerships. Their collaboration with Huawei encompasses AI infrastructure development and network advancement. Additionally, partnership with the UAE’s training programs has registered over one million participants, showing how international cooperation can rapidly scale skills development.

The country offers attractive tax incentives and customs exemptions for investors willing to build data centers worth over $100 million, demonstrating how policy frameworks can mobilize private capital for infrastructure development. Their approach includes establishing data lakes that collect government sector data, making it available to SMEs and startups at low cost to stimulate innovation.

Trade and Economic Opportunities

Johanna Hill from the World Trade Organization highlighted AI’s potential economic impact, projecting that AI and trade integration could increase global trade by 40% by 2040 – what she termed the “40 by 40 effect.” However, she emphasized that realizing these opportunities requires comprehensive digital infrastructure, skills development, and policy readiness rather than simply technological deployment.

Hill noted that regional cooperation is particularly crucial for smaller economies that cannot achieve the scale needed for major AI investments independently. Regional frameworks and trade agreements can help countries develop coordinated approaches and achieve economies of scale.

Development Finance Perspective

Mio Oka from the Asian Development Bank provided a pragmatic view of AI infrastructure investment, emphasizing that service-level applications in agriculture, water supply, and irrigation may offer more immediate impact given the scale of populations in the Global South. The ADB’s strategy focuses on mobilizing private capital through integrated approaches that combine infrastructure development with AI considerations.

Oka shared a revealing anecdote about proposing AI-based fish feeding systems for aquaculture development. The negotiations ended abruptly when government officials prioritized employment concerns over technological efficiency. This experience highlights the critical importance of understanding local development priorities and ensuring that AI solutions align with broader socio-economic objectives rather than pursuing technology for its own sake.

Skills Development Across Society

The discussion revealed sophisticated approaches to skills development that extend beyond technical training. Uzbekistan’s strategy covers all levels of society, from students and professionals to public servants, recognizing that successful AI adoption requires broad-based understanding rather than concentrating solely on the technology sector.

Indonesia’s training programs emphasize “train the trainers” approaches aimed at creating sustainable capacity building that can scale across large populations. The skills challenge extends beyond individual capacity to institutional readiness, with both countries emphasizing the importance of preparing government institutions for AI implementation.

Balancing Innovation with Employment Concerns

A significant theme emerged around balancing AI’s potential for economic transformation with legitimate concerns about employment displacement. Oka’s aquaculture example illustrates how technological solutions that seem beneficial from an efficiency perspective may conflict with development priorities focused on job creation and poverty reduction.

This tension requires nuanced approaches that consider AI’s role in economic development rather than simply technological advancement. The emphasis on human-centered AI development in national strategies reflects recognition that successful AI adoption must enhance rather than replace human capabilities, particularly in economies where employment generation remains critical.

The Path Forward

The discussion highlighted that AI infrastructure development must be understood as part of broader development strategies rather than as a standalone technological challenge. Success requires integrated approaches that address infrastructure, skills, governance, and practical implementation simultaneously while remaining sensitive to local contexts and development priorities.

Moderator Arndt Husar referenced the ITU’s framework of “3 S’s” – solutions, standards, and skills – as essential elements for AI development. He also noted the ADB’s partnership with the summit and their working group on democratization of AI compute, representing efforts to address shared infrastructure challenges for smaller economies.

The conversation demonstrated mature understanding of AI development challenges, moving beyond simple technology adoption to comprehensive ecosystem thinking. The emphasis on human-centered approaches, regional cooperation, and context-sensitive strategies suggests that developing nations are creating sophisticated frameworks for AI adoption that balance technological advancement with local development needs.

Session transcript

Dr. Saurabh Garg

models or talent, how we can ensure that it works in a federated manner. I think I’ll just, I was discussing and maybe I’ll just focus on one piece, which is on AI -ready data, if I can focus on that and leave it for the esteemed panelists on the large number of issues. Some of the elements that we are focusing on include, one is on how to make it more discoverable. That would be a very basic point to ensure that it’s discoverable. Second is how to ensure that the data sets are trustworthy, and that would be the second element. The third would be on the interoperability, and the fourth would be on the usability across systems.

So on discoverability. On discoverability, the metadata of that structure is extremely important, so that would help to, that’s a first. element of having a metadata structure which is understandable and well defined and can be used across the second on the trustworthy part would be the quality assessment so we’ve developed as kind of a quality assessment framework which focuses on the quality of the data so that to ensure credibility on the data interoperability a lock would depend on whether data can talk to each other what is the unique identifiers that we have which will ensure that the different data sets whether are they talking about the same thing or different we are able to identify that and the fourth would be on the usability across systems would be based on the standards and classifications that we have whether it’s a common definitions and common standards so that two sets of data don’t refer to the same thing and I suppose this really forms the bedrock of making a data AI ready and that That’s something that we’re working with ministries and governments and state governments across the country.

And given the importance of data sets in the AI infrastructure, it has an important part to play. The other aspect on data is also on its dissemination and access, on how we are able to ensure that data sets in themselves have value beyond AI and what kind of dissemination and mechanisms can be there which will make it usable for people to leverage them for business while preserving the privacy aspects of individual data. One other thing, since we are talking about and the panel will be having discussions on AI infrastructure, I just wanted to focus on one thing that I think discussed, over the past couple of days. has also come up that the existing models seem to be extremely data infrastructure, infrastructure heavy, whether compute infrastructure, data infrastructure.

And every time a new query is put out to the model, is it necessary for the billions of bytes to be again run through again and the gigawatts of power that we need? And are there alternative mechanisms available? And I just want to highlight yesterday one comment which stays with me, is what Vishal Sikka had made, that when we talk in terms of AI infrastructure, we talk in terms of gigawatts of power. Compared to that, a human being requires 2 ,000 calories, which is only 100 watts. So are we missing something out there in the infrastructure? And perhaps a greater focus on the models going forward is there. So I’ll stop here. Thank you for inviting me.

Thank you.

Arndt Husar

Thank you so much, Secretary. And I’m now going to join the fireside chat here. The discussion that we have planned will cover various different aspects of digital infrastructure. So when you hear digital infrastructure, you might be first thinking of the data centers and the compute. But we actually want to have a conversation that also encompasses the solution side, the skill side, so that we really look at the whole spectrum of infrastructure, even standards. So these three S were introduced yesterday by ITU’s head, the three S of solutions, standards, and skills. Kind of a nice way to open up to the panel. We have different perspectives here today. And we’re going to try to stick to time.

But let me introduce you to this panel by asking the first question. And I would request that each of the panelists then quickly states their name and their institution to shorten the time. What we would like to hear from each of you is that from your vantage point, what do you see as the most critical gap, the most exciting opportunity for the global south in generating positive impact through AI? So we’ve asked each of them to think about a concern or opportunity and then also to maybe link it to strategy or vision. So maybe I’ll first go to the lady on my right from the WTO. May I request you for your perspective on the big challenge or opportunity?

Johanna Hill

Thank you so much to the Asian Development Bank for the invitation and the organization to this interesting conversation. My name is Johanna Hill. And I… I am with the World Trade Organization. And let me start out with the opportunity side of the equation. We really are seeing that AI and trade, when they work together, can offer important opportunities for developing countries and low -income economies. Our projections at the Secretariat have led us to believe that by the year 2040, trade could grow by almost 40%. So that would be the 40 by 40 effect. But then here come the caveats, right? For that to happen, for those opportunities to really be realized, one element that is really important is the digital infrastructure, the skills that you mentioned, and policy readiness.

You know, we’ve heard throughout this conference and before the important opportunities and applications in different sectors, in agriculture, health care, new services being developed as we speak, new services and goods that are becoming more AI -related, more tradable, and we are also seeing that that can have important opportunities. for the smaller firms in developing economies and in the big economies also. We did a survey with the ICC that we published last year on the opportunities for businesses and for small and medium enterprises. And of those respondents, many of them were saying that they’re already using AI, of course, from bigger companies, more developed economies. But even the smaller firms are also seeing opportunities in areas like market intelligence.

So we do see that it can be a game changer.

Arndt Husar

Fantastic. And one of the things that I’ve been hearing a lot at this summit is that specifically the SMEs, the technology has moved so fast that there’s a huge adoption gap and understanding of how they can actually integrate the AI into their business models, into their little shop that takes a picture of a product and uploads it quickly. AI can be super helpful in this but hasn’t yet reached that audience. Maybe I’ll turn it over to you. Maybe I’ll turn to the other side and request our colleague from Uzbekistan to share his perspective.

Zuhriddin Shadmanov

Thanks for the question. Thanks for having me here. Let me talk about the gaps which exist in our country. I think the first one is unequal access to compute capacities and I think advanced AI digital skills. So in that sense, these foundations play a crucial role because if you don’t bridge those gaps, many countries, nations will be just the consumers of AI rather than creators of AI value. So in that sense, Uzbekistan is advancing strategic ideas. First one is developing human skills. So in that sense, Uzbekistan is advancing strategic ideas. First one is developing human skills. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas.

So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. So in that sense, Uzbekistan is advancing strategic ideas. all stratus of our nation, starting from students, professionals, and public servants. So we are not concentrating on the tech sector, but also we try to cover all the spheres of our nation.

And secondly, we are developing our infrastructure. For that reason, our government is allocating around 200 million USD. So to create our own government data center with supercomputers, GPUs, acquiring from NVIDIA. And also we are working with DataVault, a Saudi Arabian company, to create an energy -efficient data center, which is based on renewable energy. It is a very big project. It’s around 5 billion USD. And the data center will be… put into operation within two, three years. hopefully and also we are trying to develop our government strategy we adopted a strategy 2030 last year and this by by the year to 2030 we’re trying to get there early reach the export of AI related products by five 1 .5 billion USD so

Arndt Husar

fantastic so either by coincidence or planning you touched on the 3s the solutions the skills and the standards the policies fantastic thank you so a very comprehensive view with multi -pronged strategy that you didn’t introduce yourself so I just say you with the Ministry of Digital Technology and an institution quite focus the center of the development of AI and the digital economy Fantastic. Okay, let me turn back to this side. So from Indonesia, we have someone who’s actually in this skills domain. Would you like to share with us what you, from your vantage point, perceive as the key opportunity or challenge?

Hamam Riza

All right, thank you. Hi, everyone. I am Professor Hamam Riza. I am the co -chair of the National AI Roadmap Indonesia 2030. And also I am the president of the Collaborative Research and Industrial Innovation in Artificial Intelligence, the organization that was founded in 2020 when we launched our first national AI strategy towards Indonesia 2045. That is the vision. And I think AI will take us there, really. So from my vantage point, I think… We are… we have no we need to move beyond numbers even though we understood that AI economy will create millions of jobs and also potential economy of up to 1 trillion so from my vantage point from the Indonesian perspective global south they are basically triple deficit in terms of what we are going to it is the most challenging one the first one is that certainly about the data and infrastructures, the compute infrastructures we are still lacking the connectivity the networks but as I have marked down here in order for us to smooth up all the AI use cases for public services for health services, for agriculture, and many other things, you need to basically solve this triple deficit.

And that is also regarding how you need to develop the AI talents. There is a significant lack and scarcity of high -quality localized data centers tailored to Indonesia, as well as shortage of AI skilled talents that limits the capacity of long -term innovations capabilities. So our government is addressing these gaps through the national roadmap that I co -chaired. And our primary concern is how the digital divide and how the AI divide, which is the digital divide. Which is created by this. generatively I didn’t think I towards many of the public sectors in Indonesia in an in general in the global self that we can tackle so that while there is nine to two percent of our skill knowledge worker but we are still using you know a very basic AI tools and needs to be aware of all the and all the risks you know applied to the output of this AI tools so those those things are that I think will be my point of view towards closing the gaps for the global south and especially for Indonesia

Arndt Husar

thank you so much and it’s of course one of the most populous countries in Southeast Asia It’s a very young workforce. 270 million. Yes, and startup buzz in Indonesia is also palpable, so lots of potential in Indonesia. Last but not least, I want to go to our Asian Development Bank Country Director for India. Mio, can I request you to share your views?

Mio Oka

Thank you. From ADB’s perspective, of course, foundation is important. We need to have a power supply, stable power supply, and the devices that people have access to, and reliable broadband, even in our office. So that’s a foundation. But we are in India. Do we expect India to put so much money on foundation to have a ground -level impact? But India has a scale. So what we need to focus is, as others already said, is a service. So we work on agriculture sector, water supply, and even irrigation sector where AI is widely applied. Because of the scale of the people that we have in the global south, while we work on the foundational infrastructure, at the same time we really have to work on how AI can be applied at the service level.

And this is where ADB would like to support. Thanks.

Arndt Husar

Thank you, Mio. So we have, as you can see, different perspectives at the same space of how do we get at grappling with this massive development opportunity that AI represents. For this first round of questions after the opening, let’s go into the foundations a little bit. I want to go to Uzbekistan again. And you already mentioned you have… ambitions on infrastructure, policy and skills. Now, how do you actually balance this in terms of priority setting? Can you go for all of them all at once? How do you finance it? Does this keep you up at night, how you balance these three different strategic objectives?

Zuhriddin Shadmanov

It’s a tough question because we are a developing nation and money is always a scarce element for us. So anyway, our government is trying to allocate enough resources so we can cover all the aspects of AI development to create the AI ecosystem. First one, as I already mentioned, it’s a strategy. 2030, which sets our priorities, which is human -centered AI. And secondly, with the government trying to allocate enough resources overall now the government announced about 300 million USD for development of AI and the money goes to first of all implementing projects in the government sector in the social sphere healthcare, education, transportation cyber security and etc and also government is trying to provide necessary infrastructure building data centers acquiring GPUs and also we are now creating a data lake which will be collecting the data of the government sector so SMEs, startups and other who wants the data they can use those data for free or for some money usually free and anyway we’re trying to work with other countries as I already said that we are we have a good project five million AI leaders so United Arab Emirates they helping us to helped us to build this program and it was launched now over 1 million people already registered and go to training certifications there so also we are trying to attract foreign foreign investments and now government announced very good in tax incentives and other incentives to for example if you are want to invest in a you know in Uzbekistan and try want to build a data center which costs over 100 million USD you will get very cheap and take the intensive and customs exams and etc so going trying to balance with cautiously but still providing necessary conditions for the to build a ecosystem

Arndt Husar

very impressive and since I had the opportunity to chat with somebody else from Uzbekistan this week I also know that in your KPI as a public servant AI roll it out has entered that KPI space so that’s always going to make a difference let’s go back to Indonesia now I’m gonna you know of course skills is your comfort zone but can I ask you about the infrastructure side I know that the hyperscalers the big cloud companies international companies have invested significantly in Indonesia now how do you see that now moving into the AI age and is that a big step forward for you? is there a lot of activity on additional infrastructure build out what do you see happening in that

Hamam Riza

yes so I see these questions and I’m really eager to answer this because suddenly our infrastructure is undergoing a transformation really to meet the demands of the AI demand and certainly with the ability of many of these new infrastructures coming out of the government and also from the business I think benchmarking with many other countries including you know in the regional ASEAN take for example the presence of the global hyperscaler in the country have established actually multiple cloud regions in Indonesia. But certainly this needs to be amplified because as you know 10 years in many other technologies, one year in AI, right? That’s what they are saying. So how do you can fulfill this demand of AI compute massive data for training because you need to build up our own for example large language model that can align with our cultures.

So those hyperscalers needs to move beyond just being a single a host for this you know many of the AI models from outside of the country right so and the infrastructure readiness is also being federated by our chief toward the sovereign AI we are now preparing the presidential regulation actually to push forward the innovations the investment and we need to collaborate with many of the hyperscalers and we are ensuring that the physical infrastructures like the GPU data center and localized edge computing yeah is going to be present in the country. And one thing that the Vice Minister of Communication and Digital Affairs mentioned to me yesterday that we are struggling building up the ecosystem. That means there will be special economic zone for these hyperscalers and new data centers being brought forward in order to align and be part of our national AI roadmap, AI journey in Indonesia.

And we are going to prepare ourselves in this AI transformation so that our data… digital consumer… is going to be part of our transformation. The technology is accessible for all. Even, you know, what we are right now, you know, participating in the India AI Summit says about democratizing AI for all. So I think that is a very significant theme that is also part of our national AI roadmap. Thank you.

Arndt Husar

Fascinating. And, again, I think you as a large economy, you have that opportunity similar to how India is also portraying it this week of really wanting to develop your own, you know, language models and really playing in that league. However, there are many countries. also countries we work with who don’t have that kind of scale and who need to look at it quite differently. So the different nuanced strategy that you mentioned of investing into the big AI, the small AI, the edge AI, all these different pieces, very interesting. With this dynamic, can I turn to WTO? How do you see trade competitiveness evolve? That’s really your space where you are at. What are those interesting approaches that are emerging which could help support maybe the cross -border collaboration while you also, of course, respect data sovereignty?

Countries will need to collaborate, right? There’s not enough money to go around for everybody to play in that top league. So trade competitiveness, what do you see there?

Johanna Hill

So I was talking about the opportunities of trade growing by the use of AI. Okay. And if you think about… That growth comes from the lowering of trade costs. It comes from powering AI -enabled goods and services crossing borders. And it’s… Also, new products and services that are going to be invented are being invented by AI. And when you talk to business, when we asked through the survey, some of the constraints that they are having and doubts in the use of AI have to do with competing regulations and having a high cost in trying to comply. And fragmentation is actually an area of data, for example, that can become a problem. And so we developed and published last year in the World Trade Report what the Secretariat calls the AI Trade Policy Openness Index to help regions measure how they’re doing in that space.

And in there, you can see, for example, that some of the lower income economies can seem quite open in that space. But it might be because of the lack of regulation. And when you talk about AI, I think what a lot of countries and customers are saying is that AI is not a good thing. What customers are looking for is, you know, it’s AI. that is responsible, you know, trading AI with trust. So just not having regulation can also be a disadvantage to your competitiveness. So starting to look at those things that way. And then in the part of the solution side, definitely the regional approaches are important, those collaborations, and sharing infrastructure, for example.

When you don’t have those economies of scale, those huge investments come in your way. And then not every single company or every single country is looking to be on the edge of things necessarily, but we do want to adopt AI to boost our economy and our competitiveness.

Arndt Husar

Well, thank you so much. And I don’t know whether people heard about this new initiative that the working group on the democratization of AI, of compute, has come up with. ADB is actually supporting that. Really, this is… This has not yet evolved, right? This collaboration on the infrastructure. How do you share that properly across borders? It’s still new territory and very interesting to see. Can I turn to my colleague, Mio, and request her to talk a little bit about the engagement that we’ve had with member countries. What does demand to ADB actually look like in this space?

Mio Oka

try to invest in the township planning and the implementation. Also, we can have a water supply road project that can be connected to the industrial parks so that private sector can invest in the digital -related facilities. So mobilization of private capital is one. And the second is it’s an application across sectors. We just don’t look at the single sector project. As I said, we can work on road and water at the same time. And while we work on the Agri -AI project, we work with the building capacity of that institution as well so that they can handle the AI. And the third is the knowledge. As I said, we support quite a bit of this master planning or the strategy development at the municipality or the state or even at the regional level.

We see India. And you’ve been coding on science. India. And of course we always bring in the international experts so that India can learn and also this is a good opportunity for India to expand their capacity to outside countries. Thank you.

Arndt Husar

Thanks, Miyu. I actually had a follow -up question for you that would have touched on this de -risking and catalyzing investment topic. Maybe I’ll let you ask you to repeat that now, but let me just add our digital sector office being fairly new. They are getting a lot of demand for guess what? Data centers. And, you know, we welcome that. We have conversations with government but I’m truly impressed with the conversations at the summit here. Earlier this morning I attended one where the state of Telangana was sharing what they’re thinking about and they’re really quite cognizant of the kids to school not many kids fit into a Ferrari milk doesn’t make sense so we need to look at what type of compute is needed for what and I think we in ADB are also learning more and more how to engage in these conversations properly we’re learning alongside everybody else in this room probably and that’s an important distinction to make because it will influence the financing bit how much do we need, what do we actually need and when and how do we make that investment sustainable just wanted to add that it’s an insight from this morning that I couldn’t not retail.

Let me go back to Indonesia and ask you about cutting -edge skills because you’re in that space. I found it very interesting that you’re actually, you said co -president or co -chairing this platform where you bring together private sector, education sector, government. And as you are looking at that, how is your organization doing it in practice? How do you bring these people together and get them into action mode? How do you do that?

Hamam Riza

Okay, thank you. Very important question here, I think. So I would like to say in three pillars that what we are doing, especially that we chair the AI ecosystem in Indonesia where the government, the industry are involved in the AI ecosystem. Within an academia as well as the… civil society as well as media we call this pentahelix platform we discuss about I think three pillars first one is the talent certainly second one is infrastructures and the third one is how basically we can articulate use cases towards all the services public services and businesses as well so Indonesia for talent we have set our target quite ambitious that we want to have at least 12 million talent by 2030 and for us this is something that are uh fairly challenging, considering that we are still lacking around 3 to 5 million talent as of now, right?

So what we are trying to achieve together with the whole ecosystem is to establish an academy, the Korika Academy, where we promote to not only upskilling and reskilling some of the civil servants and other workers, but we are also looking at how we can train the trainers. We work with several of our friends. I will note here that Elevate Indonesia, for example, part of the… Microsoft and many others big tech that are there works together with our ministry to establish this program for Thailand

Arndt Husar

and it’s a digital academy or is it a physical?

Hamam Riza

it’s a digital academy with the LMS learning management system and many other things we also established the Kodika chat actually it’s a chatbot for this training and upskilling program that we do with the government beyond Thailand basically we are aggressively looking at how we can nurture this talent to work in data centers, in many others startups and incubators as well as to establish some of the most diverse demanding use cases So the third one is we try to work on climate health nexus in establishing how we counter and predict the climate sensitive infectious disease such as malaria and dengue. And we have established for the past three years the Climate Smart Indonesia which have attracted many of the universities as well as NASA pollution and air quality programs to look into these use cases.

So we can basically reach out to many of the areas where the… …the health, the disaster prone area because Indonesia is a supermarket for disaster. You can have the hydromelectorological disaster, you can have ecological, you can have many things. So you need to…

Arndt Husar

I’m not buying any of them.

Hamam Riza

Of course, we don’t want to be shopping.

Arndt Husar

So really amazing this focus also on the use cases, right? And prioritizing those that match with your country needs. Yes, thank you. Give the highest impact, right? Super. I’ll turn back to Uzbekistan and just wanted to ask you to elaborate a little bit in terms of private sector capital mobilization. You have all these ambitions you shared across the board really in terms of infra, in terms of skills and so on and so forth. Uzbekistan as an economy has still… a good chunk of traditional economy but also has a very active startup sector that I’m learning more and more about how dynamic people are around the region, Central West Asia going and finding scalable solutions but these are the still growing companies for mobilizing capital for your infra you’re going to need the big ones or you’re not going to need the international partners or what are you thinking about this private capital mobilization, what’s your strategy there?

Zuhriddin Shadmanov

First of all I should mention that according to the documents adopted by year 2030 we are planning to attract around 1 billion USD for investments for creating AI related digital infrastructure and part of this goes to creating data centers and we’re going to need to And also we’re working with our Chinese partners also. It’s the biggest IT company, Huawei. So they’re also involved in creating AI ecosystem in Uzbekistan. Mainly, first one is upskilling public servants to help them to adopt AI adoption and also creating the necessary training programs for the specialists and also creating the AI infrastructure like data centers, data lakes. And also we need to get, we are transferring to 5 .5G and also working on 6G also with Huawei.

So, yes. And also, as you mentioned already about startups, we are developing our own. startup ecosystem and we established many venture funds funds of funds and also there are many emerging private funds so they are now trying to invest in startups attracting private funds, private investments so currently we have allocated around 50 million USD for AI startups so they are already providing services both for public and for businesses so trying to balance and attract all the stakeholders of the ecosystem

Arndt Husar

Fantastic, so you’re mixing also your public funds that you invest for example in the fund of funds and then bringing in more investment domestically from your investors but also from abroad That’s amazing. And then having large industry partners that are interested in the market, bringing them in like Indonesia did with some of the hyperscalers. You are bringing in Huawei and Chinese partners. So basically it’s a mix of different strategies you mentioned. Also, that’s fascinating. Again, Uzbekistan being one of the larger countries in Central West Asia and Indonesia, both fairly large in their region. And then, of course, again, I want to come back to this point about diversity of country context. That’s both a challenge but also an opportunity.

I mean, for us at ADB, it adds, of course, complexity because we need to respond to these different needs. But from the perspective of WTO, is there like a specific area such as maybe interoperability standards or AI talent mobility? Or the shared data set? joint research, where do you see regional cooperation making the biggest difference? I

Johanna Hill

think that it’s a bit of a matter of context, right? At the regional level and at the national level. We’ve talked about the divide in the digital divide and how do we overcome that and the role of infrastructure and skills and the rest. And at the WTO Secretariat, we’ve been very concerned on this issue. And so we partnered with the World Bank and we did a study called Digital Trade in Africa, a general one, and then we did some country pilot studies to look at the situation. And we did see that some of the regional work, like the ACFDA and the digital protocol, really made a difference in how it helped bring them along and to set a certain standard in many of the countries that we studied.

Then we did a similar study with the World Bank in Latin America and the U.S. The Caribbean and the Inter -American Development Bank partnered with us. And we saw there that the situation was a bit different, more diversity in terms of regulation and trade policy, infrastructure needs. So there’s basically not one size fits all. But we have seen regional banks playing a very important role in helping countries that want to go in the regional way. I know ASEAN has done important work in AI policy, for example, and other regions are also working in that sense. And I do think that that brings economies of scale to a certain extent. It helps you resolve questions on electricity sometimes.

And so I think there’s a lot of opportunity and further work to be done at the regional level.

Arndt Husar

Thank you. And I think with the regional cooperation integration agenda being also top of mind for ADB, I just ask my colleague also to… tell me a little bit about her perspective. Of course, she represents ADB in a very large economy in South Asia, but we do have regional cooperation happening around the region. Mio, what do you see as opportunities with regards to regional cooperation integration on this digital infrastructure space?

Mio Oka

Right, thank you. So again, ADB, we support India. I’m in India. Our office covers India. So we are here to support Vixie Bharat so that India can grow at the pace to become a developed country by 2047, and the AI is a necessary means to do that. But again, as everybody knows, we are the regional bank. Nobody around us should be left alone or left behind. So ADB, through this kind of forum, has to be a catalyst. A catalyst for the global south. So we are here. Of course, there are many countries who cannot invest in scale. What are the solutions? So we are here to support the solutions, and also we support big tigers like India to support those countries too.

That’s number one. And number two is the balance approach. When we talk about regional cooperation or the work in a small country, I was quite shocked about five years ago. I went to the small neighbor country here, and I was working in the agriculture sector, and I was proudly introducing, I want to introduce aquaculture using the AI -based fish feeding system. And my negotiation ended in three seconds because the government said, no, we are interested in employment. What are you talking about? What AI -based feeders will just reduce the people who are going to work there? So that is a big lesson learned for me. We need an ecosystem, but even we talk about AI, the solution may be elsewhere.

so as you introduced the skill is super important and since that understanding again going back to India we’ve invested more than like 5 billion in the skill including the PM set and working over 10 states and now AI based skill is the big part of it so we are always mindful that the regional cooperation and we should not forget should not leave any country to be left behind but solution again may not be as direct as we expect thank you

Arndt Husar

thank you Mio and we have one minute left on the clock that throws a spanner into my closing with the thought that AI may not be the solution for everything but I think it’s a fair ending looking for a name We need to understand the problems and see how AI, if it can be deployed, if it can make a difference, how it should be supported through skills development, infrastructure investments, regulation. So I want to thank my panel for a very interesting tour de force of this topic. Also thought I’d take the opportunity to thank the audience and India for hosting this amazing summit. As ADB, we’ve been proud to be a partner of it, and it’s been truly fascinating, and we’re quite proud to have been part of this journey.

Thank you all for attending, and thanks to the panel. Let’s give them a round of applause for sharing their views. Thank you. Thank you. Thank you very much. Thank you. Thank you.

Dr. Saurabh Garg

Speech speed

143 words per minute

Speech length

569 words

Speech time

238 seconds

AI‑Ready Data Framework

Explanation

Dr. Garg outlines a comprehensive AI‑ready data framework that includes discoverability, trustworthiness, interoperability and usability. He stresses that metadata standards, quality assessment and unique identifiers are essential for data to be credible and interoperable across systems, forming the bedrock for AI deployment.

Evidence

“element of having a metadata structure which is understandable and well defined and can be used across the second on the trustworthy part would be the quality assessment so we’ve developed as kind of a quality assessment framework which focuses on the quality of the data so that to ensure credibility on the data interoperability a lock would depend on whether data can talk to each other what is the unique identifiers that we have which will ensure that the different data sets whether are they talking about the same thing or different we are able to identify that and the fourth would be on the usability across systems would be based on the standards and classifications that we have whether it’s a common definitions and common standards so that two sets of data don’t refer to the same thing and I suppose this really forms the bedrock of making a data AI ready and that That’s something that we’re working with ministries and governments and state governments across the country” [1]. “So on discoverability” [2]. “The third would be on the interoperability, and the fourth would be on the usability across systems” [3]. “On discoverability, the metadata of that structure is extremely important, so that would help to, that’s a first” [4]. “Second is how to ensure that the data sets are trustworthy, and that would be the second element” [5]. “The other aspect on data is also on its dissemination and access, on how we are able to ensure that data sets in themselves have value beyond AI and what kind of dissemination and mechanisms can be there which will make it usable for people to leverage them for business while preserving the privacy aspects of individual data” [9].

Major discussion point

AI‑Ready Data and Efficient AI Infrastructure

Topics

Data governance | Artificial intelligence | Information and communication technologies for development

AI Model Resource‑Intensiveness Concern

Explanation

Dr. Garg notes that current AI models are heavily dependent on data and compute infrastructure, implying high resource and energy demands. This raises concerns about the sustainability of AI deployments and the need for less resource‑intensive approaches.

Evidence

“has also come up that the existing models seem to be extremely data infrastructure, infrastructure heavy, whether compute infrastructure, data infrastructure” [25].

Major discussion point

AI‑Ready Data and Efficient AI Infrastructure

Topics

Environmental impacts | Artificial intelligence

Arndt Husar

Speech speed

130 words per minute

Speech length

1910 words

Speech time

875 seconds

3 S Framework for Digital Infrastructure

Explanation

Arndt Husar emphasizes that digital infrastructure must be addressed through three inter‑linked pillars – Solutions, Standards and Skills – to ensure a holistic approach. He links these pillars to the need for data centers, compute capacity and coordinated policy actions.

Evidence

“So these three S were introduced yesterday by ITU’s head, the three S of solutions, standards, and skills” [19]. “So when you hear digital infrastructure, you might be first thinking of the data centers and the compute” [21]. “The discussion that we have planned will cover various different aspects of digital infrastructure” [17].

Major discussion point

Digital Infrastructure Gaps and the “3 S” Framework

Topics

Capacity development | The enabling environment for digital development | Information and communication technologies for development

Johanna Hill

Speech speed

161 words per minute

Speech length

861 words

Speech time

319 seconds

AI as Driver of Trade Growth

Explanation

Johanna Hill argues that AI‑related products and services are becoming increasingly tradable, creating significant economic opportunities. Realising these gains requires robust digital infrastructure, skilled labour and policy readiness.

Evidence

“You know, we’ve heard throughout this conference and before the important opportunities and applications in different sectors, in agriculture, health care, new services being developed as we speak, new services and goods that are becoming more AI -related, more tradable, and we are also seeing that that can have important opportunities” [7]. “For that to happen, for those opportunities to really be realized, one element that is really important is the digital infrastructure, the skills that you mentioned, and policy readiness” [20]. “We’ve talked about the divide in the digital divide and how do we overcome that and the role of infrastructure and skills and the rest” [16].

Major discussion point

AI as a Driver of Trade Growth in Developing Economies

Topics

The digital economy | Artificial intelligence | Information and communication technologies for development

Zuhriddin Shadmanov

Speech speed

118 words per minute

Speech length

902 words

Speech time

457 seconds

Uzbekistan AI Strategy – Skills and Infrastructure

Explanation

Shadmanov highlights Uzbekistan’s gaps in compute access and AI talent, and outlines government actions to upskill public servants and build AI‑ready infrastructure such as data centres and data lakes.

Evidence

“Mainly, first one is upskilling public servants to help them to adopt AI adoption and also creating the necessary training programs for the specialists and also creating the AI infrastructure like data centers, data lakes” [8]. “I think the first one is unequal access to compute capacities and I think advanced AI digital skills” [26].

Major discussion point

Uzbekistan’s AI Development Strategy: Skills, Infrastructure, Funding, and Private Capital

Topics

Capacity development | Financial mechanisms | Artificial intelligence

Private Capital and Funding for AI Startups

Explanation

He notes that Uzbekistan is mobilising private and foreign investment, establishing venture funds and allocating USD 50 million to AI startups to foster an ecosystem that serves both public and private sectors.

Evidence

“startup ecosystem and we established many venture funds funds of funds and also there are many emerging private funds so they are now trying to invest in startups attracting private funds, private investments so currently we have allocated around 50 million USD for AI startups so they are already providing services both for public and for businesses so trying to balance and attract all the stakeholders of the ecosystem” [40].

Major discussion point

Uzbekistan’s AI Development Strategy: Skills, Infrastructure, Funding, and Private Capital

Topics

Financial mechanisms | Artificial intelligence

Hamam Riza

Speech speed

92 words per minute

Speech length

1186 words

Speech time

771 seconds

Indonesia AI Roadmap – Triple Deficit

Explanation

Riza describes Indonesia’s “triple deficit” of data, compute and talent, and explains that the national AI roadmap (which she co‑chaired) targets these gaps by 2030. The roadmap also aims to develop a digital‑consumer base to drive AI transformation.

Evidence

“We are basically triple deficit in terms of what we are going to it is the most challenging one the first one is that certainly about the data and infrastructures, the compute infrastructures we are still lacking the connectivity the networks but as I have marked down here in order for us to smooth up all the AI use cases for public services for health services, for agriculture, and many other things, you need to basically solve this triple deficit” [29]. “So our government is addressing these gaps through the national roadmap that I co -chaired” [28]. “And we are going to prepare ourselves in this AI transformation so that our data… digital consumer… is going to be part of our transformation” [10].

Major discussion point

Indonesia’s AI Roadmap and Ecosystem Building

Topics

Capacity development | Information and communication technologies for development | Artificial intelligence

Infrastructure Transformation and Accessibility

Explanation

Riza stresses that new AI‑related infrastructure is being built, with an emphasis on making technology accessible to all and leveraging global hyperscalers to create multiple cloud regions in Indonesia.

Evidence

“The technology is accessible for all” [14]. “yes so I see these questions and I’m really eager to answer this because suddenly our infrastructure is undergoing a transformation really to meet the demands of the AI demand and certainly with the ability of many of these new infrastructures coming out of the government and also from the business I think benchmarking with many other countries including you know in the regional ASEAN take for example the presence of the global hyperscaler in the country have established actually multiple cloud regions in Indonesia” [38].

Major discussion point

Indonesia’s AI Roadmap and Ecosystem Building

Topics

Closing all digital divides | Information and communication technologies for development

Mio Oka

Speech speed

143 words per minute

Speech length

664 words

Speech time

278 seconds

ADB Role – Foundational Infrastructure and Service‑Level AI

Explanation

Mio explains that ADB views foundational infrastructure (stable power, broadband) as essential and supports AI applications in agriculture, water and irrigation to deliver tangible services.

Evidence

“From ADB’s perspective, of course, foundation is important” [35]. “So we work on agriculture sector, water supply, and even irrigation sector where AI is widely applied” [30].

Major discussion point

ADB’s Role in Foundational Infrastructure, Service‑Level AI Applications, and Regional Cooperation

Topics

Financial mechanisms | Artificial intelligence | Social and economic development

ADB as Catalyst for Regional Cooperation

Explanation

She highlights ADB’s function as a catalyst that mobilises private capital, supports master planning and promotes regional integration for AI‑related digital infrastructure.

Evidence

“So ADB, through this kind of forum, has to be a catalyst” [31]. “And this is where ADB would like to support” [34]. “ADB is actually supporting that” [32].

Major discussion point

ADB’s Role in Foundational Infrastructure, Service‑Level AI Applications, and Regional Cooperation

Topics

The enabling environment for digital development | Capacity development | Artificial intelligence

Agreements

Agreement points

Critical importance of skills development and capacity building for AI adoption

Speakers

– Zuhriddin Shadmanov
– Hamam Riza
– Mio Oka
– Arndt Husar

Arguments

Comprehensive approach covering all strata from students to professionals and public servants, not just tech sector

National AI roadmap targeting 12 million AI talents by 2030, establishing Korika Academy for upskilling and reskilling programs

ADB invested over $5 billion in skills development including AI-based skills as part of regional cooperation

SMEs face a huge adoption gap in understanding how to integrate AI into their business models despite technology advancing rapidly

Summary

All speakers emphasized that skills development is fundamental to AI success, requiring comprehensive programs that go beyond just technical sectors to include all levels of society and business

Topics

Capacity development | Artificial intelligence | Closing all digital divides

Need for comprehensive infrastructure development beyond just compute power

Speakers

– Dr. Saurabh Garg
– Zuhriddin Shadmanov
– Hamam Riza
– Arndt Husar

Arguments

Four key elements for AI-ready data: discoverability through metadata structure, trustworthiness through quality assessment, interoperability through unique identifiers, and usability through common standards and classifications

Government allocating $300 million USD for AI development, focusing on human-centered AI with projects in healthcare, education, transportation, and cybersecurity

Triple deficit exists: data and compute infrastructure gaps, connectivity issues, and shortage of AI skilled talents limiting long-term innovation capabilities

Digital infrastructure encompasses not just data centers and compute, but also solutions, standards, and skills – the three S’s introduced by ITU

Summary

Speakers agreed that AI infrastructure requires a holistic approach including data governance, standards, connectivity, and human capacity, not just computing power

Topics

The enabling environment for digital development | Artificial intelligence | Information and communication technologies for development

Importance of regional cooperation and collaboration for AI development

Speakers

– Johanna Hill
– Mio Oka
– Arndt Husar

Arguments

Regional approaches important for sharing infrastructure and collaboration when lacking economies of scale for huge investments

ADB serves as catalyst for global south, supporting both large economies like India and ensuring smaller countries aren’t left behind

Regional cooperation and infrastructure sharing is still new territory with initiatives like the working group on democratization of AI compute being supported by ADB

Summary

All speakers recognized that regional cooperation is essential for countries that cannot achieve economies of scale individually, particularly for infrastructure sharing and ensuring no country is left behind

Topics

Financial mechanisms | The enabling environment for digital development | Artificial intelligence

Need for balanced approach considering different country contexts and capabilities

Speakers

– Johanna Hill
– Mio Oka
– Arndt Husar

Arguments

AI Trade Policy Openness Index developed to help regions measure performance, showing lower income economies can appear open but may lack necessary regulation

Foundation infrastructure important but service-level AI applications in agriculture, water supply, and irrigation offer more immediate impact given the scale

Countries need to balance different types of compute infrastructure based on actual needs rather than pursuing high-end solutions universally

Summary

Speakers agreed that AI strategies must be tailored to different country contexts, capabilities, and actual needs rather than applying one-size-fits-all approaches

Topics

The enabling environment for digital development | Artificial intelligence | Social and economic development

Data governance | The enabling environment for digital development | Human rights and the ethical dimensions of the information society

Unexpected consensus

Arguments

Current AI models are extremely infrastructure-heavy, requiring gigawatts of power compared to human brain’s 100 watts, suggesting need for alternative mechanisms

Foundation infrastructure important but service-level AI applications in agriculture, water supply, and irrigation offer more immediate impact given the scale

Summary

Dr. Garg questions the efficiency of current AI infrastructure and suggests exploring alternative mechanisms due to massive power consumption, while Mio Oka argues that despite foundational infrastructure being important, the focus should be on service-level applications for immediate impact

Topics

Artificial intelligence | Environmental impacts | The enabling environment for digital development

Scale and scope of AI development strategies

Speakers

– Zuhriddin Shadmanov
– Hamam Riza

Arguments

Comprehensive approach covering all strata from students to professionals and public servants, not just tech sector

Triple deficit exists: data and compute infrastructure gaps, connectivity issues, and shortage of AI skilled talents limiting long-term innovation capabilities

Summary

Shadmanov emphasizes a broad societal approach covering all sectors beyond technology, while Riza focuses on addressing specific technical deficits in infrastructure and skills that limit innovation capabilities

Topics

Capacity development | Artificial intelligence | Closing all digital divides

Unexpected differences

Role of employment considerations in AI adoption

Speakers

– Mio Oka
– Other panelists

Arguments

ADB serves as catalyst for global south, supporting both large economies like India and ensuring smaller countries aren’t left behind

Various arguments about AI development and adoption

Explanation

Mio Oka’s anecdote about AI-based aquaculture being rejected due to employment concerns reveals an unexpected disagreement about whether AI adoption should prioritize efficiency or employment preservation, which other speakers didn’t directly address in their technology-focused approaches

Topics

Social and economic development | The digital economy | Artificial intelligence

Overall assessment

Summary

The main areas of disagreement center around infrastructure investment priorities (efficiency vs. immediate impact), the scope of AI development strategies (broad societal vs. technical focus), and approaches to international cooperation (regional frameworks vs. bilateral partnerships vs. sovereign development)

Disagreement level

Moderate level of disagreement with significant implications – while speakers share common goals of AI development for the Global South, their different approaches could lead to fragmented strategies and inefficient resource allocation if not coordinated properly

Partial agreements

All speakers agree on the importance of international partnerships and regional cooperation for AI development, but they disagree on the specific mechanisms – Hill emphasizes regional trade policy frameworks, Shadmanov focuses on bilateral partnerships with specific countries, and Riza emphasizes sovereign AI with special economic zones

Speakers

– Johanna Hill
– Zuhriddin Shadmanov
– Hamam Riza

Arguments

Regional approaches important for sharing infrastructure and collaboration when lacking economies of scale for huge investments

Planning to attract $1 billion USD by 2030 for AI-related digital infrastructure, working with Chinese partners like Huawei for ecosystem development

Hyperscalers establishing multiple cloud regions in Indonesia, with government preparing presidential regulation for sovereign AI and special economic zones

Topics

Financial mechanisms | The enabling environment for digital development | Artificial intelligence

Both agree on the need for massive government investment in AI skills development, but disagree on approach – Shadmanov focuses on human-centered AI across government sectors, while Riza emphasizes ambitious talent targets through dedicated academy structures

Speakers

– Zuhriddin Shadmanov
– Hamam Riza

Arguments

Government allocating $300 million USD for AI development, focusing on human-centered AI with projects in healthcare, education, transportation, and cybersecurity

National AI roadmap targeting 12 million AI talents by 2030, establishing Korika Academy for upskilling and reskilling programs

Topics

Capacity development | Artificial intelligence | Social and economic development

Both agree on the importance of balancing data accessibility with protection, but disagree on focus – Dr. Garg emphasizes technical mechanisms for data access while preserving privacy, while Hill focuses on regulatory frameworks and trade policy measures

Speakers

– Dr. Saurabh Garg
– Johanna Hill

Arguments

Data dissemination and access mechanisms needed while preserving privacy aspects of individual data

AI Trade Policy Openness Index developed to help regions measure performance, showing lower income economies can appear open but may lack necessary regulation

Topics

Data governance | Human rights and the ethical dimensions of the information society | The digital economy

Data governance | The enabling environment for digital development | Human rights and the ethical dimensions of the information society

Takeaways

Key takeaways

AI infrastructure development requires a balanced approach across four key pillars: AI-ready data (with proper metadata, quality assessment, interoperability, and standards), compute infrastructure, skills development, and policy frameworks

Global South countries face a ‘triple deficit’ in data/compute infrastructure, connectivity, and AI skilled talent, risking becoming AI consumers rather than creators without strategic intervention

AI and trade integration could grow global trade by 40% by 2040, but success depends on digital infrastructure readiness, skills development, and appropriate policy frameworks

Current AI models are extremely energy-intensive (requiring gigawatts vs. human brain’s 100 watts), suggesting need for more efficient alternative approaches

Regional cooperation and collaboration are essential for smaller economies to share infrastructure costs and avoid being left behind, though approaches must be context-specific

Private sector collaboration and mixed financing strategies (combining government investment, international partnerships, and private capital) are crucial for scaling AI infrastructure

Skills development must be comprehensive, covering all sectors and skill levels rather than focusing only on technical specialists

Service-level AI applications in sectors like agriculture, healthcare, and water supply may offer more immediate impact than foundational infrastructure investments for developing countries

Resolutions and action items

Uzbekistan committed to allocating $300 million for AI development and targeting $1 billion in AI infrastructure investment by 2030

Indonesia established target of training 12 million AI talents by 2030 through the Korika Academy platform

Uzbekistan launched partnership with UAE’s ‘5 million AI leaders’ program with over 1 million people already registered

Indonesia preparing presidential regulation for sovereign AI development and special economic zones for hyperscalers

ADB committed to continuing support for regional cooperation as catalyst for Global South AI development

WTO Secretariat developed AI Trade Policy Openness Index to help regions measure their performance in AI trade readiness

Unresolved issues

How to effectively balance competing priorities of infrastructure, skills, and policy development with limited financial resources

Specific mechanisms for cross-border collaboration on shared AI infrastructure and compute resources remain underdeveloped

The challenge of ensuring AI adoption doesn’t reduce employment opportunities, particularly in agriculture and traditional sectors

How to develop more energy-efficient AI models that don’t require massive compute and power resources

Standardization and interoperability challenges across different national AI strategies and regional approaches

How smaller economies without scale advantages can meaningfully participate in AI value creation rather than just consumption

Balancing data sovereignty concerns with the need for cross-border data sharing and collaboration

Suggested compromises

Mixed financing strategies combining government investment, international partnerships, and private sector capital to address resource constraints

Regional cooperation approaches that respect national sovereignty while enabling shared infrastructure and standards

Phased implementation focusing on service-level applications while building foundational infrastructure over time

Pentahelix collaboration model (government, industry, academia, civil society, media) to ensure inclusive stakeholder participation

This anecdote powerfully illustrates the disconnect between technological solutions and actual development priorities. It highlights how AI implementation must consider broader socio-economic contexts, particularly employment concerns in developing countries.

Impact

This story fundamentally grounded the entire discussion in reality, reminding all participants that AI solutions must align with local development priorities. It served as a cautionary tale that influenced the closing tone of the discussion, emphasizing that AI is not automatically the right solution for every problem.

Overall assessment

Explanation

This addresses the need for AI systems that reflect local cultures and contexts rather than relying solely on foreign-developed models

Disclaimer: This is not an official session record. DiploAI generates these resources from audiovisual recordings, and they are presented as-is, including potential errors. Due to logistical challenges, such as discrepancies in audio/video or transcripts, names may be misspelled. We strive for accuracy to the best of our ability.

Regional Leaders Discuss AI-Ready Digital Infrastructure

Summary

Keypoints

Major Discussion Points:

Overall Purpose:

Overall Tone:

Speakers

Foundational Infrastructure Requirements

National Strategies: Indonesia’s Comprehensive Approach

Uzbekistan’s Investment and Partnership Model

Trade and Economic Opportunities

Development Finance Perspective

Skills Development Across Society

Balancing Innovation with Employment Concerns

The Path Forward

AI‑Ready Data Framework

Explanation

Evidence

Major discussion point

Topics

AI Model Resource‑Intensiveness Concern

Explanation

Evidence

Major discussion point

Topics

3 S Framework for Digital Infrastructure

Explanation

Evidence

Major discussion point

Topics

AI as Driver of Trade Growth

Explanation

Evidence

Major discussion point

Topics

Uzbekistan AI Strategy – Skills and Infrastructure

Explanation

Evidence

Major discussion point

Topics

Private Capital and Funding for AI Startups

Explanation

Evidence

Major discussion point

Topics

Indonesia AI Roadmap – Triple Deficit

Explanation

Evidence

Major discussion point

Topics

Infrastructure Transformation and Accessibility

Explanation

Evidence

Major discussion point

Topics

ADB Role – Foundational Infrastructure and Service‑Level AI

Explanation

Evidence

Major discussion point

Topics

ADB as Catalyst for Regional Cooperation

Explanation

Evidence

Major discussion point

Topics

Agreement points

Critical importance of skills development and capacity building for AI adoption

Speakers

Arguments

Summary

Topics

Need for comprehensive infrastructure development beyond just compute power

Speakers

Arguments

Summary

Topics

Importance of regional cooperation and collaboration for AI development

Speakers

Arguments

Summary

3 S Framework for Digital Infrastructure