Sovereign AI: How Emerging Markets Are Rewriting Big Tech Rules

Somewhere in the desert outside Riyadh, construction crews are laying the foundations for what Saudi Arabia hopes will become one of the most powerful AI computing facilities on Earth. Backed by more than $100 billion in planned investment from HUMAIN, a company launched in May 2025 under the Saudi Public Investment Fund, the project envisions 11 data centres with a combined capacity of 2,200 megawatts, powered by several hundred thousand NVIDIA GPUs. Two large campuses are already under construction, with the company targeting 50 megawatts of operational capacity by the end of 2025 and adding another 50 megawatts every quarter into 2026. It is, by any measure, an audacious undertaking. But the truly radical part is not the hardware. It is the politics.

HUMAIN is not just building servers. It is building sovereignty. The company's flagship Arabic large language model, ALLAM 34B, has been independently verified by Cohere on the MMLU benchmark as the most advanced Arabic LLM built in the Arab world. HUMAIN Chat, the consumer-facing application powered by ALLAM 34B, launched in August 2025 as a national milestone before beginning its regional rollout across the Middle East. When HUMAIN CEO Tareq Amin told CNBC that the company's ambition was to become “the third-largest AI provider in the world, behind the United States and China,” he was articulating something more than corporate aspiration. He was describing a geopolitical strategy, one rooted in the Kingdom's Vision 2030 framework and the broader conviction that AI sovereignty is, as analysts at the Saudi Data and AI Authority have put it, “not isolationism, but intelligent independence.”

This is the new reality confronting Amazon, Microsoft, Google, and every other hyperscaler racing to deploy AI infrastructure across the Global South. The old playbook of building data centres, plugging them into the global cloud, and letting the algorithms flow freely is dead. In its place, a far more complicated game is emerging, one in which governments are demanding not just local servers but local intelligence, not just data residency but data sovereignty, not just computing power but the right to understand, audit, and ultimately control what those computers are doing.

Six Hundred Billion Dollars Chasing a Moving Target

The numbers are staggering. According to CreditSights, capital expenditure by the five largest hyperscalers (Amazon, Alphabet, Microsoft, Meta, and Oracle) is projected to reach approximately $602 billion in 2026, a 36 per cent increase over 2025 and a dramatic acceleration from the $256 billion spent in 2024. Roughly three quarters of that spending, around $450 billion, is directed specifically at AI infrastructure. McKinsey projects that global demand for data centre capacity will grow at a compound annual rate of 22 per cent through 2030, reaching 220 gigawatts, nearly six times larger than demand in 2020. Of that total, AI workloads are expected to account for 156 gigawatts by 2030, up from 44 gigawatts in 2025. The costs are almost incomprehensible: McKinsey estimates between $5.2 trillion and $7.9 trillion in cumulative capital expenditure will be needed to build out AI data centre capacity through the end of the decade.

Amazon alone raised its 2025 capital expenditure guidance to $125 billion, a 61 per cent increase year over year. AWS announced more than $30 billion in combined investments in Pennsylvania and North Carolina for what it calls “AI innovation campuses.” By the end of 2025, the company had added 3.8 gigawatts of capacity over the previous twelve months, more than any other hyperscaler. AWS, Microsoft Azure, and Google Cloud collectively accounted for 66 per cent of global cloud infrastructure spending.

But here is the twist that matters. Those headline investments are overwhelmingly concentrated in the United States and established European markets. The truly consequential decisions are happening elsewhere, in markets where the infrastructure is thinner, the regulatory landscape is more volatile, and the stakes for getting it wrong are considerably higher.

Synergy Research Group counted 1,297 operational hyperscale data centres worldwide as of late 2025, nearly triple the number from early 2018. Yet Africa accounts for less than one per cent of global data centre capacity, despite industry estimates indicating the continent needs at least 1,000 megawatts of new capacity across 700 additional facilities to meet demand. Latin America ranks fifth globally in hyperscale investment, with Brazil and Mexico leading. Southeast Asia, despite explosive internet growth, is still catching up. Indonesia alone has 280 million citizens, 200 million internet users, and a $1.4 trillion GDP growing at 5.1 per cent annually, yet 70 per cent of its data centre capacity remains concentrated in the Jakarta region. These areas represent the next great frontier for cloud and AI deployment. They also represent the places where the tension between global technology ambitions and local sovereignty demands is most acute.

Sovereignty Is Not a Feature Request

The concept of data sovereignty has shifted from a niche concern among privacy advocates to a central organising principle of national technology policy. In 2026, the landscape is defined by a rolling wave of legislation that is reshaping how hyperscalers operate in virtually every emerging market. Seventy-one per cent of organisations now cite cross-border data transfer compliance as their top regulatory challenge, reflecting the complexity of navigating what has become a genuinely fragmented global framework.

India's Digital Personal Data Protection Act entered its enforcement phase following the release of operational rules in November 2025. Organisations must now implement mandatory encryption, masking, tokenisation, and access controls, with breach notifications required within 72 hours and penalties reaching 250 crore rupees (approximately $30 million). But India's ambitions extend far beyond data protection. The IndiaAI Mission, launched in March 2024 with an initial budget of 10,300 crore rupees ($1.24 billion), has blown past its original target of 10,000 GPUs. Abhishek Singh, Additional Secretary at India's Ministry of Electronics and Information Technology, revealed at the Accel AI Summit 2025 that the first two rounds of GPU tenders alone secured commitments for 34,000 units. A third tender added approximately 3,850 more, including for the first time 1,050 Google Trillium TPUs, pushing total capacity past 38,000 units available at a subsidised rate of just 65 rupees per hour. Sarvam AI was selected in April 2025 to build India's sovereign LLM ecosystem, developing an open-source 120 billion parameter model. BharatGen AI, India's first government-funded multimodal large language model, now supports 22 Indian languages. In the private sector, Reliance Industries is constructing a one-gigawatt data centre in Gujarat powered by NVIDIA Blackwell processors, estimated at $20 to $30 billion, while AWS has committed $12.7 billion to India through 2030 and Microsoft has pledged $3 billion for 2025 and 2026.

Brazil is pursuing an equally ambitious path. The Brazilian Artificial Intelligence Plan 2024 to 2028, titled “AI for the Good of All,” allocates approximately 23 billion reais to develop national computing infrastructure, sovereign cloud capabilities, and Portuguese-language foundation models. Of that, 5.7 billion reais is earmarked for a sovereign cloud operated by state firms Serpro and Dataprev, hosting a supercomputer for training Portuguese-language AI models. Another 2.3 billion reais goes toward domestic data centres powered by renewable energy, with priority given to the North and Northeast regions. The SoberanIA initiative, a collaboration between the government of Piauí, the Ministry of Science, Technology and Innovation, Telebras, Modular, and Scala Data Centers, has built the world's largest Portuguese-language database for commercial AI use, expanding from 130 billion to 350 billion tokens. The initiative's infrastructure includes a Piauí AI Factory for model training and a Data Vault in Brasilia operated by Telebras in a Tier IV data centre within a military area, functioning as the sovereign repository for the state's strategic databases. The technology already powers applications like “Piauí Oportunidades,” which delivers personalised learning paths, and “BO Fácil,” which allows incident reports to be filed by voice through WhatsApp.

Indonesia is carving its own regulatory path. Government Regulation No. 71 of 2019 requires data for public services to be processed and stored domestically, reinforcing the need for local data centres. The country's National AI Strategy 2020 to 2045 positions artificial intelligence as a primary driver of the “Golden Indonesia 2045” development goal, with priority sectors in healthcare, bureaucratic reform, education, food security, and smart cities. The government is implementing a risk-based classification system similar to the EU AI Act. BDx Indonesia launched the country's first sovereign AI data centre in December 2024, powered by NVIDIA accelerated computing, and a $15 billion foreign direct investment pipeline is being attracted through regulatory reforms and tax holidays of up to 20 years for strategic projects.

Vietnam is writing its own chapter with its first statutory personal data protection law (Law No. 91/2025/QH15) and a national AI law taking effect on 1 March 2026. The UAE has established expectations under its Personal Data Protection Law, including consent requirements, transparency mandates, and cross-border transfer controls. Across Africa, 35 data protection authorities had become operational by 2024, though 15 jurisdictions still lack established regulatory frameworks despite having enacted comprehensive privacy legislation.

The pattern is unmistakable. Governments are no longer content to be passive consumers of technology produced elsewhere. They want to shape how AI is built, deployed, and governed within their borders.

Building the Sovereign Cloud

For hyperscalers, the response has been to invest heavily in what the industry now calls “sovereign cloud” infrastructure, physically and logically separated computing environments designed to keep data within national borders while still offering the scale and capability of global cloud platforms.

On 15 January 2026, AWS officially launched the European Sovereign Cloud, a completely independent infrastructure isolated from other AWS regions worldwide. Built with a committed investment of 7.8 billion euros, the first region went live in Germany. The offering features dedicated compute regions, a separate identity and access management system, and an independent billing system operating entirely within the European Union. The accompanying Sovereign Reference Framework describes how AWS implements and validates sovereignty controls, with each criterion treated as binding and subject to independent third-party auditing throughout 2026. AWS also introduced the Digital Sovereignty Well-Architected Lens, a framework designed to help organisations build workloads that are sovereign, compliance-aligned, and auditable while remaining interoperable and portable.

But the European Sovereign Cloud is just the most visible manifestation of a broader trend. AWS is investing $5.3 billion to build a new infrastructure region in Saudi Arabia, scheduled to launch in 2026 with three Availability Zones. The company has committed $12.7 billion to India through 2030, launched its Asia Pacific Malaysia region in August 2024 with a commitment of $6.2 billion, and is building a Chile region with more than $4 billion committed for expected completion by the end of 2026.

Microsoft is on a parallel track. The company invested $2.2 billion in Malaysia for cloud and AI infrastructure, pledged $300 million for AI infrastructure in South Africa with an aim to provide AI skills to one million South Africans by 2026, and is building a $1 billion geothermal-powered data centre in Kenya alongside a partnership with Abu Dhabi's G42 for a broader digital ecosystem investment. In Saudi Arabia, Microsoft completed three data centres and Azure Availability Zones in the Eastern Province as part of a broader $2.1 billion investment. In Nigeria, Microsoft is deploying in Lagos to provide low-latency services to fintech and oil and gas enterprises while aligning with the country's data localisation requirements.

Google has launched its Johannesburg cloud region, estimated at 2.5 billion rand in investment and part of a broader $1 billion digital initiative across Africa. Google's President for Europe, the Middle East, and Africa, Tara Brady, highlighted the potential to create 300,000 jobs and contribute 1.7 trillion rand to the South African economy. The company expanded to 42 cloud regions with 127 Availability Zones by 2025. Google's connectivity investments include the Equiano submarine cable along Africa's western seaboard and Umoja, the first fibre optic route to directly connect Africa with Australia. The company aims to reach 500 million Africans with AI-powered innovations by 2030.

The sovereign cloud market itself is projected to grow from $154 billion in 2025 to $823 billion by 2032. Nearly half of technology buyers surveyed expect their use of sovereign cloud for AI workloads to increase over the next two years.

When Explainability Becomes Law

Building data centres within national borders is one thing. Making AI systems transparent and explainable to local regulators is something else entirely, and it may prove to be the harder problem.

The EU AI Act, which became fully applicable on 2 August 2026, established the world's first comprehensive AI transparency framework. The Act defines transparency as meaning “that AI systems are developed and used in a way that allows appropriate traceability and explainability, while making humans aware that they communicate or interact with an AI system, as well as duly informing deployers of the capabilities and limitations of that AI system and affected persons about their rights.” High-risk AI systems must be designed to be “sufficiently transparent to enable deployers to interpret a system's output and use it appropriately.” Non-compliance carries penalties of up to 35 million euros or seven per cent of global annual turnover.

But the EU is far from alone. Australia's Privacy Act reforms, introduced in 2025, established mandatory transparency requirements for automated decision-making systems, requiring organisations to disclose algorithmic logic, data sources, and decision criteria to affected individuals. India's regulatory framework increasingly demands that AI systems used in governance be auditable and explainable. Brazil's pending AI legislation (PL 2338/2023), approved by the Federal Senate in December 2024 and forwarded to the Chamber of Deputies in March 2025, would mandate impact assessments for high-risk AI systems and proposes the creation of a national authority to oversee AI governance.

For hyperscalers, meeting these requirements means investing in explainability infrastructure adaptable to local regulatory contexts. Amazon's approach centres on two complementary platforms. SageMaker Clarify provides model-agnostic feature attribution using techniques such as SHAP to provide per-instance explanations during inference, and includes fairness and bias detection tools. Amazon Bedrock offers Chain-of-Thought reasoning traces through Bedrock Agents, showing step-by-step logic behind each decision. Bedrock Guardrails provides configurable safeguards including model cards with detailed bias metrics and built-in evaluation datasets like BOLD, which assesses fairness across categories including profession, gender, and race. The platform also features Bedrock Data Automation, which offers visual grounding with confidence scores for explainability and built-in hallucination mitigation. All of these capabilities integrate with monitoring and logging systems in scope for compliance standards including ISO, SOC, CSA STAR Level 2, and GDPR.

These tools are not decorative. They are increasingly becoming prerequisites for operating in regulated markets. The challenge is that different jurisdictions define “explainability” differently. What satisfies European regulators may not meet Indian requirements for citizen-facing AI systems, and neither may align with Saudi Arabia's vision for sovereign AI that operates within the cultural and linguistic frameworks of the Arabic-speaking world. Model cards, which are rapidly becoming a cornerstone of responsible AI strategy, illustrate this tension. Originally designed as static documentation, they are evolving into dynamic tools integrated directly into the AI lifecycle. But their content, structure, and level of detail must be adapted to each jurisdiction's expectations.

The result is a growing demand for what might be called “regulatory localisation” of AI systems. It is not enough to build a model that works. You need to build a model that can explain itself in ways that satisfy the specific legal, cultural, and institutional expectations of each market where it operates.

Language, Culture, and the Limits of One-Size-Fits-All

The sovereignty movement is forcing hyperscalers to confront a deeper truth about AI: that language and culture are not merely surface-level features to be localised through translation. They are structural elements that shape how AI systems understand and represent the world.

Saudi Arabia's ALLAM 34B is a case in point. Built specifically for Arabic, it represents a fundamentally different approach from taking an English-language model and bolting on Arabic capabilities. SDAIA, the Saudi Data and AI Authority, established in 2019, is working with NVIDIA to deploy up to 5,000 Blackwell GPUs for a sovereign AI factory, while also training government and university scientists on developing models for physical and agentic AI. NVIDIA CEO Jensen Huang, speaking at the partnership announcement, framed the stakes plainly: “AI, like electricity and internet, is essential infrastructure for every nation.” The ambition is to create AI systems that understand Arabic not as a foreign language but as a native one, with all the cultural nuance, historical context, and institutional knowledge that entails.

India's BharatGen AI, supporting 22 languages, and Brazil's SoberanIA project, built on a 350 billion token Portuguese-language database with assured governance for commercial use, represent similar philosophical commitments. These are not merely technical projects. They are assertions of cultural and linguistic independence in a domain overwhelmingly dominated by English-language systems developed in Silicon Valley.

For hyperscalers, this creates a strategic dilemma. Their business model depends on scale, on building standardised platforms that can serve customers anywhere with minimal customisation. But the sovereignty movement pushes in the opposite direction, toward fragmentation, localisation, and the proliferation of distinct regulatory and cultural requirements that cannot be satisfied by a single global architecture.

The most sophisticated response so far has been to create layered platforms combining global infrastructure with local adaptation. AWS's approach through Bedrock allows customers to fine-tune foundation models on proprietary data within sovereign cloud environments, meeting both performance and compliance requirements. Microsoft's commitment to processing Microsoft 365 Copilot interactions in-country for 15 nations by the end of 2026, with Australia, India, Japan, and the United Kingdom gaining in-country AI processing in 2025, represents a similar acknowledgement that proximity matters, not just for latency but for trust.

But layered platforms only solve part of the problem. The harder question is whether hyperscalers can genuinely accommodate governance models that may conflict with their commercial interests.

How Localised Rules Are Reshaping Global Standards

Here is where the story gets genuinely interesting. The conventional wisdom assumes that the proliferation of local AI regulations will create a fragmented, burdensome compliance landscape that benefits nobody. But there is a plausible alternative scenario in which local experimentation eventually drives convergence toward higher global standards.

The EU AI Act is already serving as a template. Brazil's pending legislation explicitly adopts its risk-based architecture. Vietnam's new AI law draws on similar principles. Indonesia's evolving regulatory framework distinguishes between prohibited practices, high-risk applications, and limited-risk systems in a manner that closely mirrors the European model. The pattern resembles what happened with data protection after the GDPR: the European regulation became the de facto global standard, not because every country copied it exactly, but because the cost of building separate systems for each jurisdiction incentivised companies to adopt the highest common standard everywhere.

The UN's Global Digital Compact, adopted by 193 member states in September 2024, and the Global Dialogue on AI Governance, established by the UN General Assembly in August 2025 through Resolution A/RES/79/325, are creating multilateral forums for exactly this kind of convergence. The first annual gathering is planned for the 2026 AI for Good Global Summit in Geneva. UN Secretary-General Antonio Guterres outlined clear goals: building safe, secure, and trustworthy AI systems grounded in international law; promoting interoperability between governance regimes; and encouraging open innovation accessible to all. An Independent International Scientific Panel on AI, the first global scientific body of its kind, has been established to assess how AI is transforming societies.

The Global Cross-Border Privacy Rules Forum, which launched its certification systems on 2 June 2025, represents another vector of convergence. With nine member jurisdictions (including Australia, Canada, Japan, Singapore, and the United States) and four associate members spanning six continents, the Forum establishes harmonised requirements for sensitive data processing, children's protection protocols, and standardised breach notification timelines. Nigeria recently joined as an associate member, while the Dubai International Financial Centre became a full member, signalling that emerging markets are actively shaping these international frameworks. The Global Cooperation Arrangement for Privacy Enforcement, established in October 2023, facilitates cross-border enforcement actions between participating authorities, including the UK Information Commissioner's Office and the US Federal Trade Commission.

The question is whether hyperscalers will resist this convergence or embrace it as a means of reducing compliance complexity. The early evidence is mixed. The United States came out in strong opposition to multilateral AI governance initiatives at a UN Security Council debate in September 2025. Yet companies like Amazon, Microsoft, and Google continue to invest billions in sovereign infrastructure that implicitly accepts the legitimacy of local regulatory authority.

The Monetisation Gap and Its Political Consequences

There is one more dimension to this story that deserves attention: money. AI-related services are expected to deliver only about $25 billion in revenue in 2025, roughly ten per cent of what hyperscalers are spending on infrastructure. Only about 25 per cent of AI initiatives have delivered their expected return on investment. Morgan Stanley and JP Morgan project that the technology sector may need to issue $1.5 trillion in new debt over the next few years to finance AI infrastructure construction. The disconnect is stark: hyperscalers are transforming from historically cash-funded business models into leveraged ones, betting that AI adoption will eventually catch up to the enormous capital outlays.

This gap between investment and return creates political vulnerability, particularly in emerging markets. Governments that have welcomed hyperscaler investment with tax incentives and expedited permitting will eventually demand results. Indonesia's experience offers a cautionary tale: research has demonstrated that the country's increasing data restrictiveness between 2013 and 2018 led to measurable economic harm, reducing trade output by 9.1 per cent, decreasing productivity by 3.7 per cent, and increasing downstream prices by 1.9 per cent over five years. If AI infrastructure does not deliver measurable economic benefits to local populations, the backlash could be severe.

The smartest hyperscalers understand this. Microsoft's commitment to providing AI skills to one million South Africans by 2026, Google's target of reaching 500 million Africans with AI-powered innovations by 2030, and India's programme to train 115,000 civil servants in AI fundamentals are all attempts to ensure that the social licence to operate keeps pace with the rate of capital deployment.

But skills training alone will not be sufficient. The deeper challenge is ensuring that sovereign AI systems actually work, that they deliver tangible improvements in healthcare, education, agriculture, and governance that justify the enormous investments being made. Brazil's SoberanIA already demonstrates what this can look like in practice, with applications ranging from personalised learning to voice-enabled government services now available to public managers from any region. India's BharatGen AI is being deployed across multiple sectors with explicit goals around inclusive growth. Saudi Arabia's HUMAIN is targeting energy, healthcare, manufacturing, and financial services. Africa's AI market is expected to grow from $4.51 billion in 2025 to $16.53 billion by 2030, at a compound annual growth rate of 27.42 per cent, with Cassava Technologies rolling out NVIDIA-powered AI factories starting in South Africa and expanding to Egypt, Kenya, Morocco, and Nigeria.

Intelligent Independence as a Design Principle

The phrase that keeps surfacing in policy documents and executive presentations across the Global South is “intelligent independence.” It captures something important about this moment. The countries building sovereign AI infrastructure are not trying to cut themselves off from the global technology ecosystem. They are trying to participate in it on their own terms. As one analysis of Brazil's approach noted, the pursuit of digital sovereignty “should not be confused with the notion of technological self-sufficiency, which remains unrealistic in the short and medium term.” What is emerging instead is an agenda of relative autonomy, in which countries seek to reduce critical vulnerabilities without isolating themselves from global innovation networks.

For hyperscalers, this means that the era of building technology and exporting it wholesale is giving way to something more collaborative and more constrained. The companies that thrive will be the ones that can operate as genuine partners, contributing infrastructure, expertise, and capital while ceding meaningful control over governance, transparency, and cultural adaptation to the countries where they operate.

This is not a comfortable position for companies accustomed to setting the rules of the game. But it may be the only viable strategy in a world where sovereign cloud spending is projected to quintuple in seven years, where the next billion internet users will predominantly come from Asia, Africa, the Middle East, and Latin America, and where every one of those users will be governed by data protection and AI transparency laws that did not exist five years ago.

The infrastructure is being built. The regulations are being written. The models are being trained in Arabic, Portuguese, Hindi, and 22 other Indian languages. The question that remains is whether the global AI ecosystem that emerges from this period of intense localisation will be more fragmented or more resilient than the one it replaces. The answer will depend less on the technology itself and more on whether the companies building it can learn to operate in a world where sovereignty is not an obstacle to be overcome but a feature to be designed for.

References and Sources

  1. IEEE ComSoc Technology Blog, “Hyperscaler capex > $600 bn in 2026,” December 2025. https://techblog.comsoc.org/2025/12/22/hyperscaler-capex-600-bn-in-2026-a-36-increase-over-2025-while-global-spending-on-cloud-infrastructure-services-skyrockets/

  2. Data Center Knowledge, “AI-First Hyperscalers: 2026's Sprint Meets the Power Bottleneck,” 2026. https://www.datacenterknowledge.com/hyperscalers/hyperscalers-in-2026-what-s-next-for-the-world-s-largest-data-center-operators-

  3. Data Center Frontier, “Amazon Doubles Down on AI Infrastructure with $30B in New U.S. Data Center Investments,” 2025. https://www.datacenterfrontier.com/hyperscale/article/55295967/amazon-doubles-down-on-ai-infrastructure-with-30b-in-new-us-data-center-investments

  4. McKinsey & Company, “The cost of compute: A $7 trillion race to scale data centers,” 2025. https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/the-cost-of-compute-a-7-trillion-dollar-race-to-scale-data-centers

  5. McKinsey & Company, “AI power: Expanding data center capacity to meet growing demand,” 2025. https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/ai-power-expanding-data-center-capacity-to-meet-growing-demand

  6. Future Privacy Forum, “2026: A Year at the Crossroads for Global Data Protection and Privacy,” 2026. https://fpf.org/blog/2026-a-year-at-the-crossroads-for-global-data-protection-and-privacy/

  7. AWS, “European Digital Sovereignty,” 2026. https://aws.amazon.com/compliance/europe-digital-sovereignty/

  8. CIO Dive, “AWS European Sovereign Cloud goes live,” January 2026. https://www.ciodive.com/news/aws-european-sovereign-cloud-live/809794/

  9. AWS Security Blog, “Exploring the AWS European Sovereign Cloud: Sovereign Reference Framework,” 2025. https://aws.amazon.com/blogs/security/exploring-the-new-aws-european-sovereign-cloud-sovereign-reference-framework/

  10. EU Artificial Intelligence Act, “Article 13: Transparency and Provision of Information to Deployers,” 2024. https://artificialintelligenceact.eu/article/13/

  11. EU Artificial Intelligence Act, “Article 50: Transparency Obligations,” 2024. https://artificialintelligenceact.eu/article/50/

  12. CIO, “AWS to invest $5.3B in Saudi Arabia data centers,” 2024. https://www.cio.com/article/1311730/aws-to-invest-5-3-to-build-data-centers-in-saudi-arabia-to-bolster-tech-in-the-region.html

  13. About Amazon, “AWS and HUMAIN announce a more than $5B investment,” 2025. https://www.aboutamazon.com/news/company-news/amazon-aws-humain-ai-investment-in-saudi-arabia

  14. CNBC, “Saudi AI firm Humain is pouring billions into data centers,” August 2025. https://www.cnbc.com/2025/08/27/saudi-arabia-wants-to-be-worlds-third-largest-ai-provider-humain.html

  15. CIO, “HUMAIN: Saudi Arabia's bold bet on sovereign AI and Arabic LLMs,” 2025. https://www.cio.com/article/3984044/humain-saudi-arabias-bold-bet-on-sovereign-ai-and-arabic-llms.html

  16. NVIDIA Newsroom, “Saudi Arabia and NVIDIA to Build AI Factories,” May 2025. https://nvidianews.nvidia.com/news/saudi-arabia-and-nvidia-to-build-ai-factories-to-power-next-wave-of-intelligence-for-the-age-of-reasoning

  17. Datacenters.com, “Microsoft's Africa Push: New Data Center Investments,” 2025. https://www.datacenters.com/news/microsoft-s-africa-push-how-new-data-center-investments-are-redrawing-global-maps

  18. Google Blog, “Investing in connectivity, products and skills for Africa's AI future,” 2025. https://blog.google/around-the-globe/google-africa/africas-ai-future/

  19. Data Centre Magazine, “Google's Strategy for AI Data Centres in Africa,” 2025. https://datacentremagazine.com/news/googles-investment-in-african-ai-infrastructure-explained

  20. Data Center Dynamics, “Indian government launches $1.2bn IndiaAI mission,” 2024. https://www.datacenterdynamics.com/en/news/indian-government-launches-12bn-indiaai-mission-plans-10000-gpu-supercomputer/

  21. Press Information Bureau, India, “India's Common Compute Capacity Crosses 34,000 GPUs,” 2025. https://www.pib.gov.in/PressReleasePage.aspx?PRID=2132817

  22. Business Standard, “India doubles down on AI with Rs 10,000 crore fund,” June 2025. https://www.business-standard.com/industry/news/india-doubles-down-on-ai-with-rs-10-000-crore-fund-50-000-gpu-ambition-125060500096_1.html

  23. DD News India, “Transforming India with AI: Rs 10,300 crore mission, 38,000 GPUs,” 2025. https://ddnews.gov.in/en/transforming-india-with-ai-rs-10300-crore-mission-38000-gpus-a-vision-for-inclusive-growth/

  24. Library of Congress, “Brazil: Senate Advances Discussions on Bill to Regulate AI Use,” May 2025. https://www.loc.gov/item/global-legal-monitor/2025-05-23/brazil-senate-advances-discussions-on-bill-to-regulate-ai-use/

  25. Chambers and Partners, “Artificial Intelligence 2025 – Brazil,” 2025. https://practiceguides.chambers.com/practice-guides/artificial-intelligence-2025/brazil/trends-and-developments

  26. TI Inside Online, “SoberanIA forms alliance to deploy the largest AI infrastructure in Latin America,” December 2025. https://tiinside.com.br/en/09/12/2025/soberania-firma-alianca-para-implantar-no-brasil-a-maior-infraestrutura-de-ia-da-america-latina/

  27. BNamericas, “Brazil inks partnership with Scala Data Centers for sovereign AI infrastructure,” 2025. https://www.bnamericas.com/en/features/brazil-inks-partnership-with-scala-data-centers-for-sovereign-ai-infrastructure

  28. UNCTAD, “Brazil Launches the Brazilian Artificial Intelligence Plan 2024-2028,” 2024. https://investmentpolicy.unctad.org/investment-policy-monitor/measures/4930/launches-the-brazilian-artificial-intelligence-plan-2024-2028

  29. ITIF, “Indonesia's Data Localization Regulation,” June 2025. https://itif.org/publications/2025/06/09/indonesia-data-localization-regulation/

  30. Introl Blog, “Indonesia's First Sovereign AI Data Center,” 2025. https://introl.com/blog/indonesia-first-sovereign-ai-data-center-market-analysis

  31. SSEK Law Firm, “Regulation of Artificial Intelligence in Indonesia,” 2025. https://ssek.com/blog/indonesia-law-update-regulation-of-artificial-intelligence/

  32. Tony Blair Institute, “Sovereignty in the Age of AI,” 2025. https://institute.global/insights/tech-and-digitalisation/sovereignty-in-the-age-of-ai-strategic-choices-structural-dependencies

  33. UN Global Digital Compact, “AI Panel and Dialogue,” 2025. https://www.un.org/global-digital-compact/en/ai

  34. CSIS, “What the UN Global Dialogue on AI Governance Reveals About Global Power Shifts,” 2025. https://www.csis.org/analysis/what-un-global-dialogue-ai-governance-reveals-about-global-power-shifts

  35. World Economic Forum, “The UN's new AI governance bodies explained,” October 2025. https://www.weforum.org/stories/2025/10/un-new-ai-governance-bodies/

  36. Inside Privacy, “Global CBPR and PRP Certifications Launched,” June 2025. https://www.insideprivacy.com/cross-border-transfers/global-cbpr-and-prp-certifications-launched-a-new-international-data-transfer-mechanism/

  37. Amazon SageMaker AI Documentation, “Model Explainability,” 2025. https://docs.aws.amazon.com/sagemaker/latest/dg/clarify-model-explainability.html

  38. Amazon Bedrock, “Build genAI applications at production scale,” 2025. https://aws.amazon.com/bedrock/

  39. Tech in Africa, “North vs. Sub-Saharan Africa: AI Investment Trends,” 2025. https://www.techinafrica.com/north-vs-sub-saharan-africa-ai-investment-trends/

  40. Cogent, “The XAI Reckoning: Turning Explainability Into a Compliance Requirement by 2026,” 2025. https://www.cogentinfo.com/resources/the-xai-reckoning-turning-explainability-into-a-compliance-requirement-by-2026

Tim Green

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk

Discuss...