AI as Infrastructure: Investment, compute, power and capital formation across Kazakhstan and the Turkic space

Introduction

This report examines artificial intelligence as a new class of economic infrastructure and assesses the investment opportunity it creates for Kazakhstan and the wider Turkic space. It reviews the global AI infrastructure buildout, compares sovereign AI policy models, evaluates Kazakhstan’s readiness in power, connectivity and compute, and considers how the five member states of the Organisation of Turkic States can complement one another. The final chapter sets out the financing structures through which AI infrastructure is funded internationally and the role the AIFC can play in adapting them for the region.

Key Findings

AI investment has become a macroeconomic variable, and electricity is its main condition

Global investment in AI data centres is expected to reach around USD 5.2 trillion by 2030, placing AI infrastructure on the scale of the energy and transport sectors. At the same time, data-centre power demand is projected to grow 2.7 times over the same period, and about one in five planned projects is waiting for grid connection. The location of AI investment is increasingly determined by the availability of power, land and a predictable delivery timeline – factors in which Kazakhstan holds a comparative advantage.

Sovereign AI models differ, but converge on one practical lesson

The United States relies on private capital, China on state coordination, and the European Union on shared public compute under common rules; the UAE, India and Singapore offer further variations. Each model works within its own conditions and reaches its own limits. For mid-sized economies the consistent lesson is to retain control over sensitive data and critical workloads while contracting the remainder with trusted partners.

The Turkic region offers complementary strengths and a basis for shared infrastructure

The five OTS member states represent close to USD 2 trillion of GDP and more than 160 million people, with at least USD 23 billion in announced and targeted AI-infrastructure investment and at least 2.5 GW of planned capacity. National strengths are complementary: demand and cloud scale in Türkiye, an investment pipeline in Uzbekistan, westbound connectivity through Azerbaijan, digital public services in Kyrgyzstan, and power-backed compute in Kazakhstan. The report proposes a model in which sensitive systems remain national while selected compute and cloud services are shared across compatible national platforms.

AI infrastructure can be financed layer by layer

AI infrastructure consists of four asset layers with different lifetimes – land, grid and power; data-centre facilities; chips and servers; and cloud services – each suited to a different type of capital, from project finance and development institutions to contract-backed private credit, bonds and sukuk. International transactions by Meta and BlackRock, CoreWeave and AirTrunk show these structures in practice.