Artificial intelligence has definitively transformed from an experimental technology into a key macroeconomic driver, launching the next phase of Industry 4.0. According to the IEA, capital expenditures of the five largest technology companies on AI infrastructure exceeded $400 billion in 2025, with a further 75% jump expected in 2026. Global data center electricity consumption will double by 2030, reaching 945 TWh — comparable to the energy consumption of all of Japan. Electricity and computing capacity are no longer a routine budget line — today they are fundamental strategic assets.

We are now witnessing how historically established digital hubs — Frankfurt, London, Amsterdam, Paris, and Dublin (the so-called FLAP-D markets) — are failing to meet market demand. As a result, grid connection queues at key European hubs are now stretching to 7–10 years, making it impossible to rapidly deploy new clusters and freezing investor capital.

Vladislav Minkevich, CEO of AKASHI Data Center being built in Astana:

“In reality everything is very straightforward. You can come up with as many brilliant algorithms as you like, but the logic builds from the bottom up. Without basic energy, concrete, and heavy iron, there will be no high technology — full stop.”

Mass technology adoption is going far beyond Big Tech. According to McKinsey’s global survey (The State of AI 2025), 88% of organizations are already regularly using AI in their business processes, and 62% are actively experimenting with autonomous AI agents. The PwC report (28th Annual Global CEO Survey) shows that already more than a third of CEOs (32%) are recording real revenue growth from Generative AI deployment, while 56% note significant improvements in employee efficiency.

The enterprise sector (Enterprise AI) is generating enormous demand for independent local capacity. The Uptime Institute 2025 report shows that classic public clouds are no longer the only solution: 46% of AI inference IT workloads are deployed in companies’ own data centers, and a further 34% in commercial colocation facilities. The main driver is the need for strict data control — nearly 70% of enterprise customers cite Security by Design and privacy as the primary factor.

However, old facilities are conceptually obsolete and unprepared for these challenges. Since power density in modern racks often exceeds 100 kW, traditional air cooling becomes completely ineffective.

Vladislav Minkevich:

“Old infrastructure physically cannot handle new workloads. A complete generational change is underway. If 5–10 kW per rack was once the norm, today clusters demand 80–100 kW.”

To ensure continuous operation of LLMs and deep machine learning, next-generation data centers (such as AKASHI at 100 MW) are conceptually different from their predecessors. Key requirements:

  • Ultra-dense computing clusters: approximately 27% of racks in the AI training segment have crossed the 50 kW threshold.
  • Liquid Cooling transition: more than half (51%) of data center operators deploying AI are being forced to upgrade cooling systems.
  • Power distribution re-engineering: 52% of data center owners are upgrading electrical networks to avoid downtime during peak loads.
  • Alternative geography: investors are shifting focus to new markets with reliable energy systems. The industry is selling megawatts, not square meters.

Building such facilities will allow Kazakhstan to address the infrastructure needs of domestic business and claim export positions in the global digital economy.

Source: ER10