中央アジア・コーカサス企業の AI 実装率が世界平均の倍に

Companies in Central Asia and the Caucasus (CAC) are putting artificial intelligence into production at twice the rate of their global peers, according to a KPMG report.

Fifty-three percent of companies surveyed in the region said they had already reached industrial-scale AI implementation with measurable returns across multiple use cases, compared with 24% globally.

For the report KPMG surveyed more than 2,500 technology executives across 27 countries, including 40 respondents from the Caucasus and 32 from Central Asia.

Konstantin Aushev, partner and Head of Technology Practice at KPMG Caucasus and Central Asia, notes an important caveat. He supposes that some respondents may have interpreted “production” broadly.

"We suspect that in our region the respondents included in their responses such fragmented cases as text and voice bots," Aushev told Euronews.

The caveat, however, does not erase the region’s progress. There are examples of genuinely large-scale projects in the region.

At Kazakhstan's national communications operator Kazakhtelecom, AI has been introduced since mid-2025 across customer service and contact centres, procurement, human resources and administration, finance, regulatory documents, occupational safety and recruitment.

"The implementation model is mixed. The central team is responsible for platforms, security, quality, and complex products. For smaller tasks, we use an agent creation platform," Yerzhan Assanov, AI officer at Kazakhtelecom told Euronews.

The platform allows an employee without programming skills to give instructions, add working materials, and assemble an agent for their needs.

Business & Technology Services (BTS), the IT company within the Eurasian Resources Group (ERG), is using computer vision to check serviceability of equipment on mining sites and data analysis in geological reports processing.

ERG is a company that operates Kazakhstan's chrome, iron and aluminium production entities.

Why the region moved quickly

The KPMG report finds that companies in the region are often choosing a proactive experimentation approach - 44% vs. 21% globally - which helps explain why Central Asia and the Caucasus may have moved faster than more mature markets in some areas of AI adoption.

Rustem Uvaliyev, BTS Head of the Digitalisation Department, told Euronews that his company runs AI pilots continuously. Around 10% of its generative AI pilots make it into production, while the figure is about 80% for “classical AI”, such as machine learning and computer vision.

Kazakhtelecom says it launches pilots of new AI solutions almost every month, but it's too early to draw conclusions about the rate of their transition to industrial operation.

Assanov said that at this stage, they assess how reliably the solutions work on real data, how employees use them, and what needs to be changed before scaling.

"We pilot scenarios with a specific process owner who is willing to promote the solution within the company and who has a clear way to evaluate the business impact," he added.

BTS has also developed a self-service platform called FAIR that allows employees to create digital agents using generative AI.

"After a short training session, each employee can create a digital assistant in a test environment. After testing, the most viable cases are transferred to production," Uvaliyev said.

Both companies are experimenting with vibecoding platforms, where the employee can describe the required solution using words and quickly assemble it without setting up servers, domains, or a technical environment. This will potentially address the issue of personnel shortage.

Selective approach to pilots

Compared to companies worldwide, the region's approach to AI appears to be more selective.

Konstantin Aushev said the region was more chaotic in its approach to AI couple of years ago, when companies competed to announce the number of AI use cases they had launched, but that “Brownian motion” is now beginning to diminish.

“Today, a certain chaos remains, but nevertheless, there is a deeper understanding of where we need artificial intelligence and where we don't,” he said.

That shift is consistent with KPMG's finding that 43% of regional companies favour “rapid but selective adaptation”, compared with 36% globally.

Assanov says that over the last year Kazakhtelecom's approach became bolder.

"We launch experiments earlier and test them faster on real-world tasks," he said.

He added that now they start by analysing the process in advance and introduce stricter quality requirements.

Uvaliyev, however, notes that "hype and pressure" to implement AI remains, and understanding that it's not a universal solution is important for scaling truly necessary AI initiatives.

"Another problem is attempting to scale AI without assessing the costs. Many large international companies have already encountered this and are beginning to evaluate their initiatives more soberly," he said.

Aushev also attributes part of the region's rapid AI adoption to its relatively high level of digital penetration. Many companies already have the infrastructure and experience needed to connect systems to data and external sources, making it easier to introduce AI into existing services.

The strategy paradox

Although the region is considerably more willing to experiment, that does not mean every experiment becomes an enterprise-wide transformation.

Only 34% of CAC companies report having a clear company-wide AI strategy aligned with business and technology goals, compared with 76% globally. A cross-functional AI governance structure exists at just 25% of regional companies, versus 69% worldwide.

Aushev called this one of the most paradoxical findings of the research: companies are deploying AI at much higher rates than globally, while far fewer have formalised how they intend to develop and govern it.

"The same applies to existing AI governance mechanisms. Only a quarter of companies in our region say they have them in place, and even fewer say they can clearly measure the return on investment in AI," he added.

The report author noted however, that since the survey had been conducted, companies have already come to the point where it is necessary to approach the issue systematically and develop policies, rules and procedures.

The same shift is visible at the government level, he added, with Kazakhstan introducing more formal rules around artificial intelligence and the digital economy.