Update (October 2, 2026): TÜİK (the Turkish Statistical Institute) has published its Artificial Intelligence Statistics, 2026 bulletin. The share of enterprises using artificial intelligence (AI) rose from 7.5% in 2025 to 14.0% in 2026; among enterprises with 250 or more employees, the share is 37.1%. Among enterprises considering adoption, the biggest barrier is once again a lack of expertise (72.3%), followed by uncertainty over legal consequences (66.4%) and data-protection concerns (65.4%). The analysis below is based on the 2025 data.
The Turkish Statistical Institute published data on AI use as official statistics for the first time on October 1, 2025. The picture is clear: among enterprises with 10 or more employees, the AI usage rate was 2.7% in 2021 and rose to 7.5% in 2025. The increase is rapid, but the starting point is low. In the same year, across the OECD countries for which data is available, 20.2% of firms reported using AI.
TÜİK's next bulletin on this subject was published on October 2, 2026; the latest figures are in the note at the top of this article. Below, we look at what the 2025 data says and how organizations can close this gap.
The picture: where does Türkiye stand?
The size gap is pronounced. According to TÜİK data, the share of enterprises using AI in 2025 breaks down by size as follows:
- 10-49 employees: 6.6%
- 50-249 employees: 9.6%
- 250 or more employees: 24.1%
A similar divide exists across the OECD: 52.0% of large firms use AI, while 17.4% of small firms use it. Because the two datasets were compiled by different institutions, the comparison is indicative; the message is nonetheless the same: AI use rises markedly as company size grows.
Looking at the purposes of use, enterprises in Türkiye that use AI turn to the technology most often in marketing or sales (46.5%), production or service processes (41.1%), and accounting, control or financial management (33.7%).
Barriers: expertise, cost, legal uncertainty
For 2025, TÜİK reported 9.0% as the share of enterprises that do not use AI but are considering it. The main reasons these enterprises cite:
- Lack of the relevant expertise within the enterprise: 74.2%
- Costs being too high: 67.4%
- Legal uncertainty over liability in case of potential harm: 62.4%
All three barriers have more to do with organizational readiness than with technology. In other words, the solution is not just choosing a tool; it is building capabilities, budget discipline and a governance framework together.
The global picture: use is widespread, value is selective
The experience of early-adopting organizations also counsels caution. In McKinsey's 2026 global survey, 80% of respondents say AI has increased their individual productivity; however, the share reporting at least some AI-attributable impact on earnings before interest and taxes (EBIT) stays at 37%. The share of "high performers" — organizations that attribute at least 5% of their EBIT to AI and describe the impact as "significant" — is about 6%.
Gartner issues a similar warning about more ambitious projects: it predicts that more than 40% of agentic AI projects will be canceled by the end of 2027 because of rising costs, unclear business value or inadequate risk controls.
These two findings also point to an advantage for late starters: by learning from the trial-and-error costs that others have already paid, it is possible to start directly in areas that deliver value.
From pilot to scale: a five-step roadmap
1. Choose the use case by value
Choose your first project by its payback period, not by the appeal of the technology. According to Gartner's survey of finance leaders, use cases such as data extraction, automation of accounts payable and receivable processes, and report generation typically deliver returns within 9-10 months. Gartner also recommends using the agentic approach only where it delivers clear value, automation for routine workflows, and assistants for simple information retrieval.
A practical checklist:
- Is the process repetitive and high-volume?
- Can the success metric (time, error rate, cost) be measured today?
- Is a human approval point defined for cases of error?
2. Establish data and governance from the start
Legal uncertainty is one of the three main barriers in the TÜİK data. In Türkiye, the first document to consult in this area is the Guide on Generative Artificial Intelligence and the Protection of Personal Data published by the Kişisel Verileri Koruma Kurumu (KVKK, Türkiye's Personal Data Protection Authority). The guide explains the principle of data minimization as keeping personal data "limited to only what is necessary" and stresses the importance of data controllers defining the purpose of processing separately at each stage.
Within the organization, this translates into:
- A usage policy that specifies which classes of data may be entered into which tools
- Clarified data controller and data processor roles with suppliers
- An approval workflow in which model outputs are reviewed
3. Map the regulatory timeline
For organizations working with the EU market, the timeline of the European Union's AI Act should also be part of planning. Under the legislation that entered into force on July 27, 2026, the rules for high-risk systems (Annex III) will apply from December 2, 2027, and for systems embedded in products from August 2, 2028. The postponed date allows time to prepare; however, starting the system inventory and risk classification today will make the work easier.
4. Build capability in-house
In the TÜİK data, lack of expertise is the most frequently cited barrier, at 74.2%. Gartner also now names low AI literacy as the most important barrier in the finance function. External support can accelerate progress; however, lasting value emerges when the teams who know the process can adapt the tool to their own work. Assigning a business-unit owner to each pilot project and planning short, role-specific training sessions is a good place to start.
5. Measure, stop or scale
McKinsey's data shows that individual productivity gains do not automatically turn into financial impact for the organization. For this reason, a metric, a time frame and a decision point should be set at the outset for every pilot. The portfolio approach Gartner recommends to finance leaders supports this: knowing where to invest and when to terminate underperforming initiatives.
A practical framework for the first 90 days
Tying the roadmap to a concrete calendar prevents decisions from being put off. The framework below offers a starting point that most organizations can adapt, whatever their size:
- Days 1-30 — Discovery: Compile a process inventory; identify three candidate processes that are repetitive, high-volume and measurable. List the AI features that already exist in your current tools but go unused.
- Days 31-60 — Framework: Put the data classification and usage policy in writing; record the success metrics for the pilot and the current baseline (time, errors, cost).
- Days 61-90 — Pilot: Run the pilot in a single process, with a limited team and a defined human approval point. At the end of the period, decide whether to continue, change or stop based on the metrics.
This sequence breaks the investment into small steps and carries what the organization learns at each stage into the next.
Conclusion
AI use in Türkiye is growing quickly, but it remains well behind the OECD-wide rate. This gap is an opportunity when approached correctly: organizations that start with short-payback areas, build the data and regulatory framework from the outset, and grow capability in-house move from experimentation to scalable use.
Our work draws on hands-on operations experience. One of our consultants is a senior executive with more than 30 years in operations, logistics and supply chain. At a technology distribution company of roughly one billion dollars in scale, he has run logistics, warehousing, technical service and administrative operations end-to-end, leading teams of 100+ people and delivering lasting gains through KPI-driven programs. That logistics and supply-chain discipline now goes into our AI and automation work at SME scale.
If you would like to assess where your organization stands on this journey and clarify your priorities with us, let's plan your transformation roadmap together.
#interimmanagement #forwarddeployedengineering