# AI in Türkiye: What Adoption Figures Leave Out

> What Türkiye's AI statistics and action plan tell business leaders about adoption—and what they cannot tell us about suppliers, operations or competitiveness.

A supplier can use AI to write a clearer quotation and still take just as long to send it. Purchasing has to confirm the material cost, production has to provide a delivery date, and the owner has to approve the price. The document improves while the customer keeps waiting.

That is a hypothetical example, but it captures the distinction I would make when assessing AI adoption in Türkiye. For a manager running a Turkish business, or evaluating Turkish suppliers and partners, AI use says little by itself about delivery reliability. We need to know what changed in the work.

AI use is highly visible in my LinkedIn feed and among the businesses I encounter. I expect a much more uneven picture across Türkiye, particularly in manufacturing, construction and large parts of the service sector. That expectation is my judgment, not a finding from a representative study. My professional circle cannot stand in for the country's businesses.

I expect use to spread. I do not expect anything approaching uniform, advanced use in the short or medium term. The risk is that competitors may adapt faster than management is prepared to.

## Reading Türkiye's 7.5% figure

In its [Artificial Intelligence Statistics, 2025 bulletin](https://veriportali.tuik.gov.tr/Bulten/Index?p=Yapay-Zeka-Istatistikleri-2025-57945), the Turkish Statistical Institute, TÜİK, reports that 7.5% of enterprises covered by the survey said they used at least one AI technology. The survey covers enterprises with ten or more employees; it does not represent smaller businesses. In information and communication, the figure was 47.1%.

These are 2025 figures, not a current assessment of every Turkish business. The definition includes chatbots, machine-learning analysis and autonomous systems. The overall percentage cannot tell us how many companies have reorganised their operations around AI.

It also does not prove that most users are doing little more than editing text. Among enterprises using AI, 41.1% reported using it for production or service processes. Those reported purposes do not show how deeply AI is used or whether it produced measurable gains. The bulletin's findings on expertise, cost and legal uncertainty concern non-users considering adoption, not all non-users.

I suspect many companies have yet to connect AI with a problem in their own business. These figures do not tell us how many. Hearing about a technology, considering it and depending on it in daily operations are different levels of adoption.

For someone evaluating a Turkish business from outside the country, the implication is practical: neither the overall survey figure nor the examples most visible online can replace looking at the company itself. Inside the business, the same distorted picture can lead management to buy tools because everyone seems ahead, or dismiss AI because the visible examples seem irrelevant. Both decisions skip the work of establishing what needs to improve.

## Where the quotation gets delayed

Using ChatGPT to improve a customer email is useful work. A small, recurring problem can deserve a simple solution. Useful AI does not have to involve a complex system.

In the hypothetical manufacturing company, the sales employee uses AI to summarise the customer's request and draft the quotation. That can save time even while the approvals remain unchanged. But if the business wants to respond to customers sooner, it also needs current costs, visibility into production capacity and a workable process for approving prices.

Existing software might address part of the problem. Clearer decision-making authority might address another. AI can help when the task is suitable; buying a more capable model will not by itself clear the approval queue.

By maturity, I mean that the company knows which work it changed and why, checks the result, and can sustain the practice beyond one employee's enthusiasm. A company doing this well in one narrow area may be further ahead than one that has bought dozens of tools.

![An AI-assisted quotation sits in the foreground while purchasing, production and owner approvals remain separate waiting points](/images/inline-ai-adoption/ai-use-is-not-operational-maturity.webp)

For a customer assessing that supplier, a more polished quotation is an improvement in communication. A quotation that arrives earlier with an accurate price and a dependable delivery date is evidence of a different operational improvement. It is worth keeping those gains separate.

## Management has to change how it works

When a business has long depended on a few people's knowledge and decisions, changing the process means challenging how authority works.

Making pending quotations visible exposes where they wait. Delegating price approval means the manager no longer decides every case. Keeping cost data current can bring longstanding pricing habits into question. Tool training alone will not settle those matters.

There is also the time required from the people doing the work. A team trying to finish today's production is being asked to organise data and test a new method. If management makes no time for the work, leaves responsibility unclear or returns to the old method after the first mistake, the application may fall out of use.

That is why I expect meaningful operational change to be harder than the spread of access. It also explains why I would hesitate to assume that a very large majority of Türkiye's businesses will achieve advanced AI maturity soon.

Computers offer a useful analogy. One company uses them to prepare documents, another to track stock and orders, another to plan production. Similar access can coexist with very different ways of managing the business.

AI may spread through features added to existing software, employees' experiments becoming routine, or a system built for a particular need. Companies need not follow the same sequence. Ready-made services may make difficult tasks easier over time, so today's lack of preparation need not be permanent.

The analogy does not give us a timetable. Earlier computerisation took years; that does not mean companies will have the same amount of time now. AI could reduce the cost of a task faster than an established business can adapt.

## How far public support can reach

Türkiye introduced its Artificial Intelligence Action Plan for 2026–2030 on 13 June 2026. The related Presidential Circular No. 2026/9 was published on 18 August 2026, according to the [Union of Municipalities of Türkiye's publication notice](https://www.tbb.gov.tr/tr/mevzuat-duyurulari/turkiye-yapay-zeka-eylem-plani-2026-2030-ile-ilgili-20269-sayili-cumhurbaskanligi-genelgesi).

The [government's presentation of the plan](https://www.iletisim.gov.tr/turkce/haberler/detay/cumhurbaskani-erdogan-turkiye-yapay-zeka-eylem-planini-acikladi) sets out four themes, broadly covering awareness, use, production and management. Announced goals include AI literacy, training people to apply AI, and AI vouchers for small and medium-sized enterprises in priority areas. These are policy intentions; the announcement does not establish that firms have received support or improved their operations.

Some companies first need help seeing where AI could fit their work. That gives the plan's emphasis on awareness a practical purpose. Discussing why a quotation is late can make that connection more effectively than a general presentation on the technology.

Public support can reduce the cost of trying something and improve access to expertise. Its effect will still depend on whether the business assigns someone to continue the work and makes time for data preparation, process changes and learning.

I would assess the plan partly by whether it reaches firms that are not ready to apply for support. If assistance mainly reaches companies that can already define a project, they may move ahead while less prepared firms remain outside the programme. That is an implementation risk to watch, not a verdict on the plan's results.

Training and usage counts need to be read alongside evidence that the change lasts: whether an application is still used after support ends, quotations go out sooner, errors decline, and the method survives an employee's departure. For a company working with a Turkish partner, a national policy announcement gives a reason to follow developments. It does not establish that the partner can yet perform differently.

![A team tracks a shared workflow with a clear owner and compares abstract before-and-after measures of operational change](/images/inline-ai-adoption/ai-adoption-needs-continuity.webp)

## Uneven adoption can still raise the competitive pressure

A low reported adoption figure does not guarantee a slow-moving company time to wait. Where AI helps competitors reduce costs or improve service, I expect not using it to become more costly. A few firms improving a relevant part of their operation can raise customers' expectations before advanced use becomes common across a sector.

The competitor that supplies an accurate price and a reliable delivery date sooner may become easier to buy from. It may also have time to consider more enquiries while another firm struggles to handle its existing workload with the same staff.

Whether AI caused that advantage needs to be measured separately. Better data, simpler approvals or a stronger sales operation could produce the same difference. Customers notice the service before they know which technology made it possible.

Pressure may show up in prices, delivery times or the requirements of a major customer. Even so, the adoption figures cannot tell us that businesses without AI will disappear, or when. Capital, product quality, relationships, local access and physical capacity can sustain an advantage. Some firms may adapt, buy ready-made services or continue in a narrower market. Others may lose orders, shrink or close.

Survival can also conceal a weaker business: lower margins, more effort from the people keeping it running and less room to grow. That is one reason I am hopeful about wider access to AI while expecting much more work from company management. Public policy can help more businesses begin; management has to make the improvement last.

Whether you run the company or depend on it as a partner, follow the quotation beyond the screen. If it still waits for the same decisions, better wording may not be the next improvement the business needs.

---

Language: English
License: CC BY 4.0
License URL: https://creativecommons.org/licenses/by/4.0/
Scope: Evren Bal-authored text, unless this article expressly states otherwise.
Excluded: Third-party material, quoted excerpts, logos, and separately marked images retain their own rights.
Attribution: Credit Evren Bal, link to the canonical source and license, and indicate changes.
Source: https://evrenbal.com/ai-will-spread-across-turkiyes-businesses