Relevant for: procurement directors and category managers at Dutch industrial companies with 250 or more employees (machine building, chemicals, high-tech, agrifood) who are including AI in their category plans and software contracts for 2027.
According to CBS, 66.2% of Dutch companies with 250 or more persons employed used AI technology in 2025 [1]. Among SMEs the figure was 29.8% [1]. The question for procurement is therefore not whether AI arrives, but how to give it a permanent place. That requires three things: usable master data, demonstrable AI literacy and contracts that settle data, costs and exit. The high-risk rules of the AI Act move to 2 December 2027 [6][7]. Use that room to amend contracts at renewal.
What is actually happening
AI is no longer an experiment in large Dutch companies. CBS measures that 66.2% of companies with 250 or more persons employed used at least one AI technology in 2025 [1]. In small and medium-sized enterprises (10 to 249 persons employed) the figure was 29.8% [1]. Across all companies with 10 or more persons employed, it is roughly one third [2].
The Netherlands is therefore above the European average. Eurostat reports that in 2025 20.0% of EU enterprises with 10 or more employees used AI, up from 13.5% in 2024 [3]. Among large EU enterprises the figure was 55.03% [4]. Denmark scores highest at 42.0% [3].
Two details from the CBS research matter most for procurement. First, SMEs and large companies use AI most often for administrative processes or management tasks, not for marketing and sales [2]. Second, AI mainly enters through commercial software. Among micro-enterprises, 48.2% obtained AI technology through direct use of commercial software [2]. CBS does not publish that share for larger companies in this report.
On the regulatory side, two dates are fixed. Since 2 February 2025, providers and deployers of AI systems must take measures to ensure sufficient AI literacy among their staff [9]. On 29 June 2026 the Council of the EU gave its final approval to the AI omnibus, which simplifies the AI Act [5]. The regulation was published on 24 July 2026 and entered into force on 27 July 2026, law firm DLA Piper reports [8].
Structural drivers
Regulation. The omnibus moves the high-risk obligations. For stand-alone high-risk systems (Annex III) the date becomes 2 December 2027, and for AI in regulated products (Annex I) 2 August 2028 [6][7]. The literacy obligation remains. According to Gibson Dunn, the agreement does shift that obligation from guaranteeing a level to supporting its development [7]. Have your own legal department assess which applications fall under which regime.
Technology. Gartner placed generative AI for procurement in the trough of disillusionment in its 2025 Hype Cycle, Procurement Magazine reports [10]. Gartner does see efficiency gains among early adopters and expects full productivity within five years [10]. The research firm, which itself sells advisory services, puts the brake on data. According to Gartner, procurement's low data maturity limits the usefulness of autonomously acting (agentic) AI [11].
Procurement model. Gartner advises CPOs to work with data and analytics leaders to make procurement data AI-ready and to set governance [11]. Among micro-enterprises, AI mainly enters through commercial software [2]. Orbecta analysis: in large companies too, AI in procurement runs through software packages, and therefore through contracts that procurement itself concludes.
Who is directly affected
Direct: procurement organisations in large companies that use source-to-pay, ERP or contract management software in which suppliers build AI functions. Here AI is already in the tools, whether or not the department has consciously chosen it.
Direct: category managers for IT and software contracts. They conclude the contracts through which AI enters the company.
Indirect: buyers in manufacturing who assess price claims, spend analyses and supplier risks. Their work changes when their tools start delivering AI analyses based on their own master data.
Less directly affected: smaller companies with simple procurement processes and without integrated procurement software. CBS measures 13.8% AI use among micro-enterprises [1]. If they buy a package with built-in AI at their next software renewal, the same contract and data questions apply to them too.
The second-order effect
Orbecta analysis: the chain runs as follows. Software suppliers build AI into standard packages. Procurement organisations therefore receive AI through their existing contracts. The negotiation shifts from the question "do we want AI?" to the question "under what conditions does AI run on our data?".
Scenario 1: AI as a paid module. This scenario applies as soon as a supplier prices AI functions as a separate module or consumption bundle. Then total licence costs rise outside the existing indexation. The budget for procurement software then grows faster than the agreed annual price adjustment. Whoever fails to fix this at renewal negotiates afterwards from a weak position.
Scenario 2: AI on polluted master data. This scenario applies as soon as an organisation deploys AI without first cleaning supplier, item and contract data. Then the tool delivers spend analyses with duplicate suppliers and wrong classifications. Category managers lose trust in the output and fall back on Excel. AI then gets no permanent place, however good the tool is.
Scenario 3: literacy as burden of proof. This scenario applies as soon as a supervisor or auditor asks how staff have been prepared for AI use. Then you must demonstrate which measures you have taken since 2 February 2025. A single e-learning without records is then too thin.
The procurement indicators are measurable: the share of AI modules in your licence invoices, the percentage of classified spend, the number of duplicate suppliers in your master data and the share of staff with recorded AI training.
Dutch procurement impact
Contract clauses with software suppliers
Direct: every renewal of an S2P, ERP or CLM contract (source-to-pay, enterprise software and contract management) becomes an AI negotiation. Fix whether the supplier uses your data to train models. Fix where that data is stored and who owns analyses created by AI. Also arrange an exit: how do you get your cleaned data and classifications back if you switch?
If the supplier prices AI separately: then include a price cap or fixed consumption bundle. Link price revision to the same index as the base licence. This prevents AI costs from rising outside your indexation clause.
Supplier conversations and transparency
Direct: ask suppliers which AI functions are switched on by default. Also ask whether new AI functions switch on automatically through updates. Procurement must know what is running before it steers usage.
If an AI function affects decisions about people: think of assessing staff from temporary staffing suppliers. Then have legal assess whether the high-risk rules apply from 2 December 2027 [6]. That is a legal question that differs per application.
Data foundation and budget
Direct: AI needs clean master data. Gartner names procurement's low data maturity as the main brake [11]. In the 2027 budget, reserve money first for data cleansing and spend classification. Only then does expansion to more AI applications pay off.
Indirect: Orbecta analysis: the more item numbers and suppliers in the Netherlands and Germany you manage, the more data quality determines the outcome of AI.
Roles and AI literacy
Direct: the literacy obligation has applied since 2 February 2025 to every organisation that deploys AI systems [9]. Even after the omnibus, a support obligation remains [7]. Define per role what a buyer must be able to do. A category manager checks an AI spend analysis personally. A contract manager places AI summaries next to the source document.
If you use AI for claim assessment: then always have the tool refer to the source index and the contract clause. AI advice without a source is no argument in a negotiation.
What is certain and what remains uncertain
What is certain
- 66.2% of large Dutch companies and 29.8% of SMEs used AI technology in 2025 [1].
- EU-wide, the figure was 20.0% of enterprises with 10 or more employees, up from 13.5% in 2024 [3].
- The AI literacy obligation has applied since 2 February 2025 [9].
- The Council gave final approval to the AI omnibus on 29 June 2026 [5]. The high-risk rules apply from 2 December 2027 (Annex III) and 2 August 2028 (Annex I) [6][7].
What remains uncertain
- How national supervisors will check the softened literacy obligation in practice.
- Whether software suppliers keep AI functions in the base licence or price them as a paid module.
- Whether generative AI for procurement reaches the productivity Gartner expects within five years [10].
What to watch
- The share of AI modules and consumption charges on your software invoices per quarter.
- The percentage of classified spend and the number of duplicate suppliers in your master data.
- Release notes from your S2P and ERP supplier: which AI functions switch on by default?
- Guidance from the European Commission and national supervisors on AI literacy and high-risk classification.
What you can do now
- This month: ask your S2P or ERP supplier in writing which AI functions are active and whether your data trains models. Then decide whether you need an AI addendum.
- This quarter: at the next software renewal, fix data ownership, an exit with data return and a price cap for AI modules. Decide which terms are non-negotiable.
- This quarter: have your ERP or data services partner run a quality scan on supplier and item master data. Then decide which two AI applications you scale up in 2027.
- This year: ask your training partner for a role-based AI programme for buyers with records per employee. Decide which roles complete it before the 2027 category plans.
How Orbecta supports
With Contract Intelligence you test your software contracts for data clauses, AI price revision and exit arrangements before renewal. Warroom Prep helps you prepare the negotiation with your S2P or ERP supplier, including the terms you do not want to give up. Company exposure: not yet calculated. Connect spend and supplier data.
Sources
Primary source:
- [1] CBS — Use of AI technology by Dutch micro-enterprises, chapter 2: Use of AI technology — March 2026
- [2] CBS — Use of AI technology by Dutch micro-enterprises (full report) — March 2026
- [3] Eurostat — 20% of EU enterprises use AI technologies — 11 December 2025
- [4] Eurostat — Use of artificial intelligence in enterprises (Statistics Explained) — December 2025
- [5] Council of the EU — Artificial intelligence timeline: Council gives final green light to simplify and streamline AI rules — 29 June 2026
- [6] European Parliament — Legislative Train: Digital Omnibus on AI — date not specified
Additional context:
- [7] Gibson Dunn — EU AI Act Omnibus Agreement: Postponed High-Risk Deadlines and Other Key Changes — 27 May 2026
- [8] DLA Piper — The Digital AI Omnibus: Proposed deferral of high risk AI obligations under the AI Act (update) — 30 June 2026
- [9] Covington (Inside Privacy) — European Commission Provides Guidance on AI Literacy Requirement under the EU AI Act — March 2025
- [10] Procurement Magazine — Gartner: Generative AI in Trough of Disillusionment — August 2025
Market research:
- [11] Gartner — 2025 Trends for Chief Procurement Officers — 2025
Orbecta analysis based on the sources above.