Substitution Risk

Where AI can actually substitute the revenue

The fear behind Wolters Kluwer's derating is that generative AI lets a professional ask a chatbot instead of consulting proprietary content. Mapping the €6,125 million 2025 revenue base by how substitutable each part is, that exposure is narrower than the multiple implies: print — the one format a chatbot most cleanly replaces — is down to 4.75% of revenue, level with RELX's 4%. What WKL lacks is RELX's structurally protected Risk division, which is ~90% machine-to-machine. Its content is more human-facing, and that is the honest reason it is discounted more.

Three disclosed cuts of the revenue base

Wolters Kluwer discloses its revenue three ways, each additive to the same €6,125 million. Together they bound the question of what AI could displace.

Source: FY2025 Annual Report — Group financial review (p.39) [1] and Note 6 Revenues (p.163) [2].

The first cut is by media format. Digital delivery is €5,250 million (86% of revenue), services €584 million (10%), and print €291 million — just 4.75% [3]. The second cut is by revenue recognition: €4,999 million (82%) is recognised over time — subscriptions that renew — and €1,126 million (18%) is recognised at a point in time, the transactional and one-off character [4]. The third cut is by product type: expert solutions — software plus advanced information solutions — are 59% of revenue, of which software solutions are about 45%; within software, cloud has reached 46% and grew 15% organically [5] [6].

Source: FY2025 Annual Report, Note 6 Revenues (p.163) [7]; FY2024 Annual Report, Note 6 (p.170, 2023 comparatives) [8].

Print is the sliver at the top of each bar, and it is shrinking in both euros and share — €327 million to €291 million over two years, 5.9% of revenue to 4.75% [9] [10]. Digital is not one thing, though, and the substitution question turns on splitting it.

A substitution ladder

Grouping the base by how easily a general-purpose model displaces it gives four tiers. The media-format split (software sits inside digital) anchors it; the software share is derived from the disclosed 45%, and the tiering is judgment, not a reported line.

Source: media-format and product-type figures from FY2025 Annual Report (p.39, p.163) [11] [12]; software share (45%) from FY2024 Annual Report (p.8) [13]; tiering derived.

Embedded software (~45%, ~€2,756 million). CCH Axcess, CCH Tagetik, OneSumX and TeamMate are systems of record wired into a firm's tax workflow, a bank's regulatory reporting, or an audit function [14]. A chatbot does not replace the platform that files the return or runs the Basel calculation; if anything it is embedded inside it. This tier grew: cloud software rose 15% organically to 46% of software revenue [15].

Services (~10%, €584 million). Almost all of this — €535 million — sits in Financial & Corporate Compliance, where Wolters Kluwer acts as registered agent, performs the legal-entity filing, and runs the lien search [16]. The value is the completed transaction, not an answer a model could generate. Alongside it, €447 million of transactional revenue in FCC and Legal & Regulatory is the same transaction-rail character [17]. It is cyclical — it moves with corporate and lending activity — but it is not content a chatbot substitutes.

Digital expert and reference content (~41%, ~€2,494 million). This is the contested tier: online information and decision-support that a professional consults — UpToDate for a clinician, legal and tax research, regulatory reference. It is where generative AI is both the threat and Wolters Kluwer's own product. The company's answer is to convert the content into the delivery channel: nearly 70% of digital revenue is now from AI-powered solutions, and the flagship products are wrapped in the expert-in-the-loop validation examined in Expert AI [18]. Renewal above 90% held through 2025 — reassuring, but a lagging indicator [19].

Print (4.75%, €291 million). The clearest substitution target — a medical or legal reference book a model can paraphrase — is already immaterial and declining. It concentrates almost entirely in two divisions.

Print is concentrated, and matches RELX

Print books and subscriptions live in Health (medical reference, €134 million) and Legal & Regulatory (legal reference, €118 million); together those two carry 87% of group print, while Tax & Accounting (€34 million), FCC (€5 million) and Corporate Performance and ESG (nil) are effectively paperless [20].

Source: FY2025 Annual Report, Note 6 Revenues — media format by segment (p.163) [21].

The "reference-content dinosaur" version of the bear case is not supported by the number. At 4.75% of revenue, Wolters Kluwer's print exposure is essentially level with RELX's, which has spent 25 years running print from 64% of revenue down to 4% and now manages it separately as a runoff [22]. On the legacy-format axis, the two are the same business.

What RELX has that Wolters Kluwer does not

The genuine gap is one level up. RELX's Risk division — its highest-growth, highest-multiple engine — derives over 90% of revenue from machine-to-machine interactions: data and scores wired directly into customers' fraud, credit and compliance pipelines, with no human at a screen to disintermediate [23]. RELX management is explicit about why the division is AI-proof: the underlying data is "incredibly difficult to replicate" [24], and the "generative AI tools actually do not add significant value to those kinds of mathematical calculations" the business runs on [25].

Wolters Kluwer discloses no comparable machine-to-machine metric, and it has no single division built the way Risk is. Its embedded-software tier plays a similar defensive role, but its flagship products — the ones that carry the brand and the pricing power — are human-facing lookups: a clinician queries UpToDate, a lawyer runs a search. That is structurally more exposed to "the professional asks a model instead" than a data feed the customer never sees. The comparison is not print versus print; it is a human-in-the-loop content base versus a machine-in-the-loop data utility. The ~90% machine-to-machine layer is precisely what Wolters Kluwer's revenue mix is missing, and it is a fair part of why the same GenAI wave to which RELX's Risk business is less exposed is priced as a heavier threat here.

The read

On the primary record, Wolters Kluwer's revenue is less substitutable by generative AI than the derating implies. The cleanly displaceable bucket — print — is 4.75% of revenue, matches RELX, and is falling. Roughly 55% of revenue is embedded software, registered-agent services and transactional work that a chatbot does not replace. The exposure is real but concentrated in the ~41% human-facing content layer, and that layer renews above 90% and is being converted from threat into an AI delivery channel.

The strongest fact against this read is structural, not cyclical: Wolters Kluwer has no equivalent of RELX's 90%-machine-to-machine Risk division — no part of the base that is immune rather than defended [26]. Its most human-facing divisions, Health and Legal & Regulatory, are also its most print-heavy and, per Peer Benchmark, its slowest growers. What would change the read is renewal: if AI substitution is going to appear, it appears first as slipping retention or organic growth in Health and Legal & Regulatory. Those two divisions — not the group headline — are where this thesis is falsified or confirmed.