RELX profit rises 9%, but shares remain 30% below their high as AI fears refuse to disappear

RELX plc (LSE: REL) ended August 14 at £25.26, up 1.4% on the session as software and data businesses benefited from a broader reassessment of artificial intelligence disruption risk. The rebound still leaves RELX approximately 30.5% below its current 52-week high of £36.34, an unusually large valuation reset for a company whose first-half underlying revenue increased 7%, adjusted operating profit grew 9% and adjusted earnings per share advanced 11% at constant currency. The disconnect has become one of the more interesting questions in the FTSE 100 because RELX is not merely defending itself against generative AI. It is embedding AI into legal research, scientific information and risk analytics while simultaneously expanding margins and returning £2.25 billion through its 2026 buyback programme.

The market is therefore confronting two competing interpretations of the same technology shift. One argues that increasingly capable AI agents will weaken established information platforms by allowing lawyers, researchers and corporate users to obtain answers without paying premium prices for traditional professional databases. The other is that AI makes proprietary, structured and trusted information more valuable because professional users require authoritative data rather than answers generated from uncontrolled internet content. RELX’s first-half financial performance currently supports the second interpretation more strongly than the first, but the share price indicates investors are still assigning substantial probability to disruption.

That makes the August 14 rebound more significant than an ordinary one-day move. European software and data shares strengthened after reported private-equity interest in Workday encouraged investors to reconsider whether public markets had pushed AI-related disruption discounts too far. RELX participated in that recovery, but its own investment case is less dependent on takeover speculation than on whether Chief Executive Officer Erik Engstrom can demonstrate that AI-enabled products generate faster sales growth, higher customer value and stronger margins over several years. Engstrom remains RELX chief executive, a position he has held since 2009.

Why does RELX’s 9% profit growth matter more when its shares are still down about 30% from their high?

RELX generated £4.87 billion of first-half revenue, compared with £4.74 billion a year earlier, with underlying growth of 7%. Adjusted operating profit increased from £1.65 billion to £1.73 billion, representing 9% underlying growth, while adjusted EPS rose from 63.5 pence to 68.6 pence and increased 11% at constant currency. Reported operating profit reached £1.59 billion and reported profit before tax increased to £1.52 billion from £1.28 billion.

The gap between underlying revenue growth of 7% and adjusted operating profit growth of 9% matters because it shows RELX continuing to generate operating leverage rather than spending increasingly large amounts simply to defend its position. Adjusted operating margin expanded to 35.5% from 34.8%, a 70-basis-point improvement, as the company kept cost growth below revenue growth. That margin movement is particularly important during an AI investment cycle because one of the major concerns surrounding incumbent information providers is that maintaining competitive products might require significantly higher development spending.

RELX is showing the opposite outcome so far. AI product development is occurring alongside higher margins and stronger per-share earnings rather than replacing profitability with an expensive defensive technology programme. This does not guarantee that competitive pressure will remain manageable, but it raises the burden of proof for investors arguing that the business model is already being structurally impaired.

The share price nevertheless tells a very different story. RELX closed August 14 at £25.26 compared with a rolling 52-week high of £36.34, leaving the stock 30.5% below that peak. It was also about 3% below its £26.03 close on August 6, although it remained approximately 3.4% above the 2,444-pence closing level recorded on July 14. The market has therefore stabilised somewhat over the past month without materially reversing the much larger AI-related derating.

Did investors overreact when Anthropic’s AI tools triggered the RELX sell-off?

The most dramatic change in sentiment occurred in February when Anthropic released new Claude Cowork capabilities and plug-ins capable of automating tasks in legal, sales, marketing and data-analysis workflows. RELX fell around 14% during the initial European software and data sell-off, while Thomson Reuters and Wolters Kluwer also suffered sharp declines as investors questioned whether AI agents could bypass established professional-information platforms. Reuters described the episode as part of a wider software sell-off that erased roughly $1 trillion of market value from U.S. software and services companies within a week.

The logic behind the sell-off was not irrational. If a law firm can use a general-purpose AI agent to search, summarise and draft legal material at dramatically lower cost, the number of traditional database seats required across the organisation could fall. Similar arguments apply to scientific research, business information and certain analytics workflows, meaning the historical model of charging large organisations according to user numbers could come under pressure.

The weakness in that argument is that professional workflows depend heavily on the quality and provenance of the underlying information. Legal research requires accurate case law, statutes, regulatory material and citation histories, while scientific and medical researchers depend on reliable scholarly content and structured datasets. Generative AI can provide a new interface to those information assets, but producing a fluent response is not equivalent to owning, validating and continuously updating the authoritative information needed to produce it.

Reuters Breakingviews argued during the February panic that the market appeared to be applying an unusually severe disruption scenario even though established providers possess proprietary datasets and are already integrating AI into their own products. That remains the central counterargument in August. RELX does not need to outperform general AI at generating language; it needs to make its controlled content and professional workflows sufficiently valuable that AI increases the usefulness of the subscription rather than eliminating it.

Is Lexis+ with Protégé proving that AI can accelerate Legal rather than cannibalise it?

RELX’s Legal division is one of the clearest testing grounds because legal research was among the professional workflows investors initially believed could be heavily disrupted. The company said in its April trading update that Law Firms & Corporate Legal was delivering double-digit growth, driven by adoption and expansion of Lexis+ with Protégé, its AI-enabled legal research platform with an integrated agentic assistant. Renewals and new sales were described as strong across all principal Legal segments.

That performance builds on a Legal business that generated £1.81 billion of revenue in 2025, with underlying growth of 9%, and £415 million of adjusted operating profit, which increased 12% underlying. RELX said at the first-half 2026 results that Legal had achieved a further step up in growth, although the abbreviated results release did not provide a specific divisional percentage.

The strategic importance lies in how Protégé changes the customer proposition. Traditional legal databases required lawyers to search documents, review results and construct an answer themselves. AI-enabled platforms can compress parts of that process by summarising authorities, generating drafts and assisting with research, which creates direct productivity benefits that law firms may be willing to pay for if accuracy and citations remain dependable.

There is also a potential pricing shift. If one AI-enabled lawyer can complete work that previously required several junior professionals, the customer may need fewer individual seats but could assign substantially greater value to each remaining licence. RELX therefore needs to evolve monetisation alongside the technology rather than assume the historical seat-based model remains unchanged indefinitely.

The strongest evidence would be continued double-digit Legal growth accompanied by high renewal rates and margin expansion. If those conditions persist while generative AI adoption accelerates across the legal profession, it becomes increasingly difficult to argue that AI is simply cannibalising LexisNexis rather than expanding what customers can accomplish with it.

Why could RELX’s proprietary data become more valuable as general-purpose AI becomes cheaper?

The economics of generative AI are moving toward abundant processing and increasingly capable models, which means access to reliable differentiated information may become a larger part of the competitive advantage. RELX’s Risk business illustrates this dynamic because its products combine public information, industry-specific datasets and proprietary analytics to help customers manage fraud, identity verification, insurance risk and financial-crime compliance. Risk generated £3.49 billion of 2025 revenue and £1.31 billion of adjusted operating profit, with underlying growth of 8% and 10% respectively.

RELX said in April that Risk continued to achieve strong growth through AI-enabled analytics and decision tools, with Financial Crime Compliance and digital Fraud & Identity solutions among the principal drivers. Insurance products were also benefiting from contributory databases and market-specific solutions, while management expected underlying adjusted operating profit growth to exceed revenue growth for the full year.

This type of business is harder to replicate through a generic language model because the value does not come primarily from generating text. It comes from combining identities, histories, transactions and risk indicators into a structured decision that a bank, insurer or government organisation can use operationally. The model may improve how that information is analysed, but the underlying data network remains critical.

The same logic applies across much of RELX. In Scientific, Technical & Medical, the group controls major scholarly publishing and research-information assets, while in Legal it combines authoritative legal content with analytics and workflow tools. AI can lower the cost of interacting with those datasets, but that does not automatically lower the value of having access to them.

This is why the long-term investment debate should focus less on whether AI can answer professional questions and more on whether RELX retains control of sufficiently valuable information and workflow integration to charge for those answers. If the answer remains yes, cheaper and more capable AI could expand product functionality while RELX captures part of the productivity benefit.

Can Scientific, Technical & Medical turn AI into a second acceleration engine?

Scientific, Technical & Medical is another division where management reported improving momentum. In April, RELX said underlying revenue growth was improving as the business shifted toward higher-value analytics and tools, while databases and electronic reference products were benefiting from increased adoption and usage of AI-enabled products. Primary Research was also benefiting from strong growth in article submissions.

The division generated £2.71 billion of revenue and £1.04 billion of adjusted operating profit in 2025, with underlying growth of 5% and 7% respectively. At the 2026 half-year stage, Engstrom described STM as having achieved a step up in growth, suggesting the division is moving beyond the slower growth profile it historically carried compared with Risk and Legal.

AI provides several commercial opportunities within scientific workflows. Researchers must identify relevant papers, understand rapidly expanding fields, compare findings, organise references and assess the significance of new studies. Search and summarisation tools can reduce the time required for those activities, but they become substantially more useful when connected directly to high-quality scientific literature and metadata.

The risk is that publishers face increasing scrutiny around research access, subscription pricing and the economics of academic publishing independent of AI. RELX therefore needs technological productivity gains to create additional customer value rather than simply reinforce the economics of existing publishing models. Stronger growth in analytics and decision tools would provide the clearest evidence that the division is evolving beyond conventional content distribution.

Is the £2.25 billion RELX buyback creating value after the share-price collapse?

RELX plans to repurchase £2.25 billion of shares during 2026. It had completed £1.75 billion during the first half, another £100 million had been deployed after July 1 and the remaining £400 million was scheduled to be completed before year-end. The company also increased its interim dividend 7% to 20.9 pence per share.

The timing is notable because much of the programme is being executed after a substantial valuation decline. Buybacks generally create greater per-share value when a company purchases its own stock at lower valuations, assuming the underlying earnings power has not deteriorated proportionately. RELX’s adjusted EPS is currently increasing 11% at constant currency while the shares remain roughly 30% below the 52-week high, creating a very different capital-allocation proposition from repurchasing shares at the previous peak valuation.

There is a balance-sheet trade-off. Net debt increased to £8.73 billion at June 30 from £7.44 billion in the comparison presented by RELX, an increase of roughly £1.29 billion or 17%. Net debt to EBITDA consequently moved to 2.3 times from 2.2 times, while adjusted cash-flow conversion remained very strong at 98%.

That leverage level does not indicate immediate financial stress, but it means the buyback cannot be evaluated in isolation. RELX is choosing to return substantial capital while also funding acquisitions, AI development and dividends, so management must maintain high cash conversion if it wants to continue balancing shareholder distributions with investment.

The first-half numbers currently support that approach. Two acquisitions required only £103 million of consideration, while strong operating profitability and cash conversion provided substantial internal funding. The more difficult capital-allocation question would emerge if RELX identified larger data or AI assets capable of accelerating growth while the company was simultaneously committed to aggressive repurchases.

Why has RELX’s valuation not recovered even after strong first-half earnings?

One explanation is that investors are applying a lower terminal value rather than questioning current earnings. RELX can deliver another year of 7% revenue growth and 9% profit growth while the market simultaneously assumes those rates will slow materially later if AI weakens customer dependence on existing products. A share price can therefore fall substantially even when near-term earnings continue increasing.

This helps explain why the July 23 first-half results did not immediately eliminate the discount. The stock closed around 2,451 pence on the results date before rising to 2,568 pence the following session, showing that investors initially responded positively to the operating evidence. By August 14, however, RELX was at £25.26, little changed from those late-July levels and still far below its previous peak.

The current valuation debate is consequently less about whether RELX can meet its 2026 guidance and more about what growth looks like from 2028 onward. The company has reaffirmed expectations for another year of strong underlying revenue and adjusted operating profit growth, with strong constant-currency adjusted EPS growth. Those targets matter, but a full rerating requires investors to believe the business can maintain a superior growth trajectory after generative AI becomes embedded throughout professional workflows.

The October 22 nine-month trading update will provide the next important operating checkpoint. Continued strong Legal and Risk growth, further acceleration in STM and evidence that profit is still outgrowing revenue would strengthen the argument that the AI disruption discount has become excessive.

Key takeaways from RELX’s AI strategy, first-half results and 30% share-price discount

• RELX reported first-half revenue of £4.87 billion, with underlying growth of 7%.
• Adjusted operating profit increased 9% underlying to £1.73 billion, growing faster than revenue.
• Adjusted operating margin expanded 70 basis points to 35.5%, despite continued investment in AI-enabled products.
• Adjusted EPS reached 68.6 pence, representing 11% constant-currency growth, while reported EPS rose to 65.7 pence from 52.9 pence.
• Legal continues to benefit from Lexis+ with Protégé, with RELX previously reporting double-digit growth across Law Firms & Corporate Legal.
• Risk remains RELX’s largest operating business and continues to benefit from financial-crime, fraud, identity and insurance analytics powered by proprietary datasets.
• RELX plans £2.25 billion of buybacks during 2026, of which £1.85 billion had been completed by the period shortly after June 30.
• Net debt increased to £8.73 billion and net debt to EBITDA moved to 2.3 times, while cash conversion remained strong at 98%.
• RELX shares closed August 14 at £25.26, approximately 30.5% below their rolling 52-week high of £36.34 despite continued earnings growth.
• The October 22 trading update is the next major test of whether AI-enabled product adoption can continue translating into stronger Legal, Risk and STM growth.

Is RELX being disrupted by AI, or is the market undervaluing the data that AI increasingly needs?

The most important finding in RELX’s 2026 story is not that artificial intelligence presents no threat. AI is already changing how lawyers, researchers and corporate users search for information, complete analytical tasks and produce professional work, and those changes could eventually reduce demand for some traditional interfaces or seat-based subscriptions. The market is therefore justified in reassessing what customers will pay for information services when intelligent agents can perform more work automatically.

What is harder to reconcile is the scale of the valuation decline with the operating evidence available so far. RELX is generating 7% underlying revenue growth, 9% adjusted operating profit growth and 11% constant-currency adjusted EPS growth while expanding margins to 35.5%. Legal is accelerating as customers adopt AI-enabled tools, Risk continues to grow strongly and Scientific, Technical & Medical is improving rather than contracting.

That suggests the competitive advantage may be migrating rather than disappearing. The interface through which a lawyer or researcher accesses information could become increasingly conversational and agentic, but RELX still owns large quantities of specialised content, structured data, workflow knowledge and customer relationships. If general-purpose AI becomes commoditised, trusted proprietary information may become one of the scarcer components of the professional AI stack.

The next stage of the argument must be proven through commercial metrics rather than technological demonstrations. RELX needs AI-enabled products to sustain stronger new sales, renewals and usage while profit continues outgrowing revenue. If that happens through 2027 and beyond, a share price approximately 30% below the 52-week high could increasingly look like an unusually severe response to a technology transition the company is already monetising. If growth slows sharply despite widespread AI adoption, the market’s discount will have anticipated a structural change that the first-half accounts simply had not yet captured.