After Law and Medicine, Vertical AI Has Found Its Next Billion-Dollar Market

Investors have poured more than $2.2 billion into sector-specific AI search engines for legal and medical professionals. The structural thesis behind those bets now points squarely at the online gambling industry – and the first serious evidence that the disruption is already underway has emerged from within the European regulated iGaming ecosystem, backed by a publicly listed parent company, not Silicon Valley.

 

The Pattern That Preceded the Money

Three years ago, a former securities litigator named Winston Weinberg sent a cold email to Sam Altman and described an idea: what if large language models could be trained, not on the open internet, but exclusively on the structured knowledge of a single profession? What if they could answer the kind of specialised, high-stakes questions that professionals faced every day, faster than any human researcher, and with reliable citations from authoritative sources?

The result was Harvey, a legal AI platform that today counts more than 50 of the top 100 American law firms as clients, employs over 100,000 lawyers across 42 countries, and has raised more than $806 million from Sequoia Capital, Kleiner Perkins, Google Ventures, and Andreessen Horowitz. Its annual recurring revenue reached $195 million by the end of 2025, up from $50 million a year earlier – a 3.9-fold increase in twelve months, according to data from Sacra. In February 2026, Harvey entered discussions to raise a further $200 million at an $11 billion valuation, according to sources familiar with the deal reported by TechCrunch.

Concurrent with Harvey’s ascent, a parallel story was unfolding in medicine. OpenEvidence, founded by Daniel Nadler, trained its AI not on general internet data but exclusively on peer-reviewed medical journals, including formal content partnerships with the New England Journal of Medicine and the American Medical Association. It raised approximately $700 million across four funding rounds in under twelve months, its valuation climbing from $1 billion in February 2025 to $12 billion by January 2026 – a twelvefold increase in less than a year. By December 2025, the platform was supporting 18 million clinical consultations per month from verified US physicians, up from roughly three million a year prior. It reached $100 million in annual revenue faster than almost any company in history.

Then there is Glean, which applied the same principle to enterprise search. Rather than allowing employees to search the entire web, Glean built an AI layer that indexes and understands only the proprietary knowledge contained within a company’s own systems – its Slack conversations, Salesforce records, Google Workspace documents, internal wikis. It raised $765 million and reached a $7.2 billion valuation in June 2025, crossing $208 million in ARR by end of year.

The investment thesis connecting all three is not simply that AI is useful. It is something more specific: that there are large industries defined by enormous volumes of highly structured, trust-sensitive information, where the existing tools for finding and navigating that information are chronically inadequate – and where a purpose-built AI that masters the domain’s specific knowledge will attract extraordinary investor conviction and user loyalty. The corollary is obvious. Ask which large industry fits that description next. The data has a clear answer.

 

The investment thesis is not simply that AI is useful. It is that large industries with enormous volumes of trust-sensitive information, served by inadequate information tools, are systematically vulnerable to vertical AI disruption.

 

A $95 Billion Market Built on Inadequate Information

The global online gambling market generated between $78 billion and $95 billion in gross gaming revenue in 2024, depending on methodology – Grand View Research places the figure at $78.66 billion while IMARC Group estimates $93 billion. Both project the market reaching somewhere between $153 billion and $173 billion by 2030, at compound annual growth rates of between 7 and 12 percent. Europe is the largest single geography, accounting for approximately 41 percent of global online gambling revenue in 2024, according to Grand View Research.

But the headline market size understates the structural opportunity for AI. What matters more to the vertical search thesis is how players currently navigate the market – and the economics that flow from that navigation.

Online casino players face an information problem of extraordinary complexity. In a typical regulated European market, hundreds of licensed operators offer overlapping products: deposit bonuses, free spin packages, loyalty programmes, live dealer tables, and thousands of slot titles from dozens of software providers. The terms attached to those products – wagering requirements, game contribution rates, expiry windows, maximum bet restrictions during bonus play – are buried in small-print that requires sustained legal literacy to decode. Comparing across operators is a multi-hour exercise even for experienced players.

The infrastructure built to solve this problem is the affiliate marketing industry. In iGaming, the affiliate channel accounts for an estimated 22 percent of all player acquisition, according to SOFTSWISS’s iGaming Trends 2025 report, with influencer collaborations adding a further 38 percent. That is approximately 60 percent of all player acquisition flowing through what are, in essence, third-party information intermediaries. The global affiliate marketing industry was valued at $18.5 billion in 2024, per Cognitive Market Research, with the iGaming vertical representing one of its most lucrative segments – affiliate commissions in online gambling typically run between $100 and $250 per acquired player, paid on a cost-per-acquisition basis.

 

Company	Vertical	Total Raised	Peak Valuation	ARR (2025)
Harvey	Legal AI	$806M+	$8B (Dec 2025)	$195M
OpenEvidence	Medical AI	~$700M	$12B (Jan 2026)	$100M+
Glean	Enterprise Search	$765M+	$7.2B (Jun 2025)	$208M
marvn.ai	iGaming AI	Undisclosed	Parent co. listed (MSE)*	Early stage

*marvn.ai is developed and operated by Marlin Media Ltd. Angler Gaming plc, listed on the Swedish stock exchange, 

The problem is that affiliate economics are not aligned with player interests. Rankings on comparison sites are heavily influenced by operator commission rates. Welcome bonus offers – often the primary navigation tool players use to choose between casinos – are frequently optimised to be misleading rather than transparent, front-loading headline figures while burying the wagering conditions that determine whether the bonus has any realisable value. A player searching for the best casino bonus is navigating a landscape that has been commercially engineered to obscure the information they actually need.

This is precisely the kind of market dislocation that has preceded the emergence of every successful vertical AI product. Legal professionals faced it when searching for case law and regulatory guidance through systems that were built for libraries, not workflows.

 Physicians faced it when trying to stay current with 18 million new journal articles published every year, in a system that offered no synthesis layer between the primary literature and the clinical decision. In both cases, the information asymmetry was exploited commercially by intermediaries – legal publishers, medical database vendors – rather than resolved in the user’s interest.

 

In legal and medical AI, the information asymmetry was exploited commercially by incumbents rather than resolved in the user’s interest. The online gambling affiliate ecosystem operates on exactly the same structural dynamic.

 

The Regulatory Tailwind Accelerating Disruption

The analogy between legal, medical, and iGaming AI is structural but also temporal: all three sectors are experiencing simultaneously tightening regulatory environments that increase the cost of the old model while creating new demand for compliant alternatives.

European regulators have spent 2024 and 2025 systematically dismantling the affiliate marketing infrastructure that the iGaming industry depends on for player acquisition. Belgium implemented a broad ban on gambling advertising in 2023. Lithuania adopted a general ban on gambling advertising effective July 2025. The Netherlands requires affiliates to prove that 95 percent of their audience is aged 24 or over before any promotion can run. Spain restricted gambling advertising to the hours between 1am and 5am and prohibited any promotion of bonuses in public-facing channels. Italy introduced a new licensing regime in 2024 with the highest concession fees in the EU, while simultaneously deploying AI-driven compliance monitoring through its Customs and Monopolies Agency.

In November 2025, regulators from Austria, France, Germany, Great Britain, Italy, Portugal, and Spain issued a joint statement warning specifically about the proliferation of gambling advertising through affiliate networks, calling on digital platforms to strengthen enforcement mechanisms. The scale of the problem they were responding to was quantified in the same month by a Yield Security report commissioned by the European Casino Association, which found that illegal online gambling operators generated €80.6 billion in the EU in 2024 – representing 71 percent of the total EU gambling market – with the targeting of European citizens by illegal operators increasing 26 percent year-on-year, and 92 percent of online gambling content viewed by EU citizens promoting unlicensed operations.

For compliant, licensed operators, the regulatory tightening creates a distribution problem: the marketing channels that have historically delivered player acquisition at scale are being progressively closed. The affiliate ecosystem, already under pressure from algorithm changes affecting SEO-driven comparison sites, is being further squeezed by advertising restrictions that reduce the surfaces on which affiliate content can legally appear. An AI-powered discovery tool that delivers players to regulated operators through a search-and-recommendation mechanism rather than through traditional advertising falls entirely outside these restrictions – and in markets where aggressive bonus promotion is banned, a platform that helps players find the best available regulated products through transparent comparison is not just commercially attractive, it is the only compliant channel available.

 

The First Mover: European, Listed, and Operator-Built

Against this backdrop, the launch of marvn.ai in November 2025 takes on significance beyond its scale. The platform – an AI-powered conversational search engine for iGaming products – was built not by a Silicon Valley startup racing to raise a Series A, but by Marlin Media Ltd., which developed and operates marvn.ai. Marlin Media is owned by Angler Gaming plc, a company listed on the Swedish stock exchange. That corporate structure matters for several reasons.

First, it signals that the disruption of iGaming discovery is being driven from within the regulated European operator ecosystem rather than imposed on it from outside. Marlin Media’s existing infrastructure – including partnerships with more than 500 iGaming brands – gave it the data relationships to build a product with genuine breadth at launch. 

Where OpenEvidence’s early moat was its formal content partnerships with the New England Journal of Medicine and the American Medical Association, marvn.ai’s equivalent is Marlin Media’s existing operator partner network, which provides the structured, verified data on casino products, bonuses, payment methods, and licensing that makes its search results meaningfully different from a keyword-matched comparison table.

Second, Marlin Media’s partner network is built exclusively around licensed operators – the 500-plus brands in its database hold licences from the MGA, UKGC, and other national authorities. 

Ionut Constantinescu, CEO of Marlin Media, described the product’s proposition at launch: “With marvn, we aim to offer a more accurate, faster, and more safe way for players to discover casinos, games and bonuses online.” 

That framing – more accurate, faster, safer – maps almost precisely onto the value propositions of Harvey (more accurate than traditional legal research), OpenEvidence (faster than nine hours of daily journal reading), and Glean (smarter than keyword search across enterprise data).

In January 2026, the platform extended its capabilities with the addition of a Discover section: a knowledge and news hub that scans reputable sources to generate topic overviews and news summaries for the iGaming space, with integrated follow-up query capability. The architecture mirrors what OpenEvidence built with DeepConsult – a product that combines search with synthesis, allowing users to move from discovery to understanding within a single interface rather than requiring them to navigate between a search result and external sources.

 

Where OpenEvidence’s early moat was content partnerships with the New England Journal of Medicine, marvn.ai’s equivalent is Marlin Media’s operator data relationships providing verified, structured information on casino products, bonuses, and licensing across more than 500 brands.

 

The Investor Thesis: Why iGaming Fits the Vertical AI Playbook

For investors tracking the vertical AI sector, the case for iGaming as the next major disruption opportunity rests on the same four structural criteria that identified legal and medical AI before capital flooded into them.

High-stakes, complex decisions: Casino bonus selection is not trivially simple. A player choosing between two welcome packages with different base amounts, different wagering multiples, different eligible game lists, and different maximum win caps is making a financial decision that requires structured analysis to evaluate correctly. This is the same structural pressure that made Harvey’s legal document analysis and OpenEvidence’s clinical decision support valuable: the user needs the right answer, not just a fast one.

 

Trust-sensitive, source-specific data: iGaming search requires not just general web knowledge but specific, structured, frequently updated information about licensed operators, current promotional terms, payment infrastructure, and regulatory status. This data does not exist in a form usable by a general-purpose AI model; it requires proprietary curation. OpenEvidence’s key insight was that training on medical journals rather than the internet produced answers physicians could trust. marvn’s equivalent is a database built on operator partnerships rather than web scraping.

 

Incumbent infrastructure misaligned with users: The affiliate comparison site model is, as described above, commercially misaligned with player interests. Players navigating bonus offers through traditional comparison sites are not receiving unbiased information; they are receiving commercially curated information. This is the same dynamic that made Glean compelling in enterprise search – employees searching internal systems through keyword tools were not getting useful answers, they were getting results shaped by whatever had the most accessible metadata.

 

Regulatory pressure accelerating the transition: The progressive tightening of gambling advertising regulations across the EU is systematically closing the channels through which the incumbent affiliate ecosystem operates. This is not a headwind for AI-powered discovery; it is a structural accelerant. A player-facing search tool that operates outside the advertising model is not subject to the restrictions being imposed on promotional content.

 

Where the Analogy Breaks – and Why That May Not Matter

The vertical AI comparison is not seamless. Harvey and OpenEvidence serve professionals – lawyers and physicians – whose willingness to pay for premium tools is high and whose institutional affiliations provide built-in distribution channels. The iGaming user base is fundamentally consumer-facing, which creates different acquisition dynamics, lower individual willingness to pay, and higher sensitivity to the quality of the user experience.

The monetisation model is also different. OpenEvidence chose an advertising-supported model – free to verified physicians, with casino-style video advertising from pharmaceutical and medical device companies providing revenue. Harvey chose enterprise subscription pricing at the firm level. marvn.ai, operating in a market where advertising restrictions are actively tightening, will need to navigate carefully between player-facing utility and operator-facing commercial relationships. The data economics of an iGaming AI platform – where the value of a conversion is measurable in real time through affiliate tracking – may actually make this more tractable than it was for OpenEvidence’s medical advertising model.

There is also a question of data moat sustainability. Harvey’s competitive advantage deepens with every law firm onboarded, because each new engagement provides more training data on how real legal professionals use AI. OpenEvidence’s moat compounds with each new content partnership with a medical journal or clinical guidelines body. 

For marvn, the equivalent dynamic is operator data – but operators in the iGaming space are notoriously reluctant to share structured data with third parties whose commercial interests may not align with theirs. Building and maintaining 500-plus operator partnerships while preserving the independence of the search result is a long-term governance challenge that Harvey and OpenEvidence, serving institutional rather than commercial partners, have not had to manage in the same way.

These are real tensions. But they are the kind of tensions that define first-mover advantage rather than disqualifying it. Harvey operates in a legal market where multiple large incumbents – LexisNexis, Thomson Reuters, various BigTech players – have both the resources and the motivation to compete. OpenEvidence operates in healthcare, where regulatory complexity and liability considerations make building a trusted product genuinely hard. The barriers to entry in iGaming AI search are different, but they are not trivially lower.

 

A European Story in a Field Dominated by American Capital

One of the more notable dimensions of the iGaming AI thesis is its geography. Harvey was founded in San Francisco and funded by Sequoia, Andreessen Horowitz, and Kleiner Perkins – the canonical Silicon Valley investor list. OpenEvidence began in Cambridge, Massachusetts, and moved to Miami. Glean was founded in Palo Alto.

marvn.ai is a product of the European regulated iGaming ecosystem: developed and operated by Marlin Media Ltd. in Malta. It was designed from the ground up for markets that are already living with the regulatory environment that will, over the next five years, reach markets that are currently in earlier stages of the same regulatory cycle. That European-first design is not a limitation. It is a market intelligence advantage.

The EU’s patchwork of national gambling regulations – each with different licensing requirements, advertising restrictions, bonus promotion rules, and player protection obligations – creates exactly the kind of structured data complexity that an AI platform built natively within the system is better positioned to navigate than one designed for a single-market or less-regulated environment. 

For European investors and strategic observers, the vertical AI story in iGaming carries an additional dimension. The legal and medical AI booms were primarily funded by and for the American market. 

The iGaming vertical AI opportunity is, by geography and regulation, inherently European in its earliest and most developed form. That does not guarantee that the capital will flow from European sources – the global VC community is alert to the thesis at this point – but it does mean that the companies best positioned to capture the early data advantage are operating out of the continent rather than into it.

What Comes Next

The pattern in vertical AI has been consistent: a credible first product demonstrates that the technical problem is soluble, attracts a critical mass of users who generate proprietary data, and then either scales independently or becomes an acquisition target for a larger platform seeking domain-specific capability. Harvey went from a seed round to an $11 billion reported valuation in three years. OpenEvidence went from zero to $12 billion in four. Glean has nearly tripled in valuation in eighteen months.

The iGaming vertical is earlier in that cycle. The market size is demonstrably large – $78 to $95 billion in annual gross gaming revenue, growing at double digits annually, with Europe accounting for 41 percent of the global share. The structural problem with incumbent information infrastructure is well-documented. The regulatory pressure accelerating the transition is already in place and intensifying. The first platform with genuine operator data relationships and a working conversational search product has launched and is operational.

Whether marvn.ai will be the Harvey of iGaming or will be acquired by a Harvey-scale company that emerges from this space over the next three to five years is not a question this article can answer. What the evidence from legal AI, medical AI, and enterprise search strongly suggests is that the thesis – that a purpose-built AI platform, trained on trusted domain-specific data, and aligned with user interests rather than commercial intermediary interests, can displace the incumbent comparison infrastructure in a large professional or consumer vertical – has now been validated at scale in multiple sectors.

The online gambling industry is a large, complex, data-rich, regulation-heavy, user-interest-misaligned market dominated by legacy information infrastructure. It is, in other words, exactly the kind of market that vertical AI has proven it can transform.

The question for investors and strategists is not whether this disruption will happen. It is whether they are positioned to notice it before the next round of $300 million funding announcements makes the opportunity considerably more expensive to enter.

 

Key data sources and references

Harvey funding and ARR data: Sacra (sacra.com/c/harvey), TechCrunch (December 2025, February 2026). OpenEvidence funding: Business Wire (January 2026), CNBC (January 2026), OpenEvidence press releases (February, July, October 2025). Glean funding: Glean press release (June 2025), CNBC (June 2025), Crunchbase. Global online gambling market size: Grand View Research (2025), IMARC Group (2025), Statista Market Forecast (2025). European market share: Grand View Research (2024 data). iGaming affiliate acquisition share: SOFTSWISS iGaming Trends 2025 report. Global affiliate marketing market size: Cognitive Market Research (2024), Hostinger/Statista (2025). EU illegal gambling data: Yield Security / European Casino Association report (November 2025), as reported by DLA Piper (December 2025). EU regulatory developments: iGaming.com (August 2025), SOFTSWISS (September 2025), affnook.com (November 2025). marvn.ai product and corporate information: Marlin Media Ltd. / Angler Gaming plc announcements (November 2025, January 2026).