Open a modern sportsbook app on a Tuesday night and try to count all the ways you could spend money on a single mid-table football fixture. You'll likely give up somewhere past two hundred. Corners in the first 15 minutes. Throw-ins before half-time. Cards, carries, crosses, next goalscorer, both-halves-both-teams-to-score…
A decade ago, the same match might have carried 30 markets, maybe 50 if the trader was feeling ambitious. Today, some major matches carry upwards of 1,500.
For most of gambling history, the market ceiling existed for a simple reason: a human being had to build the market. Someone had to price it, monitor it and manage the liability once punters started betting on it. Markets were finite because trading desks were finite.
That constraint is dissolving.
AI, automation and better infrastructure mean that almost anything measurable can now become a betting market – and the industry's defining question has somewhat flipped. It used to be what can we offer? Increasingly, it's: what shouldn't we?
The menu that ate itself
The raw numbers tell the story better than any trend report. BetConstruct's sportsbook platform now pushes out more than 140,000 pre-match events and over 90,000 live matches every month.
Genius Sports has said plainly that its industry ingests hundreds of thousands of events a year – a volume no trading floor could ever price by hand.
Micro-market engines built for tennis reportedly generate roughly 1,500 fresh betting opportunities per match, recalculating prices every few hundred milliseconds as the ball crosses the net.
This isn't just "more of the same, but faster." It's a structural shift in what a betting market even is.
Match odds used to be the product. Now they're a one-line item on a menu – a menu so long that discovering the market you actually want has become its own skill.
What's significantly more scarce now is judgment: knowing which of the infinite available markets are actually worth offering, to whom, and when
The machine behind the menu
None of this would be commercially possible without AI taking over the trading desk. Sportradar's Alpha Odds tool, in use across 60 of the company's sportsbook clients, reportedly delivered an average 10% profit uplift in its first full year – powered by machine-learning models trained on tens of millions of daily transactions.
Genius Sports launched a rival pricing engine, Edge, with its product director candidly noting that the sheer scale of modern betting content makes it near-impossible for human traders to optimise pricing manually across every market on every match.
That's the real unlock. AI hasn't just made existing markets more accurate – it’s made the marginal market nearly free to launch. Generating a price, monitoring it for anomalies, suspending it if something looks wrong: all of this can now happen without a person in the loop until something breaks.
When the cost of adding a market approaches zero, the incentive to keep adding them approaches infinity. Restraint has to be a deliberate choice now, not a side-effect of limited resources.
When betting stops looking like betting
If sportsbooks are expanding sideways, prediction markets are expanding the whole category outward in every direction at once. Combined monthly trading volume on Kalshi and Polymarket quadrupled from under $5bn in September 2025 to roughly $24bn by April 2026, and hit an estimated $44.8bn in June – more than triple the average monthly handle of the entire regulated US sportsbook industry.
Crucially, sport is the fuel: sports contracts have made up roughly 80% of Kalshi's volume since mid-2024 – though dressed in the language of finance rather than betting.
Unsurprisingly, regulators have noticed.
In June 2026, the CFTC published a proposed rule that would generally permit sports event contracts, on the logic that they aid price discovery, while potentially banning contracts most vulnerable to manipulation – individual player injuries, referee decisions, single in-game moments.
It's a telling distinction: regulators are drawing exactly the line the industry itself keeps blurring, between a market that reflects a real outcome and a market that exists purely because it could.
Meanwhile, the infrastructure is already merging. On 8 June 2026, Sportradar signed a multi-year deal to become Kalshi's official data and solutions provider, supplying the same league data, live odds and integrity-monitoring tools it already sells to licensed sportsbooks – covering MLB, NHL, MLS and UFC. The arrangement arrived just two months after MLB struck its own exclusive data deal with Polymarket.
In other words, the two ecosystems aren't just adjacent anymore; they're even sharing suppliers.
A sportsbook of one
Layer personalisation on top and the sportsbook stops looking like a catalogue at all. Operators are increasingly building recommendation engines trained on session behaviour and bet history that decide, in real time, which of those thousands of markets a given customer actually sees first.
BetConstruct puts it bluntly in its own trend analysis: generic interfaces stuffed with endless markets and static promotions no longer drive engagement the way a tailored feed does. Two customers opening the same app on the same night may now see almost entirely different home screens – one built around player props for their favourite NBA team, the other around Champions League handicaps and crypto-price contracts.
This is a genuine reversal of the industry's operating logic. The catalogue used to be fixed and the customer had to search it. Now the catalogue is nominally infinite, and the product's real job is deciding what to hide.
The case for limits
However, all this is well and good, but there's a tension nobody in the "more markets" camp likes to sit with: proliferation has a body count.
The past couple of years have delivered a rather uncomfortable run of scandals directly tied to granular, low-liquidity markets – from the Jontay Porter lifetime NBA ban to the November 2025 federal indictment of two MLB pitchers over alleged pitch-level manipulation (which actually pushed the league to cap pitch-specific prop bets at $200 and strip them from parlays).
New York's state senate has since proposed a task force specifically to study the risks of granular prop markets.
This isn't just "more of the same, but faster." It's a structural shift in what a betting market even is
The issue is, these scandals trace back to the same root cause the industry usually frames as progress: a market thin enough, specific enough, and low-profile enough that a single person could move it.
There's a commercial argument for restraint too, not just a regulatory one. Behavioural research on choice overload – the "paradox of choice" popularised by psychologist Barry Schwartz – has long shown that, past a certain point, more options don't drive more engagement; they drive hesitation and abandonment.
A sportsbook with 1,500 markets on one match isn't necessarily more compelling than one with 150 well-chosen ones – it's just harder to use, unless the personalisation layer is doing some serious work to hide the other 1,350.
Why curation is the new creation
None of this points toward every conceivable event eventually becoming a betting market – integrity risk, regulatory attention and basic usability will keep pulling in the other direction. But the industry's centre of gravity is clearly shifting anyway.
For many years, competitive advantage in this business came from creation; who could build the market first, price it fastest, and cover the most obscure corner of the fixture list.
That skill hasn't disappeared, but it's no longer scarce – AI has made it available to anyone with the right data feed.
What's significantly more scarce now is judgment: knowing which of the infinite available markets are actually worth offering, to whom, and when.
The operators who find success in this new era likely won’t be the ones with the longest market list; they’ll be the ones who embrace the idea that, in a world of infinite supply, the product isn’t really the market anymore – it’s the filter.
Sportsbooks' use of AI has been called into question of late, with Massachusetts now investigating reports that DraftKings utilised AI to attract 'losing' customers
