Three years ago, the trading desk was the operational heart of every serious sportsbook. Experienced traders compiled lines, managed liability, adjusted odds in real time, and formed the institutional knowledge that separated a sharp book from a recreational one. Today, that model is being systematically replaced—not by cost-cutting, but by mathematics.
The sheer combinatorial complexity of modern betting markets has outpaced what any human team can price accurately at scale. The evidence is now documented in vendor financials and operator disclosures. This article examines what drove the inflection, what AI pricing delivers in measurable P&L terms, and what the competitive landscape looks like for operators who have yet to make the transition.
The ShiftFrom 4% to 48%: Three Years That Rewired the Trading Desk
The headline number from Kambi’s 2025 disclosures is striking: 48% of all bets across its 50+ operator network were priced by AI in 2025, up from 28% in 2023 and just 4% in 2022. That is a 12x increase over three years across a major B2B network serving BetMGM, Rush Street Interactive, LeoVegas, Bally’s, and others.
No other operational shift in sportsbook history has moved this fast. In the same period, the industry has debated responsible gambling frameworks, explored new state markets, and navigated a media rights arms race. None of those trends show the same rate of structural displacement as AI pricing. In 2022, AI pricing was a pilot program at most operators. In 2025, it handles nearly half of all volume across one of the industry’s largest B2B networks—and that network is not an outlier.
Kambi also quantified the financial impact directly: €7.4 million in incremental margin gains were attributed to expanded AI pricing in 2025. That figure represents the P&L delta between manual and automated pricing at current network volumes—real money, on the operator side of the ledger, not projected efficiency savings.
The adoption curve also signals something important about velocity. Going from 28% to 48% in a single year suggests the easy markets—pre-match singles on major football—were priced by AI first, and the technology is now moving into more complex territory: player props, bet builders, in-play markets, and niche sports. The hard problems are being solved in production.
Root CauseThe Combinatorial Problem: Why Bet Builders Broke Manual Pricing
To understand why AI adoption accelerated so rapidly, you have to understand what happened to the product itself. Modern sportsbook markets are not more complicated versions of the markets that existed five years ago. They are structurally different in ways that make manual pricing mathematically untenable at scale.
The data from Kambi’s 2025 trends report is illustrative. In 2025, 88% of Super Bowl Bet Builder bets included at least one player prop—up from 72% in 2022. For Champions League pre-match markets, Bet Builder share of all bets hit 24% in 2025, compared to just 8% in 2020.
These numbers represent a structural shift in how bettors engage with the product. Bet Builders are not a feature—they have become the dominant bet type for high-value customers on premium fixtures. And each Bet Builder with multiple player props creates a combinatorial pricing challenge that scales exponentially with the number of legs.
Consider a five-leg Bet Builder combining match result, both teams to score, first goalscorer, total corners over/under, and a player booking. Each leg has its own probability. The correlations between legs—especially player-level outcomes and match outcomes—are non-trivial to model and change dynamically as team news, weather, and tactical information emerges. An experienced trader can price a single such bet. They cannot price 50,000 of them across an afternoon of fixtures with consistent accuracy and speed.
Market complexity growth is the structural driver of AI adoption, not cost-cutting. The operators who understood this earliest moved fastest, and their margin data now proves the advantage was real.
What AI Pricing Actually Delivers on the P&L
The most important development in the AI trading narrative over the past 12 months is that financial impact is now documented, not projected. Vendors and operators are publishing specific numbers, and those numbers are large enough to reframe the build-vs-buy decision entirely.
Genius Sports’ fully managed AI trading model—Genius Sports Edge—reported a 22% margin improvement on football markets in the 2025/26 season through liability-driven pricing adjustments and automated line movement. This is a documented P&L outcome, delivered via a subscription model that requires no in-house trading infrastructure from the operator.
The margin mechanics are multi-dimensional. AI pricing delivers improvement on the P&L through several distinct mechanisms:
| Mechanism | How it works | Estimated P&L contribution |
|---|---|---|
| Margin optimization by market tier | Selective margin allocation by sport, league, market type—configurable in real time vs. static manual tables | Primary driver of 22% improvement |
| AI bet scoring & fraud detection | Automated pattern detection identifies sharp and syndicate activity previously requiring manual review | ~1% GGR protection |
| Coverage scale | Pricing markets that manual desks would leave uncovered or price poorly due to capacity constraints | Handle uplift on niche markets |
| In-play latency elimination | Millisecond repricing on game events vs. manual suspension windows | Margin protection on live volume |
The coverage scale point deserves emphasis. Genius Sports Edge automates pricing across 600,000+ annual fixtures across 40+ sports. No in-house trading team can replicate this volume cost-effectively. The alternative is not worse AI coverage—it is no coverage, which means either refusing to offer markets or offering them at margins that do not reflect actual risk.
The ~1% GGR protection from AI bet scoring is also significant. Before automated systems, identifying sharp bettors and syndicates required dedicated trading staff reviewing account behavior, bet patterns, and timing anomalies. AI systems flag these cases in real time, enabling selective restriction and margin adjustment at the account level rather than market-wide blunt instruments.
Market StructureManaged Trading Services: AI Pricing-as-a-Service Takes Hold
The delivery model for AI trading has converged on a clear structure: Managed Trading Services (MTS), where operators subscribe to AI-driven pricing rather than staffing and maintaining proprietary trading desks. This is not a new concept—MTS has existed in various forms for years—but the AI layer has transformed what these services can offer and at what price point.
The current vendor landscape covers the major players:
- Kambi: AI pricing integrated directly into its B2B platform, now covering 48% of network volume. Operators on Kambi’s platform receive AI pricing as part of the core product, not a premium add-on.
- Genius Sports: Genius Sports Edge operates as a fully managed model—operators outsource the entire trading function, with AI handling line compilation, real-time adjustment, and liability management across 600,000+ fixtures.
- Sportradar: Alpha Odds delivers bespoke, per-operator AI pricing with real-time financial exposure recalculation. Sportradar has been named the most AI-advanced company in gaming by industry analysts, positioning Alpha Odds as the premium tier of the MTS market.
- GR8 Tech and Altenar: Both offer MTS subscription models targeting mid-market and emerging operators who lack the scale to justify in-house trading infrastructure.
The most significant signal of where the market is heading came in March 2026: Kaizen Gaming, the operator behind Betano, acquired GameplAI. Kaizen is a Tier 1 operator with significant in-house technical capability. Their decision to acquire rather than build signals that even the largest operators now view AI trading as a buy decision, not a build decision. When Tier 1 operators with eight-figure engineering budgets are buying AI trading capabilities from specialists, mid-market operators running manual desks face a decision they can no longer defer.
The broader market context supports sustained investment: the global B2B sportsbook solutions market is projected to reach $1.98 billion by 2032, with AI-driven managed trading services identified as a core growth driver in that figure.
In-Play EdgeMilliseconds Matter: AI and the Live Betting Advantage
If the structural case for AI pricing in pre-match markets is compelling, the case in live betting is overwhelming. In-play markets operate at a speed and volume that makes manual pricing not just inefficient—but structurally exploitable by sharp bettors who can identify and act on stale lines faster than a human trader can update them.
Automated trading systems reprice within milliseconds of in-game events: goals, injuries, red cards, substitutions, penalties. The interval between event and repricing that manual traders cannot eliminate—even with dedicated in-play staff and rapid communication—is where the margin leaks. Bettors who can process game state information faster than the book can update odds extract value at the operator’s expense.
AI systems eliminate this window. Altenar’s documentation on its automated trading infrastructure describes real-time repricing tied directly to live data feeds, with no human latency in the loop. Kambi has described its AI as handling thousands of markets simultaneously in real time—a volume that would require an implausibly large manual trading team to match.
The uptime dimension is equally important. Manual trading desks routinely suspend markets during rapid-event sequences—VAR reviews, penalty shootouts, injury stoppages—because the complexity of rapid repricing across multiple correlated markets exceeds what traders can manage in real time. Each suspension window is a moment where bettors cannot place bets and the operator earns nothing. AI systems achieve near-100% in-play market uptime by automating the repricing decisions that previously required suspension.
This matters because in-play handle now represents the majority of total handle at leading operators. The margin is won or lost in live markets, and the operators with the fastest, most accurate in-play pricing hold a structural advantage that compounds with every fixture cycle.
Traders Aren’t Gone—They’re Repositioned
The narrative around AI trading often defaults to displacement anxiety: traders will be replaced, headcount will shrink, the human element will disappear from the trading desk. The operational reality is more nuanced, and understanding the actual transition model matters for operators planning their own AI adoption.
What AI systems are replacing is the mechanical, high-volume work of manual line management—compiling pre-match odds from consensus markets, making routine adjustments, managing standard liability positions. This was the majority of a trading desk’s activity by volume, but it was also the least differentiating. The work that required genuine expertise—identifying unusual betting patterns, recognizing syndicate activity, understanding why a market is moving against consensus—remains a human function, now more concentrated and more impactful.
The emerging model is supervision rather than elimination. AI systems flag edge cases—anomalous bet patterns, unusual liability concentrations, markets where model confidence is low—and human traders resolve them. This is the “virtual trading desk” paradigm: AI agentic systems replicating the routine function of a human team, with human oversight focused on exception handling and model governance.
The practical staffing implication is fewer traders, operating at higher specialization, focused on tasks where human judgment genuinely adds value: identifying new sharp accounts, evaluating model performance on niche sports, managing relationships with data providers, and making strategic decisions about market coverage and margin philosophy. The trading desk does not disappear—it evolves from a production operation to a governance and strategy function.
2026 and Beyond: AI Trading as Table Stakes
The forward-looking market data frames the scale of what is underway. The global AI sports betting market was valued at $9 billion in 2024 and is projected to reach $28 billion by 2030 at a 21.1% CAGR, with AI-powered pricing and risk management identified as core growth components. The global sports betting market itself is projected to add $221.1 billion in incremental volume between 2025 and 2029—and the majority of that volume will be priced by AI systems.
For operators, the 2026 decision frame is not “should we invest in AI trading?”—it is “which model do we choose, and when?” The three options are:
- Build proprietary AI trading capability: Viable only for Tier 1 operators with significant data science resources and historical pricing data to train models. The Kaizen/GameplAI acquisition suggests even Tier 1 operators are questioning whether building beats buying.
- Subscribe to Managed Trading Services: The dominant model for mid-market and regional operators. Sportradar Alpha Odds, Genius Sports Edge, and Kambi’s integrated AI all offer this path with documented margin outcomes.
- Maintain manual-first trading with selective AI augmentation: The current position of operators who have not yet committed. This path accepts the margin disadvantage documented above as a deliberate trade-off—a position that becomes harder to justify with each quarter of published competitor data.
World Cup 2026 will be the first major global event where AI-priced markets are the norm, not the exception. The volume concentration of a World Cup—hundreds of millions in global handle across a compressed six-week window—will stress-test every operator’s trading infrastructure simultaneously. Operators running manual desks into that event face a capacity problem that cannot be solved by hiring more traders on short notice.
The longer horizon points toward fully autonomous AI agents managing end-to-end book construction with minimal human oversight. Several vendors are already describing their roadmap in these terms. The “virtual trading desk” concept—AI agentic systems that replicate the complete function of a human trading team—is not a 2030 vision. It is being built and deployed in production today, with the Kambi and Genius Sports data as evidence.
Data Sources & Attribution
- CDC Gaming: Kambi AI pricing share data — 4% (2022), 28% (2023), 48% (2025)
- Legal Sports Report: Kambi €7.4M incremental margin from AI pricing, 2025
- Genius Sports: Trading Services product page — 22% margin improvement, 600,000+ fixtures, 40+ sports
- AGB Brief: Kambi 2025 trends data — 88% Super Bowl Bet Builder player prop inclusion, 24% Champions League Bet Builder share
- Intellias: Global AI sports betting market — $9B (2024) to $28B (2030), 21.1% CAGR
- GlobeNewswire: B2B sportsbook solutions market — $1.98B projected by 2032
- GR8 Tech: Sports trading software guide — ~1% GGR protection via AI bet scoring
- Altenar: The Age of Automated Betting — millisecond repricing, in-play market uptime