Three years ago, AI traded 4% of bets on Kambi’s network. In 2024, that figure was 28%. By end of 2025, it had reached 48%. The acceleration is not incremental — it is structural. AI has moved from a trading experiment to the core infrastructure of competitive sportsbook operations, and the operators who treat this as a future consideration are already behind.
This article breaks down the financial case, the scale pressures driving adoption, the fraud reality operators face, and the concrete steps required before World Cup 2026 — the first major international tournament that will be fully traded by AI on Kambi’s network.
The Inflection PointFrom Experiment to Infrastructure: How AI Took Over Pricing
The trajectory of AI adoption in sportsbook trading is one of the clearest trend lines in the industry. According to Kambi’s 2025 Sports Betting Trends Report, 48% of all bets across their 50+ operator network were priced by AI in 2025 — up from 28% in 2024 and a baseline of just 4% in 2022. That is a 12x increase in three years.
The financial impact is direct and measurable. Kambi’s operator client margin rose from 10% in 2024 to 10.8% in 2025, with that gain explicitly attributed to AI-powered pricing efficiency. At the business level, Kambi’s Adjusted EBITDA grew 16% in Q4 2025 to €7.4M — a result tied to the same AI expansion. These are not projections or correlation claims. They are reported outcomes from a network operating at scale.
| Year | AI share of bets (Kambi network) | Operator margin |
|---|---|---|
| 2022 | 4% | — |
| 2024 | 28% | 10.0% |
| 2025 | 48% | 10.8% |
The signal is clear: as AI pricing coverage expands, operator margin follows. The mechanism is straightforward — AI reprices faster, holds accurate lines longer, and reduces the window in which sharp bettors can exploit stale odds. Each of these effects contributes directly to the margin line.
World Cup 2026 will be the first major international soccer tournament fully traded by AI on Kambi’s network. Kambi expects approximately €5 million in incremental revenue — a roughly 3% revenue lift — from this single event. That figure is a forcing function: operators who have not built AI trading infrastructure before the tournament window will be unable to capture it.
The Financial Case Is No Longer Theoretical
The margin improvement Kambi reports at network level is supported by individual operator data. One trading services provider reported a 22% football betting margin increase using virtual AI agents in the 2025/26 season, according to Sports Betting Operator. The mechanism: AI agents continuously recalibrate prices against incoming data rather than waiting for human trader review cycles.
The CLV (Closing Line Value) advantage is equally significant. According to Intellias, top AI models beat the final market odds by 3–7% on average. This means AI-enabled operators are pricing more accurately than the market consensus — a structural edge that compounds across thousands of markets and millions of bets. Operators still on manual pricing are, in effect, offering slightly worse-priced markets than AI-powered competitors, while also being slower to correct when events shift.
The global AI-powered sports betting market was valued at approximately $9 billion in 2024 and is projected to reach $28 billion by 2030, growing at a 21.1% CAGR. Operators who delay AI adoption are not simply missing margin improvements today — they are ceding ground in a market that will be defined by AI infrastructure within this decade.
Scale ProblemBet Builders and Player Props Are Breaking Manual Trading
The scale argument for AI is as compelling as the margin argument. bet builder formats — where customers combine multiple selections from a single match into a single bet — have grown from a niche product to the dominant pre-match format at major events. According to Kambi’s data, approximately 50% of Super Bowl LIX pre-match bets were placed via Bet Builder. Of those Bet Builder bets, 88% contained a player prop.
The implications for pricing are substantial. A Bet Builder with four player prop legs requires real-time correlation modeling across all four selections — accounting for in-match dependencies, player status updates, and opponent defensive data. Human traders cannot price these combinations at volume with the speed the market demands. AI does not tire, does not queue, and does not introduce latency between a market trigger and a price update.
The Champions League Bet Builder trend shows the same trajectory over a longer time horizon: share of pre-match Bet Builder bets grew from 8% in 2020 to 24% in 2025 — a three-fold increase in five years, with no sign of plateau. Operators who want to offer competitive Bet Builder products — tighter margins, faster pricing, higher market availability — cannot do so at scale without AI infrastructure.
The scope of real-time risk monitoring reinforces the point. AI systems can monitor hundreds of thousands of bets and thousands of markets simultaneously, flagging anomalies and adjusting lines in real time. A human risk team faces physical limits: they can watch a finite number of markets, with a finite reaction time. AI operates without either constraint. For operators with significant Bet Builder and player prop volume, this is not a future consideration — it is an active gap in their current risk management capability.
Fraud RealityUp to 44% of Sportsbook Traffic Is Fraudulent — AI Is the Only Answer
Fraud is where the AI case becomes an existential argument, not just a margin optimization. According to data from TrafficGuard, based on campaign data from 100+ sportsbooks, up to 44% of traffic at the largest sportsbook operators is fraudulent or invalid. The primary mechanism: AI-powered bots creating fake accounts at scale to harvest welcome bonuses and promotional offers.
The consequence is direct marketing ROI destruction before a single legitimate player converts. An operator running a customer acquisition campaign at €50 cost-per-acquisition, with 40% of that traffic being fraudulent, is effectively paying €83 for each genuine player — before any further leakage from promotional bonus abuse by the fraudulent accounts that slip through.
Manual fraud review cannot keep pace with bot-scale attack volumes. A human compliance team reviewing suspicious accounts can process hundreds of cases per day. Bots can generate thousands per hour. The asymmetry is permanent — the only solution that matches the scale of the problem is AI detection operating at the same speed as the attack.
The industry’s leading integrity monitoring infrastructure gives a sense of the scale involved. Sportradar’s AI-powered Fraud Detection System monitors approximately 850,000 sporting events across 70 sports per year, identifying over 1,000 suspicious matches annually. That is a surveillance footprint no human team could replicate. AI-powered retention and fraud systems deliver 40% fraud reduction alongside 35% engagement lifts — the ROI is simultaneously defensive and offensive.
Hybrid ModelHuman + AI: The Operating Model That’s Winning
The correct framing for AI trading is not replacement — it is augmentation at inhuman scale. The hybrid human-AI trading model is emerging as the dominant operating structure across the industry: AI sets lines, manages routine risk, monitors markets in real time, and flags anomalies; human traders focus on novel market structures, edge cases, and decisions that require contextual judgment that AI has not yet matched.
This is not a theoretical architecture. Palms Bet in Bulgaria is launching what has been described as the world’s first fully AI-driven sportsbook, powered by SSTrader and Altenar. The competitive benchmark is being reset in real time. Operators who have not yet begun building AI trading infrastructure are not competing against a hypothetical future — they are already trailing a live deployment.
The accuracy numbers support the model. AI tools have reached 75–85% prediction accuracy in major sports as of 2026, according to WSC Sports. GenAI-powered prediction models show 300% higher accuracy compared to prior AI baselines — a step-change that makes the previous generation of AI pricing tools look like a prototype. At these accuracy levels, AI-priced lines are consistently closer to true probability than manually constructed odds, which means better margin, fewer arbitrage opportunities for sharp bettors, and more defensible book positions throughout match events.
AI Doesn’t Stop at Pricing — Retention Is the Second Lever
Operators who treat AI as a trading-only tool are leaving significant value unrealized. The same infrastructure that enables AI-powered pricing and risk management creates a foundation for AI-driven retention — and the retention metrics are as compelling as the margin numbers.
Platforms with advanced AI-driven personalization report a 35% increase in user engagement, according to Yogonet’s 2025 analysis. AI-powered interventions prevent 20–30% of churn through timely, personalized outreach that responds to behavioral signals — a falling deposit frequency, a shift in bet size, a gap in login activity — before a player goes fully dormant.
At the bet slip layer specifically, AI personalization surfaces the right markets, formats, and player props for each individual user — turning identical traffic into higher-engagement, higher-margin sessions without requiring any additional acquisition spend. A player who regularly bets on Champions League Bet Builders sees a pre-populated bet slip with relevant combinations. A player with a history of in-play football betting gets real-time suggestions during the match. The conversion lift from this kind of personalization is additive on top of any trading margin improvement.
The fraud reduction data closes the loop on the retention argument. AI-powered retention systems deliver 40% fraud reduction alongside those engagement gains — meaning the same AI investment that improves legitimate player engagement simultaneously reduces the fraudulent account activity that dilutes campaign ROI. See our research on reactivating dormant players at scale for the full financial model on AI-driven retention for European operators.
Action PlanWhat Operators Must Do Before World Cup 2026
The competitive window before World Cup 2026 is the clearest near-term forcing function in the industry. For operators who have not yet moved on AI trading infrastructure, the practical sequence is as follows.
1. Audit Current Pricing Coverage
Map which markets are still manually priced and quantify the margin exposure. Markets with high Bet Builder participation and significant player prop volume are the highest-priority gaps — these are where the manual-vs-AI speed differential is most visible to both bettors and sharp money.
2. Prioritize AI for High-Volume Formats First
Bet Builder, player props, and in-play repricing represent the highest-leverage entry points for AI trading. These formats combine high volume, high complexity, and high sensitivity to pricing latency — making them both the most expensive areas to remain on manual systems and the fastest areas to generate measurable margin improvement after AI deployment.
3. Layer In AI Fraud Detection Before the Next Major Campaign
With up to 44% of traffic at the largest operators being fraudulent, running a major acquisition campaign without AI fraud detection in place is equivalent to filling a bucket with a hole in it. The fraud detection investment should precede, not follow, the next major promotional push. The 40% fraud reduction that AI systems deliver has an immediate and direct impact on effective cost-per-acquisition.
4. Prepare for Regulatory Requirements Around Explainable AI
Regulatory pressure on automated decision-making in gambling is growing. Several jurisdictions are beginning to require disclosure of automated pricing decisions, and the trend toward explainable AI requirements will accelerate as AI share of bets continues to climb. Operators who build with interpretability in mind now will avoid remediation costs later. Build audit trails into AI trading decisions from the outset.
5. Connect Trading AI to the Bet Slip Layer
The full value of AI trading is realized when it connects through to the player-facing product. AI-priced markets are only part of the opportunity — AI-personalized bet slip presentation, surfacing the right markets and formats to each player, multiplies the conversion impact of accurate pricing. The operator who prices well and presents well captures more of the available margin than one who does only the former.
| Priority | Action | Impact |
|---|---|---|
| 1 | Audit manual pricing gaps, quantify exposure | Baseline for investment case |
| 2 | Deploy AI on Bet Builder, player props, in-play | Direct margin improvement (22% football uplift possible) |
| 3 | AI fraud detection before next acquisition campaign | 40% fraud reduction, recover effective CPA |
| 4 | Build explainability into AI pricing decisions | Regulatory readiness |
| 5 | Connect AI pricing to personalized bet slip layer | 35% engagement lift, 20–30% churn reduction |
The operators who move through this sequence before the World Cup window will enter the tournament with a compounding advantage over those who do not. The €5M incremental revenue Kambi projects from fully AI-trading the 2026 World Cup will not accrue evenly — it will concentrate in the operators who have built the infrastructure to capture it.
The global AI sports betting market is growing at 21.1% annually from a $9B base in 2024 toward a projected $28B by 2030. The structural shift is already underway. The remaining question for operators is not whether AI will define competitive sportsbook trading — it is whether they will be positioned to benefit from that shift or absorb the cost of being behind it.
For the player-facing side of this equation — how AI pricing connects to bet slip personalization and downstream retention — see BidCanvas AI Betslips. For the full product landscape, see BidCanvas Products. Related research: Prediction Markets: The Untapped Goldmine and How Sharp Money Moves Lines.
SourcesData Sources & References
- Kambi Sports Betting Trends Report 2025 — 48% AI bet share, 10.8% operator margin, World Cup 2026 projections
- Covers.com: AI accounts for nearly half of sports bets on Kambi network — 28% (2024) year-over-year comparison
- Gaming America: AI trading statistics — 4% baseline (2022)
- Legal Sports Report: Kambi Q4 2025 earnings — EBITDA +16% to €7.4M, €5M World Cup projection
- Sports Betting Operator: Virtual AI agents — 22% football margin improvement
- Intellias: AI in sports betting — 3–7% CLV advantage, Sportradar 850,000 events monitored
- WSC Sports: GenAI accuracy — 75–85% accuracy, 300% GenAI improvement
- TrafficGuard: Sportsbook fraud report — up to 44% fraudulent/invalid traffic (100+ sportsbook campaigns)
- Yogonet: AI personalization in iGaming (Dec 2025) — 35% engagement lift, 20–30% churn reduction, 40% fraud reduction
- Next.io: AI prices half of bets on Kambi network — 10.8% operator margin in 2025