Kambi Group published an ebook on its AI trading capabilities, making an argument that is less about artificial intelligence than about the data that makes it work.
The claim is straightforward: decades of betting volume processed across a shared network produces a training dataset that a single-operator in-house model cannot replicate. The ebook frames this as the structural case for outsourcing the sportsbook layer rather than building it.
The argument is worth examining carefully, because the word "AI" is doing a lot of work in this framing, and the actual competitive asset is older and less glamorous. What Kambi is describing is a data moat — the kind that compounds over time and that late entrants cannot purchase their way into. The AI component is the application layer. The moat is the accumulated position and outcome data that trains it.
This matters for how operators should read the ebook. An operator evaluating a proprietary trading model is not choosing between AI and no AI. Every serious platform has machine learning in its pricing stack now. The real question is whether the training corpus behind your model is large enough to price edge cases accurately — the obscure league, the weather-affected match, the market that opens thin and needs to move fast. A network that has priced those events thousands of times has a material advantage over one pricing them for the first time.
The commercial logic follows. If the data advantage is real and durable, then operators who build proprietary systems are not just incurring higher development costs — they are accepting worse pricing on the events where accuracy matters most, which tend to be the events where sharp money concentrates. That is where margin erosion happens.
I would weight the network-scale argument more than the market typically does. The sportsbook industry has a long history of operators believing their brand and product differentiation can substitute for superior pricing. Sometimes it can, in mass-market recreational segments. At the margin where sophisticated bettors operate, it cannot. Kambi's ebook is making a case that most operators will resist for competitive psychology reasons, and that resistance is usually more expensive than the alternative.
Kambi processes decades of betting volume across a shared network, creating a training dataset that individual operators cannot replicate with in-house models. The competitive asset is not the AI itself—every serious platform uses machine learning in pricing now—but the accumulated position and outcome data that trains it. A network that has priced obscure leagues, weather-affected matches, and thin-opening markets thousands of times prices edge cases more accurately than operators encountering those events for the first time, where margin erosion concentrates.
Operators who build proprietary systems incur higher development costs while simultaneously accepting worse pricing on events where sharp money concentrates and accuracy matters most. Kambi's ebook argues that this combination—elevated build costs plus margin erosion on high-value events—makes the economic case for outsourcing the sportsbook layer rather than substituting brand and product differentiation for superior pricing capability.
An operator pricing an obscure league, weather-affected match, or thin-opening market for the first time lacks the historical data advantage that compounds within Kambi's network. That knowledge gap translates directly to pricing inaccuracy where sophisticated bettors operate, producing margin erosion that typically proves more expensive than the cost of outsourcing would have been.
Kambi's ebook does not address whether prediction market volatility in the United States changes the calculus for its partner base, leaving that as the more interesting open question for evaluating whether the network-scale advantage thesis holds across different market structures and volatility regimes.
Continue reading.
Kalshi user sues for $580,000 after $308,000 in sports bet losses
Kalshi fights Louisiana order on two fronts as federal case opens
Kalshi seeks $65bn valuation as Trump Jr. exits advisory role
Arkansas preemption case tests limits of federal commodity law
A few more worlds
worth exploring.
From the FreeMalta ecosystem.
The AI Journey
Explore AI beyond the headlines. Tools, ideas and a place to start.
Explore the journey 02 / THE GARAGEThe Garage
The companies behind the names. Explore their stories and the people building them.
Open the garage 03 / THE DUELThe Duel
Two contenders. A closer look. Put products side by side before choosing.
Choose your duel 04 / MALTAINSIDERMaltainsider
Discover the B2B side of Malta. A different lens on business and opportunity.
Meet Maltainsider