Real organisations, measurable results, and what it means for your sector.
There is a familiar pattern in how emerging technologies gain traction. First, the academic papers. Then the analyst reports. Then the conference talks. Then we get the vendor pitches, and the LinkedIn posts from people who have read the analyst reports. Somewhere in the middle of all this, a reasonable question forms in the mind of any senior leader, executive or board member that is paying attention: “Is anyone actually using this?”
In the case of Causal AI – artificial intelligence that models genuine cause-and-effect relationships rather than statistical correlations – the answer is yes. Not universally, not at scale across every sector. But it is being used in enough places and with enough measurable impact to warrant real attention from leadership.
A survey of 400 senior AI professionals conducted by Databricks found that roughly seven in ten AI-driven organisations have either adopted causal AI techniques, are actively experimenting with them, or plan to adopt them by the end of 2026[1]. Causal AI has moved past the point where interest alone is the story. Results are starting to appear.
As an example, here is what that looks like across five sectors.
Finance: Building portfolios that explain themselves
For a long time finance has been a good, and active, place for complex data analysis. The sector has the data infrastructure, the quantitative talent, and (importantly) the commercial incentive to squeeze every available edge from its models. It also has regulators who increasingly want to understand why an AI system made the recommendation it did, not merely that it made one.
Causal AI is starting to be used in building portfolios and managing risks. Causify, a specialist platform founded by former members of Google and NVIDIA research teams, has applied causal modelling to portfolio construction. It reports achieving 25% better risk-adjusted returns compared with conventional approaches[2]. The difference lies in how the model understands market relationships. Where a correlation-based model might observe that certain asset classes tend to move together and build a portfolio on that basis, a causal model asks why they move together and, more importantly, whether that relationship will hold under stress. When it comes to stress-testing, the ability to reason about causal mechanisms rather than historical patterns is the difference between a model that extrapolates from the past and one that can reason about scenarios the past has not yet produced.
Fraud detection is another area where this distinction matters. Conventional models flag patterns that look anomalous based on historical fraud cases. Causal models go further, distinguishing between patterns that genuinely indicate fraudulent behaviour and those that are benign anomalies. Resulting in the reduction of false positives and focusing investigation resources where they will actually find something[3].
Healthcare: Making clinical trials shorter and smarter
The pharmaceutical industry spends billions on clinical trials. A significant proportion of that spending is wasted on trials that are too broad, too long, or targeted at the wrong patient populations. Causal AI is beginning to change the economics of drug development by enabling researchers to simulate ‘what-if’ scenarios on patient data before committing to a full trial.
By modelling the causal pathways through which a drug affects different patient groups, pharmaceutical companies can identify which populations are most likely to respond to a new treatment. Thus making trials shorter, more targeted, and less costly[3]. Rather than testing a drug across a broad and undifferentiated cohort and hoping for a statistically significant result, researchers can use causal models to design trials around the patients most likely to benefit, reducing both the time and the expense involved.
Beyond commercial drug development, causal methods are also being applied in public health. A Harvard-affiliated research team has used counterfactual modelling to diagnose drivers of childhood disease in Pakistan. They have identified factors that genuinely cause disease spread versus those that merely correlate with it[3]. When resources for public health interventions are limited, and they always are, the ability to focus on real problems instead of obvious ones greatly affects how many lives get better.
Supply chain: Diagnosing disruptions, not just predicting them
Supply chain management has been one of the most enthusiastic adopters of predictive analytics over the past decade. Demand forecasting, inventory optimisation, and logistics planning all benefit from the ability to anticipate what is likely to happen next. Predicting what will happen has a problem that is clearly seen. It’s good if the future is like the past, but supply chains often face new, unexpected issues.
Causal AI offers a different capability. The ability to understand why a disruption is occurring and model the likely effect of different responses. When a delay hits a supply chain, a causal model does not simply flag the anomaly. It traces the cause, identifies the downstream effects, and can suggest optimal responses, such as rerouting shipments. This is based on an understanding of the causal mechanisms involved rather than historical precedent alone[3].
Product vendors are increasingly sure this is the right way to go. Logility acquired Garvis in late 2024. This was specifically to improve its supply planning tools using AI to predict demand, mixing generative AI and causal AI[4]. Blue Yonder and Oracle are both exploring causal features within their planning platforms, with early capabilities focused on adjusting forecasts and suggesting responses to disruptions[3].
These capabilities are not yet mature in this sector. Most deployments are experimental or early-stage, but the trajectory is clear, and the organisations investing now are positioning themselves for a meaningful advantage as the tools mature.
Energy: Extracting more from existing assets
The energy sector presents a compelling use case for Causal AI, because the commercial value is so tangible. When a causal model can identify the operational parameters that genuinely drive performance, as opposed to those that merely correlate with it, the result is not abstract. It is additional output, reduced downtime, or avoided maintenance costs, all of which flow directly to the bottom line.
GE Vernova has worked with causaLens, a London-based specialist platform, to apply causal AI to wind turbine operations. By modelling the causal relationships between operational parameters and energy output, the system prescribes adjustments that extract additional energy from turbines at no extra capital cost[5]. The value proposition is simple. The same asset produces more because choices are based on real understanding of why things happen, not just past data.
In predictive maintenance, causal models are used to connect early warning signs to how things might fail, and ranking potential remedial actions by how likely they are likely to help. Causify reports that its causal approach to maintenance delivers early warning of equipment issues 14 to 21 days in advance, with reductions in unplanned downtime exceeding 15%[2]. For asset-intensive industries where unplanned downtime carries enormous costs, the ability to not only predict failure but understand its causal chain, and to intervene at the right point, represents a step-change from conventional condition monitoring.
Marketing: Understanding what actually drives the sale
Marketing has always grappled with the attribution problem. When a customer converts, which touchpoint deserves the credit? When a campaign runs and sales increase, was it the campaign or would sales have risen anyway? These are causal questions, but most marketing analytics answer them with correlational methods, resulting in misattribution and the waste of significant portions of marketing budgets.
Causal AI is enabling organisations to move beyond simple A/B testing towards continuous, individualised experimentation. If companies find out how much prices affect each customer’s buying choices, they can set the best price for everyone, instead of using general, wide, customer groups. A shift from asking “what price works for this segment?” to “what price works for this person, and why?”[6]
The broader implication is a shift from analytics that show what happened, to insights that advise what to do and why it’s effective. For organisations spending tens or hundreds of millions on marketing, the ability to distinguish effective actions from expensive coincidences is not a marginal improvement. It is a fundamental upgrade in how budget is allocated and performance is understood.
What this means for your sector
The five sectors above are not the only ones where Causal AI is gaining traction. Public sector applications in policy evaluation and fraud detection are developing rapidly. Other industries including telecommunications and insurance are beginning to explore causal methods. But these five illustrate the breadth of the opportunity and, more importantly, the common thread that runs through all of them.
In every case, the value comes from the same shift. From analytics that show you what happened, to analytics that tell you why it happened and what would change if you acted differently. That shift matters whether you are constructing a portfolio, designing a clinical trial, responding to a supply chain disruption, optimising a wind turbine, or deciding how to spend your marketing budget.
Causal AI is still maturing. Talent is scarce, data foundations matter enormously, and not every organisation is ready to adopt causal methods today. But the results emerging from early adopters suggest that this is not a question of whether Causal AI will become relevant to your sector. It is a question of when, and whether you will be among the organisations that moved early enough to capture the advantage.
This is the third in a series of articles exploring Causal AI in the enterprise.
References
[1] theCUBE Research, “The Causal AI Marketplace.”, 2024, https://thecuberesearch.com/the-causal-ai-marketplace/
[2] Causify, “Enterprise Causal AI Platform.”, 2026, https://causify.ai/
[3] Acalytica, “Causal AI Disruption Across Industries (2025–2026).”, 2025, https://acalytica.com/blog/causal-ai-disruption-across-industries-2025-2026
[4] MarketsandMarkets, “Causal AI Market Size, Share & Trends | Industry Forecast [2032].”, 2025, https://www.marketsandmarkets.com/Market-Reports/causal-ai-market-162494083.html
[5] causaLens, “Enterprise Causal AI Features” and published case studies, https://causalens.com/
[6] Dataversity, “Three Ways Causal AI Can Drive Your Business in 2025.”, 2025, https://www.dataversity.net/articles/three-ways-causal-ai-can-drive-your-business-in-2025/
Curious About 
Leave a comment