Getting Real

Many in the sector now see AI as one of several technologies that will shape how fuel is bought, delivered, consumed and monitored over the coming decade, alongside mass flow metering, electronic documentation and expanded data sharing.

Bunkering is increasingly being shaped by three overlapping forces: fuel-price volatility, tightening environmental regulation and a protracted transition toward lower-carbon shipping. Industry reports generally agree that carbon-neutral fuels are likely to remain expensive and limited in availability for years, leaving owners and charterers under pressure to improve efficiency while gradually introducing new fuels and propulsion concepts.

This environment generates growing volumes of operational and commercial data – from noon reports and sensor streams to lab certificates, port calls, invoices and contracts. For many stakeholders, AI is seen primarily as a way to turn this fragmented and sometimes inconsistent data into more reliable support for decisions on what to buy, where to lift, how much to stem and how fuel is used on board.

A broad consensus appears to be emerging that AI will not, on its own, transform bunkering overnight. Instead, it is expected to become part of the background infrastructure of the business, in much the same way that mass flow meters and electronic documentation have already done in some hubs.

Digital bunkering as the runway for AI
The steady advance of digital bunkering is widely regarded as a prerequisite for meaningful AI deployment. In Singapore and a number of other major ports, regulators are moving toward electronic bunker delivery notes and standardised data formats. In parallel, the IMO’s Maritime Single Window initiative is encouraging ports and national authorities to adopt more unified digital platforms for ship–shore data exchange.

Once bunker deliveries, customs information and port calls are captured in structured, machine-readable form, AI tools can be applied more effectively. Industry participants point to several immediate applications: benchmarking suppliers, identifying anomalies, automating reconciliation, and reducing the scope for disputes over quantity, quality or documentation.

Procurement: from price per tonne to “single point of truth”
Much of the current discussion around AI in maritime procurement focuses on data quality and governance. At GenPro’s 5th Annual Blue Day in Limassol, held under the theme “Single Point of Truth: Turning AI into Action in Sourcing and Procurement”, speakers repeatedly stressed that AI systems depend on consistent, reliable input data.

GenPro managing director Maria Theodosiou told delegates that AI “cannot fix chaos” and “is not magic”, noting that poorly structured data can lead to faster but not better decisions. That observation resonates with bunker buyers who work with multiple price sources, catalogues and contract templates. Even minor discrepancies in part numbers, currencies or product descriptions can undermine confidence in AI-generated output.

From a bunkering perspective, many observers therefore emphasise the importance of building a robust “single point of truth” for core data: vessel profiles, historical consumption, fuel grades and specifications, quality records, contract terms and counterparty information. AI is then seen as a layer on top of that foundation rather than a short-cut around it.

Christina Orfanidou, Head of Group AI at Columbia Group, highlighted the regulatory dimension, citing the EU AI Act and the introduction of ISO 42001 for AI management systems. She suggested that data governance, including classification schemes, quality indicators and clear ownership of datasets, is becoming integral to responsible AI use. For bunker buyers, this aligns with a wider move to treat price, quality and counterparty data with the same rigour historically applied to financial reporting.

Exelia Technologies CEO Margarita Maimonis, speaking at the same event, emphasised incremental implementation. She argued that successful projects tend to start with digitising specific processes and tackling realistic use cases, rather than attempting an end-to-end AI transformation from the outset. Applied to bunkering, that approach could mean beginning with AI-assisted supplier benchmarking on selected trades, or automating reconciliation between e-BDNs and invoices, before pursuing more ambitious concepts.

Beyond price per tonne: energy, risk and emissions
Traditional bunker negotiations have largely focused on the headline price per tonne. A growing number of technology providers and fuel managers are now promoting broader metrics that incorporate energy content, operational performance and regulatory exposure.

AI-enabled platforms are being developed to aggregate fuel-quality data, port-by-port, and combine it with price assessments, fleet trading patterns, weather forecasts and compliance constraints. The aim is to move towards questions such as:

Given a particular trading pattern, where should a vessel bunker, which grade should be used, how much should be lifted, and what are the implications for total cost and emissions?

By modelling energy content, expected consumption, Carbon Intensity Indicator (CII) performance and exposure to schemes such as the EU Emissions Trading System (ETS), these tools seek to present bunker managers with scenario-based options rather than a single ‘cheapest’ number.

In parallel, risk-intelligence providers are using AI to screen sanctions lists, vessel behaviour and ownership structures. For bunker traders and suppliers, such tools are increasingly seen as a way to manage a rising compliance burden, particularly in relation to sanctions, beneficial ownership and ‘dark fleet’ concerns.

Industry opinion generally reflects the view that these applications are evolutionary rather than revolutionary, extending existing analytical practices rather than replacing them.

Case study: AI on the bridge
The application of AI to navigation and situational awareness illustrates how fuel efficiency, safety and ESG considerations can intersect. Shipowner Harren Group’s partnership with Orca AI has been cited as an example of how real-time data from the bridge can be used to inform both operational and strategic decisions.

Harren’s newer heavy-lift and multipurpose vessels are equipped with highly efficient engine and propulsion systems designed to minimise emissions. The integration of marine technology company Orca AI adds an additional layer of intelligence via the SeaPod ‘digital lookout’, which combines visual and thermal sensors, and the FleetView platform, which aggregates navigational events across the fleet.

According to Harren’s managing director Nils Aden, the goal is to measure navigational behaviour in a way that was previously not feasible and to align ship and shore personnel around a common factual picture. The company has set targets to reduce close-quarters situations and near misses, on the basis that smoother and more predictable manoeuvring is positive for both safety and fuel consumption.

Orca AI co-founder and CEO Yarden Gross presents this type of deployment as representative of a broader industry trend. He argues that combining real-world navigational data with AI-driven analysis can support a more transparent safety culture while also generating ESG-relevant metrics, including fuel-use and emissions indicators.

Commentary around such projects generally suggests that, while individual implementations may vary, AI-enhanced situational awareness on the bridge is likely to become a more common feature of modern fleets, with implications for bunker use as well as safety performance.