This week the theme is foundations. Almost nothing that happened was an AI product launch — it was the layer underneath one. Korea broke ground on a purpose-built AI port and class signed off a floating data centre, which is what it looks like when a country and a shipyard start pouring concrete around the assumption. A P&I club and a vetting body agreed to stop inspecting the same ship twice, because the constraint was never the model, it was the duplicated data. DNV began teaching maritime AI and named the blockers out loud: data quality, governance, scaling, trust. OCIMF wrote the first rulebook for a system that steers the ship, eighteen months before it takes effect. Dualog gave every seafarer a verified identity, without which nothing onboard can be governed at all. And the Nautical Institute finally asked the seafarers what all of this is doing to their work — the first time anyone has. Foundations are slow, unglamorous and rarely make headlines. They are also the reason most AI pilots either scale or quietly die.

The week's most important developments in shipping & oceans — distilled into a 5-minute read.

🔗 Quick Links

• 🏗️ Korea breaks ground on its first Physical AI port at Gwangyang — Smart Maritime Network

  Seven days after unveiling the national strategy, the Ministry of Oceans and Fisheries put a spade in the ground: ₩772.4bn (about $538m), four berths, 1.36m TEU a year, completion 2029, with agentic AI optimising the yard and AI watching the AGV routes for cracks and ground subsidence

• 🖥️ Bureau Veritas approves Seatrium's 30 MW floating data centre — Digital Ship

  An Approval in Principle for a jetty-moored, non-propelled barge carrying six independent 5 MW "Data-in-a-Box" modules, with a 100 MW variant already in design — class has now assessed AI compute as a marine asset

• 🔍 RightShip and NorthStandard start sharing inspection reports — Digital Ship

  The insurer can now request existing RightShip inspection reports rather than sending its own surveyor, cutting duplicate condition surveys in dry bulk and general cargo; dry bulk ship manager Orion Reederei tested the model before launch

• 🎓 DNV opens AI courses for maritime professionals — and names the real blockers — Manifold Times

  DNV Maritime Academy's Dr Piyush Raj says the obstacles are data standardisation, governance and cyber control, and scaling past the pilot — not the models; MPA Maritime Cluster Fund training grants apply for eligible participants in Singapore

• 🧭 The Nautical Institute and Lloyd's Register Foundation ask seafarers what the technology is doing to them — Smart Maritime Network

  The STEER Project survey asks where automation, AI, digital systems and remote operations genuinely help, where they add workload, how they change decision-making, and whether training has kept up — evidence, at last, from the people using it

• 🇪🇸 CyberLogitec wins the TOS and digital twin for TTI Algeciras — Port Technology

  OPUS Terminal plus a digital twin for the B1 expansion at the Mediterranean's first semi-automated container terminal, taking it from 1.6m to 2.1m TEU by 2028 — with scenario simulation before anything goes live

• ⚓ OCIMF publishes the first implementation guidance for assisted steering — Smart Maritime Network

  Track Guidance Assistant for Inland Navigation systems steer a vessel along a predefined track; the first-edition guidance covers design, installation, training, procedures, human factors and cyber security ahead of legislation taking effect 1 January 2028

• 💬 Dualog rebuilds ship–shore messaging around individual crew identities — Smart Maritime Network

  Workspace gives every seafarer a personal identity tied to vessel, rank and rotation status instead of a shared bridge account — traceable under IACS UR E26/E27 — with an AI assistant for drafting, translation and inbox management to follow

🚀 Big Moves This Week

1. Korea Puts a Spade in the Ground: Gwangyang Becomes the First Physical AI Port

Last week we reported Korea's national "Physical AI" port strategy and told you to watch Gwangyang. Seven days later, on 30 July, the Ministry of Oceans and Fisheries broke ground on it. The facility: Gwangyang Port's Phase 3-2 Container Terminal — four berths, 1.36 million TEU a year, ₩772.4bn (about $538m), completion 2029. It is deliberately a shop window as much as a terminal: cargo-handling equipment, container transport equipment and the port operating system are all to be domestically developed, because the strategy's second target is capturing 10% of the global port equipment market, not just moving boxes faster at home. What the AI actually does here is more specific than most port announcements. Two applications are named. Agentic AI will optimise yard operations — the sequencing and stacking decisions that determine whether a truck waits twenty minutes or ninety. And AI will monitor the automated guided vehicles themselves to detect infrastructure hazards, spotting cracks and ground subsidence along the AGV routes. That second one is the more interesting: the vehicles become a continuous condition-monitoring sensor for the pavement they run on, which is exactly the kind of second-order use case that only appears once you have a fleet of instrumented machines already driving the same lines all day. Minister Hwang Jong-Woo said the terminal will "enhance Korea's import and export logistics competitiveness" while "spearheading Korea's Physical AI port innovation". The same week, Korea started exporting the shipyard version. HD Korea Shipbuilding & Offshore Engineering signed an MOU with Fraser Industries to modernise US yards, beginning with Fraser Shipyards on Lake Superior, Wisconsin as a demonstration project — yard layout design, robotics and automation, production optimisation, AI-based digital systems, supply chain and workforce training, with a contract signing planned later in 2026. Why this matters for ship managers: Gwangyang is now a dated reference site. From 2029 you will be able to compare a purpose-built AI terminal against the automated brownfield conversions, and the gap between the two will tell you a great deal about whether retrofitting automation into a live terminal was ever the right call. The honest caveat: this is a groundbreaking, not a commissioning. Three years of construction stand between the press release and a single measured crane move, and the specific figure to watch is not the 2035 productivity target but whether Gwangyang's build stays on the 2029 date. [Smart Maritime Network · HD Hyundai + Fraser Industries]

2. Class Approves a Data Centre: Bureau Veritas Signs Off Seatrium's 30 MW Floating AI Barge

Bureau Veritas Marine & Offshore has granted Approval in Principle to Seatrium Technology & Innovation for a 30 MW floating data centre — and the significant word in that sentence is "class". The design: a jetty-moored, non-propelled barge housing six independent 5 MW modules, each a self-contained "Data-in-a-Box" unit integrating IT equipment, cooling, power distribution and supporting systems for continuous operation. Seatrium is developing a 100 MW variant on the same principles. BV assessed the concept against its rules and applicable international standards, examining vessel arrangement, cooling systems, stability, marine systems and safety — the ordinary marine questions, applied to a machine whose purpose is inference and training rather than cargo. The approval follows an MOU signed between BV and Seatrium during Singapore Maritime Week on offshore power and digital technologies. EVP Aziz Merchant framed floating data centres as a scalable, resilient and sustainable answer to the land, power and cooling constraints strangling conventional sites. Why this matters for ship managers — and it is not the obvious reason. In our 5 July issue we reported MSI's finding that data-centre demand for 1–4 MW gensets is competing directly with shipyards for marine engine supply, lengthening lead times and pushing out newbuild deliveries. That was AI taking capacity away from shipping. This is the other direction: AI infrastructure becoming a marine asset class that yards build, class approves, crews maintain and marine insurers underwrite. If floating compute scales, the yard slot competition gets worse before it gets better — but so does the addressable market for anyone who can build, moor, power and service a barge. What to do with it: if you operate offshore support, mooring, or towage assets, this is a demand signal worth pricing now rather than reading about in two years. The honest caveat: an Approval in Principle is the first class gate, not the last — it says the concept is not obviously unsafe, not that anything is being built. No customer, no site, no capital cost and no deployment timeline were disclosed. Floating data centres have been announced repeatedly since 2022 and delivered rarely; treat this as the marine-assurance layer catching up with an idea, not as an order. [Digital Ship]

3. RightShip and NorthStandard Stop Inspecting the Same Ship Twice

On 29 July, vetting and risk platform RightShip and marine insurer NorthStandard announced that NorthStandard can now request existing RightShip inspection reports through RightShip's secure report-sharing capability — reusing an inspection that has already happened instead of sending another surveyor up the gangway. The problem it addresses is one every dry bulk manager will recognise: vessels in the dry bulk and general cargo trades are inspected repeatedly by different organisations — charterers, terminals, vetting bodies, insurers — with a large share of each assessment covering the same operational ground. Every one of those consumes crew time on a ship that is also trying to work cargo. Under the new arrangement, NorthStandard uses RightShip's safety and risk intelligence to strengthen loss prevention, minimising the need for a separate condition survey where suitable inspection data already exists. Dry bulk ship manager Orion Reederei ran the operating model before launch. The quotes are unusually direct. RightShip Chief Maritime Officer Christopher Saunders: "The dry bulk and general cargo industry doesn't need more inspections, it needs better use of trusted inspections." Executive Director Andrew Roberts called secure report sharing a way of "helping reduce duplicate inspections while making better use of inspection information". NorthStandard Loss Prevention Director Simon MacLeod said it "minimises duplication of inspections, eases pressure on crews, and supports a more efficient" approach to assurance. Why this matters for ship managers: this is the quiet, structural half of maritime AI that never makes a keynote. Every AI risk model in shipping is limited by the same thing — the inspection and condition data underneath it — and the industry's response for a decade has been to generate more inspections rather than share the ones it has. An insurer accepting another organisation's inspection is a small precedent with large implications: it establishes that a trusted digital report can substitute for a physical visit. What to do now: ask your vetting provider and your P&I club whether your existing inspection reports can be shared under this or a comparable arrangement, and start counting the inspection hours your crews actually lose per vessel per year — you will need that number to negotiate. The honest caveat: no figures were published — no count of duplicate inspections eliminated, no crew hours saved, no cost. It is also a bilateral arrangement between two organisations, not an industry standard, and the harder questions are unanswered: who is liable if a shared report misses a defect, and what happens to a manager whose reports are shared onward into a market that prices risk on them. [Digital Ship]

4. DNV Starts Teaching Maritime AI — and Names the Blockers Out Loud

DNV Maritime Academy has built a set of AI courses for maritime professionals, and in a Q&A published on 1 August, DNV Maritime Advisory's Head of Maritime Technology & Innovation, Dr Piyush Raj, described the obstacles in terms that should sound familiar to anyone who read last week's 11% figure. The courses pair AI fundamentals with maritime-specific application across operations, maintenance, safety, fleet management and decision support, and deliberately include the unglamorous layer: regulation, governance, cyber security, assurance and human factors. The stated aim is capability to evaluate AI — identify high-value use cases, judge vendor claims critically, understand the regulatory implications, and decide what not to adopt. In Singapore, Maritime Cluster Fund training grants are available to eligible participants through the MPA. The four blockers Raj names: data quality — organisations hold large datasets but cannot standardise them; governance — cyber security and control-and-assurance problems affecting both the data and the systems; scaling — moving past the pilot requires organisational buy-in and process integration, not better models; and trust — maritime must demonstrate AI meets the same safety bar as the technologies it replaces. His framing of the last one is the sentence worth pinning up: "Safety is the backbone of how shipping has created the modern global economy. If we lose confidence… we've lost everything." Why this matters for ship managers: last week's research found 81% of maritime companies piloting AI and 11% with policies to scale it. That gap is a skills and governance gap, and it does not close by buying another platform. When a class society starts teaching AI evaluation rather than AI enthusiasm, the market is admitting that the scarce resource is people who can tell a good system from a good demo. What to do now: identify who in your organisation is allowed to say no to an AI vendor, and check whether that person has been trained to. If the answer is your superintendent's instinct, you have a governance gap regardless of how many pilots are running. The honest caveat: this is a class society selling training, and its commercial interest in maritime AI assurance is direct. That does not make the diagnosis wrong — it matches independent research — but it does mean the course is a product, and there is as yet no published curriculum benchmark, certification standard or outcome data to judge it against. [Manifold Times]

5. Somebody Finally Asks the Seafarers: The Nautical Institute Launches the STEER Survey

The Nautical Institute has launched a global survey on how technology is reshaping maritime work, delivered with Lloyd's Register Foundation under the name STEER — Seafarer Technology Engagement, Empowerment and Resilience — and it is the direct sequel to the alarm fatigue argument LR took to Parliament two weeks ago. What it asks: the questionnaire examines the impact of automation, AI, digital systems, remote operations and connectivity on daily operations at sea — specifically where technology improves efficiency, where it creates additional workload, how it influences decision-making, and whether training genuinely prepares people for the systems they are given. It is open across merchant, fishing, offshore, inland waterways, cruise, yacht and naval support sectors, and to deck officers, engineers, ratings, trainers, educators, naval architects and remote operators — the last category being a quiet acknowledgement that "seafarer" now includes people ashore. The two quotes carry the argument. STEER Project Lead Captain Ann Pletschke: "Technology is changing maritime operations at pace, but the people who use these systems every day must be central to the conversation." Director of Projects David Patraiko: "By listening to those with direct operational experience, we can help ensure future tools and guidance are based on evidence rather than assumption." Why this matters for ship managers: almost every claim made about maritime AI's effect on crew workload is currently an assumption. Vendors assert their tool reduces burden; nobody has systematically asked whether it does. If STEER produces a real dataset, it becomes the first independent evidence base on the human cost and benefit of the technology this industry has spent three years installing — and, unlike vendor case studies, it is not funded by anyone selling a platform. What to do now: this one is free and takes almost nothing. Circulate the survey to your fleet. A manager whose crews are represented in the dataset gets to argue from evidence when the findings start shaping guidance; one whose crews are absent does not. It is also, frankly, a cheap and visible way to show seafarers that someone asked. The honest caveat: survey research is only as good as its sample, and self-selecting online questionnaires skew toward the digitally engaged and the strongly opinionated. No sample target, closing date or publication timeline has been announced, and findings will inform "guidance and practical tools" rather than anything binding. [Smart Maritime Network]

6. CyberLogitec Takes the Mediterranean's First Semi-Automated Terminal — With a Digital Twin Attached

CyberLogitec has won the contract to supply the terminal operating system and digital twin for the B1 expansion at TTI Algeciras, and the pairing is the point: the operating system and the simulator of the operating system, bought together. The deal: CyberLogitec's OPUS Terminal platform will coordinate vessel, yard, gate and equipment operations in a single environment at TTI Algeciras — jointly owned by HMM and a CMA CGM-affiliated investment company, and positioned as the Mediterranean and Southern Europe's first semi-automated container terminal. The B1 expansion takes annual capacity from 1.6m TEU to 2.1m TEU by 2028, adding 500,000 TEU. Alongside the TOS, the digital twin provides a real-time visual representation of terminal activity and equipment status and — the operationally important part — allows scenario simulation before changes are deployed live. The Port of Algeciras Bay was ranked Europe's most efficient container port in the World Bank and S&P Global 2025 Container Port Performance Index, so this is an upgrade to a terminal that already works. Why this matters for ship managers: this is the direct answer to the HHLA story we led with last week. Hamburg cut its 2026 profit guidance partly because automating a live terminal disrupted it more than expected — the automation J-curve, in a regulated disclosure. A digital twin that lets a terminal test the change in simulation before imposing it on real ships is precisely the tooling that shortens that curve. Whether it works is now testable: Algeciras is expanding with a twin, Hamburg converted without one being central to the story, and we will be able to compare. What to do now: when a terminal you call announces an automation or expansion programme, ask directly whether they are simulating the transition and what the modelled productivity dip is. That question separates operators who have planned the disruption from operators who will discover it. The honest caveat: no contract value, no timeline for the digital twin's go-live and no productivity or throughput targets were disclosed, and "semi-automated" is doing real work in that sentence — this is not a fully automated terminal. A digital twin also only simulates what it has been built to model; the failures that hurt in a transition are usually the ones nobody thought to put in the model. [Port Technology]

7. OCIMF Writes the First Rulebook for a System That Steers the Ship

OCIMF published the first edition of its implementation guidance for Track Guidance Assistant for Inland Navigation (TGAIN) systems on 30 July — and while the subject is European inland waterways, the precedent it sets is the one to watch. What TGAIN does: the system automatically steers a vessel along a predefined track, using positioning, heading and rate-of-turn data, so the master can concentrate on monitoring the environment rather than holding the line. It is, in the guidance's framing, a navigation aid — and OCIMF is emphatic that these systems "remain navigation aids and do not replace the master's judgement, situational awareness or responsibility for safe navigation." What the guidance covers: recommendations for vessel owners, technical operators, equipment manufacturers and crews, spanning system design, installation, training, operating procedures, human factors and cyber security — the full lifecycle, not just the specification. OCIMF notes that near misses and incidents have already demonstrated the need for consistent operating practices, which is a polite way of saying the systems arrived before the procedures did. Legislation governing TGAIN use takes effect on 1 January 2028. Publication and Advocacy Director Darron Biddle put the thesis plainly: "deployment of new technology alone is not enough" without clear procedures, appropriate training, and understanding of capabilities and limitations. Why this matters for ship managers, even if you never see an inland barge: this is the template. A partially automating function arrives, incidents accumulate, an industry body writes operating guidance covering training and human factors, then legislation follows eighteen months later. Every deep-sea equivalent — automated collision avoidance, AI-assisted watchkeeping, autonomous berthing — is on the same road, and the sequence here tells you what to prepare: procedures and training first, because that is what the guidance always turns out to be about. What to do now: for any system on your bridge that acts rather than advises, write down what it does, what it cannot do, and how the officer takes back control — and check that the person on watch has read it. The honest caveat: TGAIN is genuinely narrow, applying to inland navigation on European waterways, and OCIMF is an oil-major body issuing voluntary guidance, not a regulator. Its relevance to your fleet is as a leading indicator, not a compliance obligation. [Smart Maritime Network]

8. Dualog Gives Every Seafarer an Identity — the Precondition for AI on Board

Dualog has launched Workspace, a ship–shore collaboration platform whose real innovation is not the chat window but the identity model underneath it: every user gets an individual account tied to vessel, rank and crew rotation status, rather than a shared bridge login. What ships: Dualog Chat provides ship- and rank-based messaging with audio and video calling; users are grouped automatically by vessel, rank and rotation, and message history persists by role when a crew member changes position or leaves — so the incoming Chief Officer inherits the Chief Officer's conversations rather than starting blind. It integrates with crew management systems for automated user provisioning, runs on web, desktop and mobile, includes a data-saving mode for low-bandwidth satellite links and queues messages when connectivity drops. The compliance angle is the sharp end: individual credentials instead of shared accounts produce traceable access logs, which Dualog positions against IMO cyber guidance and IACS UR E26/E27. What is coming: file sharing, Dualog Mail integration, and an AI assistant for drafting reports, translating messages and managing the inbox — the same territory GTMaritime entered last week. Why this matters for ship managers: you cannot deploy useful AI on board without knowing who is asking. Role-aware assistance — surfacing the right procedure to the right rank, logging who was told what, translating for a mixed-nationality crew — depends entirely on individual identity, and most vessels still run shared accounts on shared terminals. This is plumbing, and it is the plumbing that has to exist before the AI layer means anything. It also quietly addresses a real welfare point: a personal identity that follows the seafarer across rotations treats them as a person rather than a station. The honest caveat: the AI assistant is not shipped — it is announced, and Dualog published no pricing, no availability dates, no customer count and no pilot data. Individual identity also cuts both ways: traceable logs are a cyber control and a surveillance capability, and crews will read them as both. Agree upfront what is logged, who can see it, and how long it is retained — before the AI layer starts reading the same messages. [Smart Maritime Network]

📊 Why It Matters — Strategic Impact Table

🔭 On Our Radar

• 🏗️ Does Gwangyang Hold Its 2029 Date — and Does the Korean Equipment Stack Sell Abroad? — Korea has moved from strategy to groundbreaking in a week and paired it with an MOU to modernise US yards, we monitor whether construction milestones hold against the 2029 completion target, track whether the domestically developed cranes, transport equipment and port operating system win their first export order outside Korea, and assess whether agentic yard optimisation and AGV-based infrastructure monitoring deliver measurable results or remain design intent.

• 🖥️ Does Floating AI Compute Reach Steel, or Stay at Approval in Principle? — Bureau Veritas has now assessed a 30 MW data centre barge as a marine asset with a 100 MW variant in design, we monitor whether Seatrium converts the AiP into a named customer, site and order, track which other class societies publish rules for floating compute, and assess whether this demand pulls further yard, engine and electrical capacity away from conventional newbuildings — the shortage MSI flagged in early July.

• 🔍 Does Shared Inspection Data Become an Industry Standard or Stay Bilateral? — RightShip and NorthStandard have established that an insurer will accept another organisation's inspection report, we monitor whether other clubs, charterers and vetting bodies join the sharing model, track whether anyone publishes the number of duplicate inspections actually eliminated and the crew hours recovered, and assess how liability is allocated when a shared report misses a defect.

• 🎓 Does AI Competence Become a Certified Requirement Rather Than a Course? — DNV is now teaching maritime AI evaluation while independent research puts only 11% of companies at scaling-policy maturity, we monitor whether any flag state, class society or STCW review moves toward formal AI competence requirements for shore and sea staff, track whether owners fund training as a line item rather than an afterthought, and assess whether trained buyers actually reject more vendors.

• 🧭 Does STEER Produce the First Independent Evidence on AI's Effect on Crew Workload? — The Nautical Institute and Lloyd's Register Foundation have opened the survey but published no sample target, closing date or publication timeline, we monitor whether the response base is large and representative enough to carry weight, track whether the findings corroborate or contradict LR's alarm fatigue position, and assess whether any vendor changes a product because of what seafarers report.

• 🇪🇸 Does Simulating the Transition Actually Flatten the Automation J-Curve? — Algeciras is expanding a semi-automated terminal with a digital twin attached while Hamburg has just booked the cost of converting without one, we monitor whether TTI Algeciras publishes throughput and reliability data through the B1 transition, track whether digital-twin-led transitions show smaller productivity dips than conventional ones, and assess whether lines can eventually price berth risk from a terminal's simulation maturity.

• ⚓ Does the TGAIN Sequence Repeat for Deep-Sea Automated Functions? — OCIMF has written operating guidance for assisted steering eighteen months ahead of legislation taking effect in January 2028, we monitor whether comparable guidance appears for automated collision avoidance, AI-assisted watchkeeping and autonomous berthing, track whether IMO or class bodies adopt the same design-installation-training-procedures-human-factors-cyber structure, and assess how quickly incidents translate into rules for systems that act rather than advise.

📅 Critical Maritime AI Research Areas for Managers

1. 🔍 The True Cost of Duplicate Vessel Inspections — and What Sharing Recovers: RightShip and NorthStandard have made inspection reuse possible without publishing a single figure on what it saves. Research should quantify inspection frequency, overlap and crew hours consumed per vessel per year across dry bulk and general cargo, then measure what a sharing arrangement actually recovers — giving managers a negotiating number and giving the industry a basis for deciding whether trusted digital reports can substitute for physical attendance at scale.
2. 🧭 Independent Measurement of AI's Effect on Seafarer Workload: The STEER survey is the first serious attempt to ask, but self-selecting questionnaires have known limits. Research should pair survey responses with observed data — alerts per watch, task interruptions, decision latency, overtime — across vessel types and technology loadouts, so that claims about AI reducing crew burden can be tested rather than asserted, and so alert-design standards can be built on measurement rather than argument.
3. 🇪🇸 Does Digital-Twin-Led Transition Reduce the Automation Productivity Dip?: HHLA has now put the cost of converting a live terminal into a regulated disclosure while Algeciras expands with simulation attached. Research should compare throughput, reliability and cost trajectories across terminals automating with and without digital-twin-led transition planning, producing a transition-risk model that lines can use to price schedule buffers and demurrage exposure at ports mid-conversion.
4. 🎓 Defining and Certifying AI Competence for Maritime Roles: With class societies now selling AI courses and no benchmark to judge them against, research should define what AI competence means for a superintendent, a technical manager, a master and a watchkeeping officer — the ability to evaluate a use case, interrogate a vendor claim, recognise model failure and know when to override — and propose a certifiable standard that could sit alongside existing STCW and management-system requirements.
5. ⚓ Handover and Control Transfer in Systems That Act Rather Than Advise: OCIMF's guidance insists TGAIN remains a navigation aid, but the recurring failure mode in every automating industry is the handover moment. Research should study how bridge teams detect that an acting system is wrong, how long control resumption actually takes under realistic conditions, and what interface and procedural design shortens it — before the same functions arrive at sea rather than on inland waterways.

📈 Top Investment Opportunities

1. 🏗️ Domestic Port Equipment and AI Terminal Operating Systems, Built for Export — Korea has now committed capital at Gwangyang specifically to prove a domestically developed stack of cranes, transport equipment and port operating systems, with 10% of the global port equipment market as the stated target — the investment opportunity is in autonomous cargo-handling equipment, AI terminal operating systems and the integration layer between them, in a market where Korean and Chinese stacks are now competing directly for the export reference sites that will set buyer expectations for a decade — Gwangyang Physical AI terminal
2. 🖥️ Floating and Offshore AI Compute Infrastructure — Bureau Veritas granting Approval in Principle to a 30 MW data centre barge, with a 100 MW design following, moves floating compute from concept to classed marine asset, and land, power and cooling constraints ashore are not easing — the investment opportunity spans modular data-centre integration, marine power and cooling systems, mooring and offshore support services, and the class and assurance work this asset class will require, with the caveat that AiPs have historically outnumbered deliveries by a wide margin — BV + Seatrium
3. 🔍 Trusted Inspection Data Exchange and Risk Intelligence — an insurer accepting another organisation's inspection report establishes that verified digital condition data can substitute for a physical survey, and every AI risk model in shipping is bottlenecked on exactly that data — the investment opportunity is in secure inspection-data exchange, provenance and audit tooling, and the risk-intelligence platforms that become more valuable as the shared evidence base deepens — RightShip + NorthStandard
4. 🎓 Maritime AI Assurance, Governance and Training Services — with 81% of companies piloting AI, 11% holding scaling policies and a class society now naming data standardisation, governance and scaling as the binding constraints, the scarce capability is judgement rather than software — the investment opportunity is in AI governance tooling, assurance and verification services, and role-specific training and certification for the people who must evaluate, deploy and override these systems — DNV Maritime Academy
5. 💬 Identity, Access and Role-Aware Collaboration for Vessels — Dualog rebuilding ship–shore messaging around individual seafarer identities addresses the unglamorous precondition for every shipboard AI feature and for IACS UR E26/E27 traceability at the same time — the investment opportunity is in maritime identity and access management, crew-system integration, low-bandwidth collaboration tooling and the role-aware assistant layer that only becomes possible once shared accounts are gone — Dualog Workspace

📅 Top Monthly Picks

1. 📧 Sedna Buys Bridge Labs to Turn Ten Million Daily Emails Into Shipping's Operational Layer — Sedna's platform now processes more than 10 million emails a day, and its acquisition of AI engineering firm Bridge Labs — whose founder Alex Crooks becomes Chief Product Officer — is a bet that shipping's real operational record lives in the inbox; Bridge Labs has already shipped five commercial AI products for Louis Dreyfus, Glencore, IFCHOR Galbraiths, MOL Chemical Tankers, Bunge, Oldendorff and Fednav, and Sedna stresses each customer's email data stays under that customer's control — Sedna + Bridge Labs
2. ⚖️ Marcura Buys Fairway Maritime and Becomes the Largest Laytime Processor in the Industry — following the HubSE and Shipdem deals, acquiring the US demurrage and marine-claims specialist makes Marcura Claims the industry's largest laytime processor by volume, running three tiers from self-serve tools through AI-assisted claims to fully managed services, with the AI reading documents, extracting clauses and standardising calculations while human experts supply judgement and correct the models — the clearest working example this period of the divide between what AI reads and what people decide — Marcura + Fairway Maritime
3. 🧮 Veson Folds Its AI Assistant Into the System of Record — With Fuel Predictions Across 1,300+ Ports — the unified Veson Platform brings IMOS together with the CoCaptain AI assistant, AI-powered Mail, Vessel Insights and Bunker Insights in one interface, with core CoCaptain bundled for all IMOS users at launch and Bunker Insights trained on hundreds of thousands of bunker transactions to generate daily fuel-price predictions across more than 1,300 ports — the incumbent system of record becoming the AI layer, with all the lock-in that implies — Veson Platform
4. 🚛 China Merchants Port Hands Berth, Crane and Yard Planning to AI Across Its Global Terminals — one of the world's largest terminal operators is rolling out a Smart Logistics Suite that pulls berth planning, crane scheduling, yard allocation and truck appointments into a single AI-driven layer, recomputing optimal sequences as ships arrive early or late, with an AI berth-planning agent that redraws harbour plans in seconds and a reported truck turnaround of 30 minutes rather than 90 — inside China's state-backed plan to standardise AI stacks across roughly 60 automated terminals through 2030 — China Merchants Port
5. 🔐 One in Eight AI Breaches Now Comes From an Autonomous Agent — at $4.7m Each — HiddenLayer's 2026 AI Threat Landscape Report finds autonomous agents behind more than one in eight reported AI breaches at an average cost of about $4.7m, with 48% of security professionals ranking agentic AI as their top attack vector for 2026 ahead of deepfakes, only 14.4% of AI agents going live with full security and IT approval, 76% of organisations citing shadow AI as a growing problem and 31% unable to say whether they were breached at all — the number to put in front of anyone deploying agents on your fleet — HiddenLayer 2026 report

Have feedback or a story tip? Email us or connect on LinkedIn.