Private LLM hosting for maritime and shipping companies
A dedicated DGX Spark for reading charter parties, voyage logs, and port documentation with a language model that never phones home. The instance is yours alone, in EU-Central, at $0.79/hour for a single node or $1.79/hour for a linked 256 GB cluster.
What shipping companies actually put in front of an LLM
An owner, operator, or charterer that wants an LLM in its workflow is usually working through one of a few recurring document problems: reading a charter party against a fixture recap to catch a clause that doesn't match, summarizing a voyage's noon reports and deck logs into a clean account of what happened, or pulling structured data, port, cargo, laytime, out of a stack of bills of lading and statements of facts. None of this is exotic AI work, but the documents behind it are commercially loaded.
A charter party sets the commercial terms of a fixture: hire rate, laytime, demurrage, off-hire conditions, sometimes with negotiated riders that differ from the standard form. Those terms are usually confidential between owner and charterer, and a broker or a competing operator who saw a run of your fixtures would learn your rate history and your negotiating room. Voyage and deck logs are more operational than commercial, but they still record vessel positions, speeds, bunker consumption, and incidents, the kind of detail an insurer, a P&I club, or opposing counsel in a dispute would want and that you'd rather not have sitting in a third party's chat logs.
Route and routing decisions, weather routing choices, port rotation, bunkering strategy, are closer to institutional knowledge than any single document. They reflect years of a fleet's operating pattern, and feeding that pattern into a shared API a little at a time is a slow way to hand a competitor your playbook.
What dedicated hardware changes
A DGX Spark from GPUwerk is single-tenant: no other customer's workload runs on the machine while it's yours, and the container and workspace are removed from the node before it's offered to anyone else, as described in our Privacy Policy, section 12. You choose the model, whether that's an open-weight model you run yourself or a commercial model you self-host under its own license, and the charter parties, logs, and cargo documents stay on that instance. We don't access, read, copy, index, or analyse what runs on it.
128 GB of unified memory on a single Spark runs a 70B-class model comfortably for the document work most operators need: check a charter party's laytime and demurrage clauses against your standard terms, extract cargo and port data from a bill of lading, or draft a first pass at a voyage summary from noon reports. For heavier work, running extraction across a full year of fixtures or logs for a fleet review, the linked 256 GB cluster gives you more throughput without a different workflow.
Where GDPR applies
Crew lists, master's reports, and some cargo documentation name individuals, a master, an agent, a consignee. If that counts as personal data under the GDPR, our Data Processing Agreement under Article 28 applies to how we handle the infrastructure it runs on; the DPA states plainly that GPUwerk hosts the machine but does not access the content of the instance, and that the controller decides what runs on it. We are not your lawyer and this isn't legal advice: flag questions on charter party interpretation, laytime disputes, or which jurisdiction's rules apply to counsel who knows maritime law. What we can say concretely is where the data physically sits and who can reach it.
Practical starting points
Most operators don't start with a fleet-wide rollout. A common first task is charter party review: comparing a new fixture's terms against your standard form to flag what's different, laytime, demurrage rate, off-hire wording, before it's signed. A 70B-class model on a single Spark handles that comparison well, and you can check its output against a broker's read before trusting it on a live negotiation.
A second starting point is voyage reporting: turning noon reports and deck log entries into a readable summary for a P&I claim or a charterer's query, without the underlying log data leaving your infrastructure. A third is bill of lading and statement of facts extraction, pulling cargo quantity, port, and laytime-relevant timestamps into a structured record your ops team can check against demurrage calculations. None of these need the model to be flawless; they need it to give you a fast, private first pass on documents you'd rather not paste into a public chatbot.
Getting started
Deploy a Spark directly from the console, or if you'd rather scope the workload first, an AI Opportunity Session covers your document types, your existing systems (voyage management, chartering platforms, AIS feeds), and what a first pilot should look like before you commit to a rollout.
Back up before you terminate. Terminating deletes the workspace and GPUwerk's recovery copy of it. Export what you need before you stop paying for it.
Related pages
Run your fleet's document workload on hardware that's only yours.
Single-node Spark from $0.79/hour, no shared GPUs, no egress fees.
Deploy a Spark Book an engagement