From the Seabed Up: Terradepth and the Ocean Operating System

The problem with maritime autonomous systems is not the robots. The problem is what you do with the data they collect, how you move it, aggregate it, make sense of it, and deliver it to someone who can act on it, in time for that action to matter. For years, the defense and commercial maritime communities have talked themselves into believing that acquiring the platform was the same thing as acquiring the capability. It is not.

Joe Wolfel, Founder and CEO of Terradepth, has built his company around this distinction. In a recent conversation, Wolfel, a Naval Academy graduate and former SEAL officer who spent time at the McChrystal Group before founding Terradepth in 2018, laid out a vision that is less about robots than about ecosystems: an Ocean Operating System that turns fragmented subsea data into persistent, decision-ready intelligence for defense and commercial users alike.

The Gap No One Closed

Terradepth’s founding premise was straightforward: the ocean represents 71 percent of the planet’s surface, most of it unmapped and poorly understood, with whatever data exists scattered across hard drives, siloed by organization, and accessible to almost no one n real time. The technology to change this, autonomous underwater vehicles, edge computing, cloud-native data platforms, was beginning to mature.

“The company’s conviction,” Wolfel explains, “is that if you can build an operating system, an ecosystem that creates a fast, effective flywheel of ocean information, you can really create better, faster decision making.” The Ocean Operating System has three layers: data acquisition (autonomous vehicles and partner platforms), a data aggregation and visualization platform (Absolute Ocean), and applications built on top of that platform for specific operational or commercial use cases.

The data problem he describes is not abstract. Before Absolute Ocean, even organizations that were collecting subsea data were not aggregating it anywhere useful. “All this ocean data, to the degree that you were collecting it, sat on a hard disk drive somewhere. That is not usable, it’s not scalable, and it does not, in a military context, provide a common operating picture at all for commanders or decision makers.”

Absolute Ocean provides a unified repository that ingests sonar, oceanographic, and unmanned system data, enables near-real-time analysis, and supports collaboration across distributed teams.

The Robot Is Not the Capability

The conceptual heart of Wolfel’s argument and the place where Terradepth’s approach most sharply diverges from how the defense community typically acquires autonomous systems is the distinction between a product and a capability. The defense instinct, reinforced by decades of platform-centric acquisition, is to ask for the robot. Give me the UUV. Tell me where to deploy it. Hand it off to a contractor who will collect data and deliver it on a hard drive six weeks later.

“A robot by itself does not create a capability,” Wolfel says flatly. “The robot exists within an ecosystem. It provides an important part of the capability, data acquisition, but you need the entire ecosystem. Now I’ve collected the data. How am I processing it? How am I turning it into something that humans can ingest and make decisions from? And how do I sustain this in an economical way?”

The answer to each of those questions is not another platform. It is the operating system.

The analogy Wolfel reaches for is the commercial space economy. “We’re probably 20 to 30 years behind the space economy in maritime, for a whole bunch of reasons. But I think we’re starting to see that change.” In the space domain, the transition from government-owned, government-operated satellite systems to company-owned, company-operated constellations that sell data by subscription transformed what was available to whom and at what speed. The same transition from “procure the platform” to “pay for the intelligence the platform generates” is what Terradepth is working to accelerate in the maritime domain.

Absolute Ocean and Lattice: Not Competitors, but Complements

A natural question in discussions of maritime autonomous systems is how different software architectures relate to each other and specifically how a data platform like Absolute Ocean relates to command-and-control frameworks like Lattice, the Anduril-developed autonomous systems management architecture increasingly present in U.S. defense programs.

Wolfel is clear on the distinction: “Our software architecture isn’t competitive with Lattice. It’s actually very compatible. What Lattice is, is a command-and-control mechanism for multiple autonomous assets. What Absolute Ocean provides is the environmental ground truth layer that enables Lattice to command and control those autonomous assets with a higher degree of efficacy because now the robots understand exactly what environment they’re operating within.”

This is not a minor point. One of the persistent problems in the autonomous maritime domain is that every vehicle manufacturer wants to run its own C2 system for its own vehicles, creating interoperability problems that prevent the family-of-systems approach that operational commanders actually need. Absolute Ocean’s positioning as the environmental intelligence layer rather than another command-and-control stack makes it a potential connective tissue for a broader ecosystem rather than another vertical silo.

AI at the Edge: Why Subsea Autonomy Is Different

The physics of operating underwater create a problem that surface and air autonomy do not face: once a vehicle goes below the surface, the high-bandwidth connectivity required for remote control or real-time human command largely disappears. An autonomous surface vessel can in many cases be operated as a remotely controlled boat because it has surface connectivity. An AUV operating at depth cannot.

This makes on-board AI not a feature but a necessity. “Agentic AI residing on premise, controlling vehicle behaviors based on incoming in situ information, is actually really important,” Wolfel explains. “The shift from generative AI to agentic AI does look promising for making what I would classify as our dumb underwater robots a lot more capable.”

The vision is not AI as a decision-maker but AI as an enabler of genuine autonomy, a vehicle that can observe its environment, identify features of interest, and retask itself without waiting for instructions from an operator who may be hours away and a thousand meters above.

The Operational Demand Signal: Ports, Cables, and the Arctic

The strategic case for persistent subsea intelligence has sharpened considerably in recent years. Nord Stream demonstrated that critical undersea infrastructure, pipelines, fiber-optic cables, energy chokepoints, is vulnerable in ways that had been systematically ignored. Mining of maritime chokepoints through which the majority of seaborne energy transits is no longer theoretical. Port security is an extant challenge and problem.

“People think that because it’s underwater, it’s just down there and we don’t really think too much about it,” Wolfel observes. “Nord Stream showed us that’s not the case. The cable issues showed us that’s not the case. Mining critical choke points where 70 percent of our seaborne energy goes through is showing us that’s not the case.”

Terradepth’s commercial client base — offshore energy, submarine fiber-optic cable operators, critical offshore infrastructure owners — is built precisely around this problem set. The defense use case extends the same capability into contested waters.

The near-term operational opportunity, as Wolfel sees it, is not global ocean mapping but area-specific, operationally relevant coverage built around defined problem sets: submarine base perimeter security, critical infrastructure monitoring, littoral approaches in the Indo-Pacific. The technology is not yet ready to provide continuous global coverage, nor does it need to be. What it can provide now is persistent, real-time environmental intelligence for specific areas of operational significance, sustained 24/7 in ways that neither manned surveys nor episodic UUV deployments can replicate.

Capability Delivery, Not Prototype Production

Wolfel is emphatic that Terradepth is not a prototype company. “Part of our narrative is that we’re not another defense tech startup chasing prototypes. We’re focused on being a lifecycle capability provider.” The distinction matters. The defense community has a well-documented tendency to fund demonstration after demonstration without ever transitioning a capability into sustained operational use. The result is an ecosystem of technologies that have been shown to work but are never actually fielded at scale.

What Terradepth offers instead is vertical integration across what has historically been a fragmented supply chain: vehicle design, data collection, edge processing, cloud aggregation, and application development, all within a single ecosystem that can be operated under a subscription or data-services model. The government does not need to own the robots, manage the fleet, or process the data. It needs the intelligence the system generates, delivered when and where it is needed, at a price it can sustain.

“Thinking in terms of capability delivery instead of product delivery” that is Wolfel’s formulation. It is also, as he notes, the model that has proven most successful in analogous domains: the company-owned, company-operated constellations of the commercial space economy, the Scan Eagle data services model that gave the Marine Corps persistent ISR in Afghanistan without locking the Navy into a government-owned asset it could not evolve, the Australian Maritime Border Command model of tightly coupling user communities directly with technology providers, eliminating the acquisition intermediary that typically prevents either side from learning fast enough.

The Ocean Operating System as Kill Web Infrastructure

Placed within the broader framework of distributed maritime force design — the security webs, deterrence webs, and kill webs that define the operational logic of the Indo-Pacific theater — what Terradepth is building looks less like a defense technology product and more like foundational infrastructure. Kill webs require domain knowledge. They require persistent situational awareness across the battlespace. They require environmental ground truth that tells distributed forces not just where adversaries are but what the operating environment constrains and allows.

An Ocean Operating System that provides persistent, AI-enabled subsea intelligence for specific operational areas, compatible with C2 frameworks, accessible at the edge without cloud connectivity, scalable through a data-services model that does not require government ownership of the collection fleet, is exactly the kind of capability that the undersea dimension of that architecture currently lacks.