Panthalassa closed a $140 million Series B on May 4, 2026, led by Peter Thiel, with participation from John Doerr, Marc Benioff's TIME Ventures, Max Levchin's SciFi Ventures, Susquehanna Sustainable Investments, Hanwha Asset Management (USA)'s venture fund, Anthony Pratt, Fortescue Ventures, Future Positive, WTI, Nimble Partners, Super Micro Computer, Sozo Ventures, Dylan Field, Planetary VC, Leblon Capital, Resilience Reserve, Portland Seed Fund, and Intrepid Oregon Fund.

Returning investors in the Series B included Founders Fund, Gigascale Capital, Lowercarbon Capital, Unless, and WovenEarth.

Prior to the Series B, Panthalassa had raised approximately $70 million across earlier rounds. Total capital raised as of May 2026 stands at approximately $210 million.

Panthalassa builds autonomous offshore nodes, large steel marine platforms that sit mostly submerged in the open ocean, harvest wave energy, and use that power on site to run AI compute rather than sending electricity back to shore.

The core product is the Node. As ocean swells lift and lower the structure, water moves through internal fluid systems and drives a turbine-generator inside the hull. That electricity powers onboard compute hardware sealed within the node, while surrounding seawater provides passive cooling. The system is self-contained: wave energy in, AI inference results out.

Rather than exporting electrons over expensive subsea cables, Panthalassa exports data. Inference outputs travel back to land via low-Earth-orbit satellite links, shifting the constraint from energy transmission to data transmission.

The current generation is Ocean-3, with pilot deployments planned for the northern Pacific in 2026. Earlier prototypes, Ocean-1, Ocean-2, and Wavehopper, validated the power generation, propulsion, autonomy, and at-sea computing subsystems in 2021 and 2024. Ocean-3 nodes are designed to operate together as a distributed offshore compute cluster, with the ocean acting as a cooling sink for the GPU-class hardware inside each hull.

Each node is self-propelled and autonomous. It does not require a fixed anchor or seabed mooring, and onboard navigation and control systems allow it to hold station or return to port when needed. Panthalassa is also building the operating stack around the hardware, including fleet control software, node controllers, deployment and retrieval equipment, and cloud-based telemetry pipelines for continuous fleet monitoring.

For customers, the product is closer to managed offshore compute capacity than marine hardware. The relevant variables are available inference throughput, service reliability, and cost per compute unit, while Panthalassa manages the underlying stack from wave energy conversion to satellite backhaul.

Panthalassa is a vertically integrated infrastructure platform that owns the stack from marine structure and wave-energy conversion to autonomy, onboard compute, fleet operations, and satellite-linked delivery of results.

Its go-to-market is B2B, targeting AI infrastructure buyers and compute operators that face power, cooling, or siting constraints on land. The monetization model is not selling commodity electricity into a grid, but converting offshore wave energy into a higher-value service, powered offshore compute capacity. Customers are expected to buy capacity or outcomes, reserved node throughput, hosted inference workloads, or dedicated fleet allocation, rather than purchasing hardware outright.

That design choice addresses a core constraint in wave energy economics. Traditional wave-energy projects struggle because generating power offshore is only half the problem, getting it ashore through subsea transmission is the other half. Panthalassa removes that step by colocating the load with the energy source, which also avoids land acquisition, grid interconnection queues, and freshwater cooling costs.

The cost structure is capital-intensive and hardware-heavy up front, but it avoids several cost buckets that burden terrestrial data centers while introducing others, including marine engineering, offshore deployment and retrieval, corrosion protection, biofouling management, vessel logistics, and remote maintenance. The key margin question is whether those offshore operating costs remain manageable at fleet scale. Panthalassa's target node manufacturing cost is roughly $1 million to $1.5 million per unit, and the company is building toward factory-style production from plate steel at a pilot manufacturing facility near Portland rather than treating each node as a bespoke offshore project.

The model depends on a feedback loop: AI infrastructure bottlenecks increase the value of alternative powered capacity, that demand attracts capital, capital funds manufacturing iteration that wave-energy startups have historically never received, manufacturing scale lowers node cost and improves reliability, and better economics make offshore compute viable for a broader set of workloads. In this framing, compute demand is the catalyst that could allow wave energy to reach commercial scale.

Panthalassa operates across marine energy, floating offshore infrastructure, and AI compute capacity, so its competitive set extends beyond wave-energy companies. The closest comps range from offshore compute platforms to marine power providers and shipbuilding-led floating data center projects.

The closest direct rival is Aikido Technologies, which is also selling offshore AI compute but built on a semi-submersible floating offshore wind platform with battery storage rather than a wave-powered autonomous node. Aikido's bet is that floating wind supply chains and pre-designated offshore sites are more bankable and faster to scale than a wave-first system. For customers prioritizing deployment certainty and megawatt-scale near-term capacity, Aikido is a credible alternative even if Panthalassa's long-run cost thesis is stronger.

SeaPower is earlier-stage but conceptually close, pursuing autonomous offshore data-center platforms stabilized and powered by wave energy. Its presence means the same questions Panthalassa must answer, survivability, mooring durability, and offshore O&M economics, will be asked of any wave-powered compute entrant, and that category ownership is still unsettled.

C-Power and Ocean Power Technologies (OPT) represent a different approach: autonomous offshore power for defense, remote industrial loads, and maritime intelligence rather than floating compute. OPT in particular has over a decade of field deployments and an established maritime customer base, which gives it operating credibility that Panthalassa still needs to build. The risk is less that these players will match Panthalassa in offshore AI capacity, and more that their deployment histories make Panthalassa's full-stack compute bet look higher-risk by comparison. Saildrone, which Sacra estimates reached $43 million in revenue in 2024 by monetizing an ocean platform across climate science, offshore energy, and defense customers, indicates that ocean platforms may commercialize faster when they serve multiple demand pools rather than waiting for a single end market to mature, a point relevant to Panthalassa's sequencing choices.

CalWave and CorPower Ocean reinforce a related dynamic: wave-energy companies are generally pursuing stepwise energy commercialization with third-party validation and utility procurement, rather than Panthalassa's combined energy-plus-compute leap. CorPower has already exported power to grid, demonstrated survivability in extreme wave conditions, and laid out funded farm roadmaps through 2030. That progress strengthens the case that wave companies often become credible energy suppliers before adding compute, a question Panthalassa will need to answer with Ocean-3 field results.

The newest threat class is industrial-scale entrants with no wave-energy ambition. Samsung Heavy Industries received classification body approval for a 50 MW floating data center design in May 2026, and a partnership with Mousterian gives it a U.S. commercial path. MOL and Hitachi are pursuing a repurposing model, converting second-hand vessels into floating data centers. HiCloud's offshore-wind-powered underwater data center in Shanghai reached full commercial operation at 24 MW in May 2026, with a roadmap to 500 MW.

These players are effectively arguing that customers mainly want coastal capacity with strong cooling and less land pressure, and that offshore wind or cable power can satisfy the use case while avoiding Panthalassa's autonomy and wave-harvesting complexity. Nautilus Data Technologies adds another angle: liquid-cooled floating data center infrastructure with tested support for next-generation AI GPU clusters, no independent generation required. If hyperscalers prefer vendor modularity over one integrated offshore node provider, these disaggregated approaches could win on deployment speed even if they lack Panthalassa's integrated energy thesis. Microsoft's Project Natick, now discontinued, validated parts of the ocean compute concept while also reminding buyers that large tech firms have already explored this space and stepped back when serviceability economics did not justify continuation.

Panthalassa's expansion logic runs in two directions: more workloads per node and more end markets for the node itself. The company's TAM broadens if it can raise revenue per deployed node while also selling into buyers beyond AI infrastructure.

AI inference is the initial wedge because it is geographically flexible and can tolerate the satellite-linked backhaul that offshore nodes require. But the node architecture is designed as a general offshore energy platform, not an inference-only appliance. Earlier company materials pointed to synthetic renewable fuel production as a second use case, with product shipped back to shore rather than data. If the node can produce green hydrogen, ammonia, or other energy-dense fuels offshore and export them by vessel, the addressable market expands beyond compute into industrial decarbonization.

The DOE's marine energy program identifies remote communities, ocean-based industries, and desalination as natural applications for offshore power systems, all of which map onto Panthalassa's autonomous, untethered architecture. Each additional payload type that can be hosted on a node increases revenue per unit of deployed capital and reduces dependence on any single end market.

Today's likely early customers are AI infrastructure buyers constrained by grid access, cooling, or siting. But the same node attributes, off-grid power, autonomous operation, and modular deployment, are also relevant to defense, maritime surveillance, disaster response, and sovereign compute use cases. Panthalassa's team background spans aerospace, defense, and military institutions, and the node's combination of autonomy, remote power, communications, and offshore persistence aligns with government procurement priorities. The dual-use framing that has driven valuation premiums for maritime autonomy companies like Saronic and Anduril is also available to Panthalassa if it chooses to pursue it.

As terrestrial grids tighten, DOE projects data centers could reach up to 9% of U.S. electricity generation by 2030, and the buyer pool for alternative powered capacity widens. Panthalassa can sell not just green offshore compute, but capacity that exists when terrestrial capacity does not, a distinct value proposition for enterprise and government buyers facing power constraints.

Panthalassa's near-term geographic plan is the northern Pacific, targeting the world's most energy-dense wave zones rather than following conventional data-center hubs. That geographic freedom is a structural advantage: land data centers are constrained by transmission queues, water availability, and local opposition, while Panthalassa's nodes can be deployed where wave energy is abundant and cooling is intrinsic.

The manufacturing expansion path is industrial as much as oceanic. Nodes are mass-produced from plate steel in coastal factories, and the Series B funds a pilot manufacturing facility near Portland. If the company reaches repeatable node production, it can expand by replicating that manufacturing template in other coastal regions and pairing factory rollout with nearby deployment corridors. Fluidstack's experience shows that compute buyers will sign large, long-duration capacity agreements when critical IT load is guaranteed. The same dynamic could support Panthalassa's factory scaling if Ocean-3 delivers credible uptime data. Super Micro Computer's participation in the Series B also creates a hardware partnership path for ruggedized, marine-optimized server packaging that could accelerate the compute payload side of the platform.

Offshore O&M economics: Saltwater environments impose corrosion, biofouling, storm survivability, and remote maintenance costs that could overwhelm the unit economics advantage of free wave energy and free ocean cooling if nodes require frequent marine intervention instead of the rare-repair, self-return operation Panthalassa is targeting.

Workload-market fit: Satellite-linked offshore inference is better suited to latency-tolerant, geographically flexible workloads than to the bandwidth-heavy, tightly networked training and cloud workloads that currently represent the largest and fastest-growing slice of AI infrastructure demand, which could confine Panthalassa to a narrower addressable compute market than its TAM narrative implies.

Integration risk: Panthalassa must simultaneously prove wave-energy reliability, marine autonomy, data-center uptime, and scalable offshore manufacturing, a stack of hard problems that most rivals attack one at a time, while competitors like Aikido, HiCloud, and Samsung Heavy can reach customers faster by borrowing proven components from floating wind, shipbuilding, and conventional data-center supply chains.