A Multi-Megawatt Prototype at Sea: Drogue's 2028 Plan for AI Compute

The most urgent constraint for scaling artificial intelligence is not the silicon, but the power and cooling needed to run it. A new startup, Drogue, is proposing a radical relocation to solve both problems at once: moving the data center to the ocean. Founded in 2026 and emerging from stealth in November with a $20 million seed round, the company is developing floating offshore platforms that use the ocean's own thermal gradient to generate electricity and provide cooling for AI workloads [TechCrunch, November 2026].

It is a deeply technical, capital-intensive bet on a novel form of energy generation, ocean thermal energy conversion (OTEC), married to one of the world's most power-hungry industries. The premise is that by colocating power generation and compute at sea, Drogue can bypass land-based grid interconnection queues and use the ocean as a nearly infinite heat sink. The company's first multi-megawatt prototype is slated for deployment aboard a conventional ocean-going ship in early 2028 [TechCrunch, November 2026].

The technical wedge: OTEC meets AI compute

Drogue's core innovation is not the computing hardware, but the integrated energy system designed to support it. Ocean thermal energy conversion is a century-old concept that exploits the temperature difference between warm surface water and cold deep water to drive a heat engine and generate electricity. For a traditional power plant, the efficiency is often too low to be economical. For a floating data center, however, the calculus changes. The same seawater that provides the temperature differential for power also serves as the direct cooling medium for the servers, eliminating the need for massive, energy-intensive chiller plants [TechCrunch, November 2026].

The company is targeting what it sees as a critical wedge: the multi-year delays plaguing new land-based data center construction, largely due to strained electrical grids and lengthy permitting processes. By moving offshore, Drogue argues it can deploy scalable compute capacity faster, and in closer proximity to coastal population centers and undersea fiber optic cables, without waiting for grid upgrades [TechCrunch, November 2026].

The solo founder and his coalition

Drogue is the vision of solo founder and CEO Sam Sipe, whose background stitches together threads from law, defense, and AI infrastructure. A former partner at law firm Steptoe LLP, Sipe later served as head of engineering for the public sector at Scale AI [Bloomberg Markets, retrieved 2026] [Linear, Unknown]. Perhaps most critically for a venture involving seaborne deployment, he is a former U.S. Navy pilot who flew E-2D Hawkeye aircraft from aircraft carriers, an experience that provides firsthand operational knowledge of complex systems at sea [MLQ, November 2026].

While no other technical co-founder is named in public reporting, the company has attracted a network of advisors and supporters from adjacent fields, including robotics, solar construction, and compute infrastructure [Rick Marquardt LinkedIn, retrieved 2026] [Andrew Zellman LinkedIn, retrieved 2026] [Sam Denton LinkedIn, retrieved 2026]. The investor syndicate, which includes XYZ Venture Capital, Toyota Ventures, and Tamarack Global, suggests a coalition betting on deep tech infrastructure over pure software [TechCrunch, November 2026].

The capital and the roadmap

The $20 million seed round, disclosed in November 2026, provides the war chest to move from design to a physical prototype. The company is actively hiring for core engineering roles in thermodynamics, electrical systems, and simulation, indicating a push toward hardware integration and testing [Drogue careers page, November 2026].

The stated roadmap is aggressive. Drogue plans to acquire a ship, outfit it with its multi-megawatt OTEC and computing system, and send it to sea for initial deployment within roughly 15 months of its funding announcement [MLQ, November 2026]. This would represent a significant acceleration for a first-of-its-kind deep tech system.

Seed Round (Nov 2026) | 20 | M USD

The formidable counterfactuals

For all its ambition, Drogue's path is lined with technical, commercial, and regulatory hurdles that have historically stalled OTEC projects. The company's public disclosures to date are strong on vision but light on the granular details that would allow for independent technical assessment. No named data center customer or power offtake agreement has been announced, leaving the commercial model unproven [Proofstory, November 2026].

The technical risks are substantial and can be broken into three core challenges:

• System reliability. OTEC systems require massive seawater intake pipes and heat exchangers that are vulnerable to biofouling and corrosion in a harsh marine environment. Maintaining high availability for critical compute workloads in this setting is an unsolved problem at scale.
• Economic viability. The capital expenditure for a ship and a custom OTEC plant is enormous. The company has not disclosed projected costs per megawatt or compute unit, making it difficult to compare its economics to land-based alternatives, even with grid delays.
• Operational complexity. Running a high-density data center on a moving vessel introduces unique challenges for network latency, physical stability, and remote maintenance that terrestrial facilities do not face.

Drogue's most plausible answer to these concerns is its timeline. By aiming for a ship-based prototype by early 2028, the company is forcing a near-term, real-world test. Success will be measured not by a whitepaper, but by the prototype's ability to generate stable power, provide effective cooling, and run commercial AI workloads for a sustained period.

What to watch before 2028

The next twelve months will be decisive for Drogue. The key milestones are less about software updates and more about heavy industry: securing a ship, building and integrating the OTEC plant, and locking in a first customer or partner willing to beta-test its compute capacity. A partnership with a cloud provider, a large AI lab, or a data center operator would be a strong signal of commercial intent.

Financially, the $20 million seed is significant, but it is likely only a down payment on the capital required to build and deploy a multi-megawatt floating platform. A larger Series A round, potentially involving strategic investors from the maritime, energy, or compute sectors, is a probable next step as prototype costs mount.

The patient population here is the global AI industry, suffering from acute power anemia. The standard of care today is a frantic race to secure gigawatts of power, often by building new substations, lobbying utilities, and navigating years of interconnection studies. It is a land-bound, grid-constrained process that is becoming a primary bottleneck for innovation. Drogue is proposing a departure from that paradigm, betting that the solution lies not in fighting for space on the grid, but in taking the compute to where the power and cooling are inherently abundant. The ocean, long a source of energy and a sink for waste heat, may yet become the foundation for the next phase of AI.

Sources

1. [TechCrunch, November 2026] Exclusive: This young startup wants to power floating data centers with the ocean’s heat, and investors are already on board | https://techcrunch.com/2026/11/13/this-young-startup-wants-to-power-floating-data-centers-with-the-oceans-heat-and-investors-are-already-on-board/
2. [MLQ, November 2026] Drogue raises $20 million to power floating AI data centers with the ocean’s heat | https://mlq.ai/news/drogue-raises-20-million-to-power-floating-ai-data-centers-with-the-oceans-heat/
3. [Bloomberg Markets, retrieved 2026] Sam Sipe, Steptoe LLP: Profile and Biography | https://www.bloomberg.com/profile/person/3800927
4. [Linear, Unknown] Sam Sipe profile | (source referenced)
5. [Proofstory, November 2026] Drogue research review | https://proofstory.co/research/drogue/index.html
6. [Drogue careers page, November 2026] Open positions | https://jobs.ashbyhq.com/drogue
7. [Rick Marquardt LinkedIn, retrieved 2026] Profile | https://www.linkedin.com/in/rick-marquardt-6381a857/
8. [Andrew Zellman LinkedIn, retrieved 2026] Profile | https://www.linkedin.com/in/andrew-zellman/
9. [Sam Denton LinkedIn, retrieved 2026] Profile | https://www.linkedin.com/in/sam-denton-/