The Pattern: Fiber, Cloud, Orbit, Ocean
PriyankaEvery major infrastructure cycle follows the same arc:
Fiber unlocked the internet. Before submarine cables and terrestrial fiber, the bandwidth to run global networks didn't exist. The infrastructure preceded the applications.
Cloud abstracted compute. AWS, Azure, and GCP turned servers into API calls. The infrastructure preceded SaaS.
Orbit is doing the same for connectivity. Starlink has 10 million subscribers. The space economy hit $630 billion in 2025. SpaceX just filed for IPO. Starcloud raised $170 million at a $1.1 billion valuation to build data centers in orbit. Infrastructure preceded the killer apps.
Ocean is next. The same forces -- falling hardware costs, AI-driven autonomy, geopolitical urgency, and capital looking for the next platform layer -- are converging on the deep.
The thesis isn't "ocean resources are valuable." Everyone knows that. The thesis is that the ocean is becoming a programmable infrastructure layer, and the asymmetric returns will accrue to the middleware companies building services on top of it -- not the extractors.
Why Now: Four Forces Converging
1. The Mineral Bottleneck
The energy transition runs on nickel, cobalt, manganese, and rare earths. Land-based reserves are concentrating in geopolitically hostile jurisdictions. The Clarion-Clipperton Zone alone -- a patch of Pacific seafloor between Hawaii and Mexico -- contains more nickel, cobalt, and manganese than all known terrestrial reserves combined. In April 2025, President Trump issued an executive order permitting deep-sea mining in international waters, signaling that the US views subsea minerals as a strategic priority.
2. The Compute Power Problem
Global data center capacity is straining. Cooling is the bottleneck. The ocean is a natural heat sink -- seawater cooling is 40-50% more efficient than air cooling. Microsoft's Project Natick proved the concept: a sealed underwater data center deployed off Scotland's coast showed an 8x lower failure rate than land-based equivalents. Microsoft shelved the project but validated the physics. The underwater data center market is valued at $3.8 billion in 2026 and is projected to reach $36 billion by 2035, growing at 28% CAGR.
3. The Defense Imperative
Subsea cables carry 95% of intercontinental data. They are the most critical and least protected infrastructure on the planet. Russia, China, and others are actively mapping and probing these chokepoints. The submarine cable systems market is valued at $20.5 billion and projected to reach $57 billion by 2034. Autonomous underwater vehicles (AUVs) have moved from research curiosity to strategic military asset. The US Department of Defense is actively investing in low-cost, modular subsea autonomy.
4. The Tech Stack Is Finally Ready
AI-powered autonomous robotics. Advanced materials for deep-pressure environments. Modular hardware architectures. Satellite-linked communication for remote ocean operations. The enabling technology for ocean infrastructure has crossed the threshold from lab to deployment -- exactly where space tech was 10-15 years ago.
The Ocean Stack: A Company Map
Think of the ocean like a cloud stack. Four layers, each with emerging companies staking out territory.
Layer 1: Exploration and Mapping
Before you can build on the ocean, you need to see it. This layer is about autonomous data collection -- mapping the seafloor, monitoring currents, measuring conditions at depth.
Terradepth -- Raised $25 million (Primary Ventures led). Building autonomous submarines that map the ocean floor and collect deep-ocean data. Think of them as the "satellite imagery" company for the ocean -- except looking down, not up. Their thesis: an accurate, continuously updated model of the ocean floor is the foundation everything else gets built on.
Apeiron Labs -- Raised $29 million (latest round). Building swarms of autonomous underwater robots designed for persistent ocean observation. Where Terradepth goes deep, Apeiron goes wide -- flooding ocean zones with low-cost, expendable sensor platforms. The surveillance satellite equivalent for water.
Tethys Robotics -- Raised €3.5 million pre-seed. Zurich-based. Autonomous underwater inspection drones for hazardous environments. Targeting the "last mile" of subsea inspection -- the dangerous, repetitive tasks currently done by human divers.
Layer 2: Energy and Extraction
The resource layer. Energy generation (wave, tidal, OTEC, geothermal) and mineral extraction from the seafloor.
Panthalassa -- Public benefit corporation building a "planetary-scale energy platform from the middle of the ocean." Clean, zero-emission ocean energy designed to power compute infrastructure and more. Team comes from SpaceX, Google, Blue Origin, Tesla, NASA, Boeing. Currently running a Pacific Ocean pilot deployment (Ocean-2). The framing is deliberate: this isn't an energy company, it's an infrastructure platform.
CorPower Ocean -- Raised €32 million Series B1. Commercializing wave energy converters. The DOE has invested $41 million in marine energy R&D through the National Marine Energy Centers. Wave and tidal are still early but gaining real traction as baseload power sources.
Deep-sea mining operators -- The Metals Company, UK Seabed Resources, and others are positioning around polymetallic nodule extraction. Politically contentious, but the mineral economics are becoming impossible to ignore as EV and battery demand scales.
Layer 3: Subsea Infrastructure
The physical backbone -- data centers, energy storage, cable networks, and maintenance systems that operate at depth.
Underwater data centers -- The $3.8 billion market is being built by companies like Highlander (China's operational subsea data center), Subsea Cloud, and others picking up where Microsoft's Natick left off. The value proposition is simple: free cooling, proximity to submarine cable landing points, and lower failure rates.
Submarine cable operators -- SubCom, NEC, Alcatel Submarine Networks. The $20.5 billion market is being driven by hyperscaler demand (Google, Meta, Microsoft all building proprietary submarine cable systems) and the need for redundancy against sabotage.
Layer 4: Middleware and Autonomy Software
This is where the alpha likely lives. The companies building the operating systems, fleet management platforms, and data services that make the ocean programmable.
Ulysses -- a16z Series A (April 2026, American Dynamism fund). Irish-founded, San Francisco-based. Building modular autonomous underwater vehicles -- "LEGO-like" architecture where you snap on capabilities (more range, more compute, different sensors) as needed. Started in commercial markets (their first competitor was "a human diver with a shovel planting seagrass"), then got pulled into defense. The platform is fundamentally modular -- an API for underwater operations. This is the middleware thesis incarnate.
Vatn Systems -- Raised $60 million Series A (December 2025). Now the largest AUV manufacturer in the US. Building autonomous underwater vehicles for military and commercial applications at scale. The defense pull is real: the Navy needs low-cost, expendable subsea assets, not $3 billion Virginia-class submarines.
Nauticus Robotics (KITT) -- Raised $12 million. Building "augmented autonomy" for subsea robots -- AI-powered software that makes underwater robots smarter and more independent. The operating system layer for ocean robotics.
Where the Alpha Is
The pattern from every prior infrastructure cycle holds: the platform layer wins.
In cloud, AWS beat the server manufacturers. In space, the satellite service companies are outperforming the rocket builders. In ocean, the same dynamic is emerging.
Raw extraction (mining) is capital-intensive, politically volatile, and geopolitically contested. Energy generation (wave, tidal) has long development cycles. Physical infrastructure (cables, data centers) requires massive capex.
But the middleware -- the autonomy software, the fleet orchestration, the ocean data APIs, the modular hardware platforms -- is where venture-scale returns are possible:
- Low capex, high margin. Software and modular hardware scale differently than mining rigs.
- Platform dynamics. Once you're the operating system for underwater operations, every new use case (defense, energy, mining, research, cable maintenance) runs through you.
- Defense as a beachhead. Military budgets de-risk the technology. Commercial markets provide the scale. Ulysses, Vatn, and Nauticus all demonstrate this dual-use playbook.
The Risks
Environmental backlash. Deep-sea mining is politically radioactive. The International Seabed Authority has yet to finalize regulations. Any company in the extraction layer faces regulatory and reputational risk.
Capital intensity. Ocean hardware is expensive. Saltwater is corrosive. Pressure is unforgiving. Maintenance cycles are long. The ocean punishes bad engineering.
Regulatory fragmentation. Who owns the ocean? The UN Convention on the Law of the Sea governs international waters, but enforcement is thin. The Trump executive order on deep-sea mining signals that unilateral action is the direction of travel -- which creates uncertainty, not clarity.
Long development cycles. Unlike software, ocean infrastructure can't iterate in two-week sprints. Hardware needs to survive at depth for years.
What We're Watching
The ocean is where space was in 2012. The physics works. The economics are starting to work. The geopolitics demand it. The question is timing.
Signals that the thesis is accelerating:
- More a16z-scale checks into subsea companies (Ulysses is the leading indicator)
- Hyperscalers building or acquiring underwater data center capacity
- DOD budget line items for autonomous underwater systems growing
- Subsea cable sabotage events increasing urgency for monitoring infrastructure
- Ocean data marketplaces emerging (the "Planet Labs for water")
We mapped the sky. Now we need to map the water. Article one is live. Now let me research and write the second thesis -- Machine Markets. Strong data. Let me get one more round of specifics on the AI x crypto gaming angle and stablecoin usage. Incredible data. $28 trillion through the agent economy, $479M in "Agentic GDP," ERC-8004 live on mainnet, 150K+ agents registered. This thesis is already materializing. Let me write the second article now. Both articles are live. Here's your full set for this week's call email: