The Great Blue Frontier — Notboring 重點整理

Ocean industrialization thesis: the last great frontier is pre-industrial

—The ocean is at the same inflection point space was when SpaceX was founded — a $2.6T pre-industrial economy running on an 1800s expedition model, where expensive humans on expensive ships do expensive campaigns then retreat to dry land
—Four technology curves crossed critical thresholds simultaneously (2020-2025): Starlink maritime connectivity (100x price-performance improvement), battery energy density doubled + costs collapsed 90%, onboard AI compute democratized, and EV/drone supply chains made AUV components 10-100x cheaper — this is Packy's 'Curve Convergence' applied to the ocean
—The governing metric is the cost of ocean work ($/km²-month); today it costs millions per month to cover a meaningful area with incumbent methods; collapsing that metric unlocks everything downstream

AI基建 ↑看多長期

Ocean infrastructure is the next AI基建 buildout frontier — the same trends (cheap compute, autonomy, vertical integration, scale pricing) that defined data center build-out are now enabling autonomous ocean systems

Undersea cable demand and sabotage risk: Meta Waterworth, Big Tech buildout

—99% of intercontinental internet traffic flows through subsea cables, yet cables break 100-200 times/year and over a million km sit on the ocean floor with essentially zero persistent surveillance
—Baltic Sea sabotage (Nov 2024), Estlink cable cut (Christmas 2024 by Russian shadow tanker), UK formally accusing Russia of covert operations over cables and pipelines — deliberate sabotage is escalating from isolated incidents to systematic strategy #風險
—NATO launched Operation Baltic Sentry in Jan 2025 specifically because no assets existed to detect or deter these attacks — infrastructure gap is acute

$META ↑看多長期

Announced Project Waterworth in 2025 — a 50,000km subsea cable spanning five continents, the longest in history, driven by AI data demand #催化劑

AI基建 ↑看多長期

AI demand is driving a near-doubling of subsea cable investment between 2022-2027; tech giants (Meta, Google, Amazon, Microsoft) now control roughly half the subsea cable market, building their own transoceanic routes #催化劑

Critical minerals / deep-sea mining: TMC files first commercial license

—The Clarion-Clipperton Zone alone contains more cobalt than all known land reserves, plus polymetallic nodules rich in nickel, manganese, and copper — exactly the minerals EV batteries, wind turbines, and AI hardware are consuming at accelerating rates
—The regulatory dam is cracking: after years of ISA stalemate, The Metals Company bypassed the ISA via a US 1980 regulatory pathway and filed the first-ever commercial deep-sea mining license application #催化劑

$TMC ↑看多長期

First mover in commercial deep-sea mining: filed the first-ever commercial license via a US regulatory pathway in 2025, bypassing the stalled ISA process; estimates NPV of first two projects at $24B #催化劑 #價位

Maritime defense: US Navy autonomy budget surge, incumbent primes as Innovator's Dilemma victims

—Strait of Hormuz conflict (April 2026) is a live demonstration of the infrastructure gap: Iran is unable to find the mines it just placed because it lacks persistent ocean surveillance — the same blindness that affects everyone
—Houthi Red Sea attacks forced global shipping to reroute around Africa, adding 11,000nm and ~$1M per ship; container rates Shanghai-to-Europe quadrupled; $1T of goods disrupted in first six months — maritime domain awareness has become a strategic imperative #風險
—FY2026 Pentagon budget: $13.4B for autonomous systems as its own budget line for the first time; Navy alone gets $5.3B, a 70% YoY increase #催化劑
—China is building submarine-sized AUVs with 18,000km range, cable-cutting tools that operate at 4,000m depth, and an 'Underwater Great Wall' sensor network across the Indo-Pacific #風險

$LMT ↓看空長期

Classic Innovator's Dilemma: ocean defense primes (Lockheed, RTX, L3Harris, Leidos, GD) sell cost-plus, crewed, campaign-based systems; their best customers (navies) demand mil-spec redundancy which prevents pivoting to autonomy-first models; a new cohort will take share as autonomous systems become good enough #比較

$RTX ↓看空長期

Incumbent maritime defense prime facing Innovator's Dilemma from autonomy-native entrants; rationally serves current customers well, but incentivized structure prevents pivoting to scale-priced autonomous platforms #比較

$LHX ↓看空長期

Incumbent maritime defense prime; same Innovator's Dilemma dynamic as $LMT and $RTX — cost-plus model and crewed-vessel heritage are structural disadvantages against autonomy-native entrants #比較

$LDOS ↓看空長期

Incumbent subsea services prime (alongside Oceaneering, Fugro, Subsea 7); campaign-based vessel model is structurally disrupted by persistent autonomous fleets that collapse the $/km²-month cost basis #比較

$GD ↓看空長期

General Dynamics Electric Boat is a legacy shipbuilder (cost-plus, multi-year hull contracts) — the model Saronic and Blue Water Autonomy are rebuilding from scratch around autonomous surface vessels #比較

New Ocean cohort: autonomy-native companies as the New Space analogy

—The 'New Ocean' cohort shares characteristics: vertically integrated (not subcontracted), autonomy-native (not crewed-first), software-defined, built on commercial supply chains (drone motors, EV battery packs, Jetson-class compute), priced for scale not scarcity
—Ulysses (private, $46M from a16z American Dynamism) is building full-stack: Mako AUV ($50k vs $500k-$5M incumbent), Leviathan autonomous surface craft, Kraken autonomous launch/recovery platform — 10-100x cost reduction vs incumbents
—The AUV market produces ~1,000 vehicles/year at $500k-$5M each; Ulysses claims to build Makos for $50k — a demand request just came in for 10,000 vehicles as a service and another for 1,000
—Kraken Robotics (TSX-V: PNG) is cited as an infrastructure enabler — builds synthetic-aperture sonar that turns a cheap AUV into a survey-grade mapping platform

AI算力 ↑看多長期

Edge compute is central to ocean autonomy: because underwater comms bandwidth is ~1M times slower than 5G, AUVs must run data-center-class GPUs onboard to make autonomous decisions without surfacing — AI算力 demand extends from cloud to seabed

每天早上,收件匣見

53 位中英文財經 KOL 的觀點,每天自動整理成一份日報。免費,隨時可退訂。

免費訂閱