The Future Caught Up: A 2026 Scorecard for an October 2024 Demo
In October 2024, I stood in front of the Navy League of Ft. Lauderdale and ran a live AI simulation that was supposed to take 20 minutes. It took 90.
The demo was simple in concept, aggressive in implication: nuclear-powered vessels — already floating fortresses of energy — serving a dual role as autonomous, off-grid data centers. Surplus reactor power running AI systems for predictive maintenance, sensor fusion, tactical response, and fleet coordination. No cloud. No terrestrial data center. No network dependency. The ship thinks for itself.
I got a plaque and a challenge coin. The audience asked sharp questions. Someone made a Skynet joke. I went home.
In April 2025, I got a call from Frank — a Navy insider who doesn't pick up the phone to make small talk. He said the concepts from the demo were now being discussed inside the Navy, including at the War College. I wrote a piece called "Ahead of the Fleet" about that call. I said it wasn't about being right. It was about being ready.
In August 2026, I got another call. A Navy client. He'd been in the room for conversations that confirmed what the simulation showed two years ago — and he wanted to talk about it. Not theoretically. Operationally. As in: this is what we're doing now. So I had AI run a verification pass against the public record. Here's the scorecard.
PREDICTIVE MAINTENANCE: ACTIVE NAVY ENGINEERING PROGRAM
In the demo, I showed onboard AI continuously examining propulsion and machinery data, identifying developing problems before they disabled the vessel. NAVSEA is now publicly describing machine-learning systems for predictive machinery health — designed to tell sailors which components genuinely need attention instead of relying on fixed maintenance schedules. Their framing is almost exactly mine: reduce unnecessary labor, catch faults earlier, keep vessels operational.
That part didn't just land. It became an active program.
AUTONOMOUS MARITIME SCOUTS: IN SUSTAINED FLEET OPERATIONS
The underwater drones and autonomous scouts from the demo are no longer demonstrations. During 2025, the Fourth Fleet moved unmanned surface vessels into sustained operational use under Operation Southern Spear — integrating as many as 20 Saildrone vessels into fleet activity. Fifth and Sixth Fleet efforts expanded too.
By 2026, the Chief of Naval Operations was publicly describing swarms of robotic and autonomous systems as the "spine of a future hedge force." Not a novelty attached to the fleet. A structural layer of it. The Navy even created a dedicated Portfolio Acquisition Executive for Robotics and Autonomous Systems. Autonomy now has its own institutional acquisition structure.
DECENTRALIZED INTELLIGENCE: YES, UNDER DIFFERENT LANGUAGE
The key idea was that a vessel couldn't assume permanent access to a cloud or terrestrial data center. It needed enough local intelligence to continue interpreting sensors, diagnosing equipment, and supporting tactical decisions while disconnected.
That's where naval computing is heading: resilient edge processing, local decision support, distributed sensors, reduced dependence on vulnerable communications links. They may not publicly call it an "autonomous off-grid data center." But architecturally, that's increasingly what a modern warship is becoming — a mobile power plant, sensor platform, communications node, and local compute environment that must continue functioning when the network is degraded or gone.
AI INSIDE THE SHIPYARDS: THE ONE THAT'S STARTLING
In the April 2025 follow-up, I wrote that AI could improve shipyard scheduling, inspection, construction, and supply-chain coordination.
In December 2025, the Navy announced a $448 million Shipbuilding Operating System — "Ship OS" — with Palantir. It's being introduced across four public shipyards and several private yards to bring AI directly into ship construction and repair. The objectives: optimize shipyard work in real time, connect suppliers through intelligent logistics, give program managers better visibility into cost, schedule, and risk.
The Navy reported that one submarine-planning process dropped from 160 manual hours to under 10 minutes. A material-review process fell from weeks to less than an hour.
That is almost a direct operationalization of the shipyard section I wrote. Not vaguely right about "AI helping industry." Right about the precise bottleneck where the Navy is now spending hundreds of millions of dollars.
THE NUCLEAR-POWERED AI PLATFORM: DIRECTIONALLY RIGHT, NOT PUBLICLY COMPLETED
I don't see public evidence that the Navy has formally designated carriers or nuclear submarines as floating AI data centers powered by surplus reactor capacity in the exact form I demonstrated. That specific packaging remains ahead of the public record.
But the underlying engineering thesis has held. Nuclear vessels possess extraordinary persistent power generation. Modern combat systems require increasing onboard compute. Ships must process growing sensor volumes locally. AI inference and autonomous decision support must survive network disruption. The Navy continues investing heavily in operational nuclear-power systems while its AI, autonomy, sensor, and edge-compute requirements advance alongside them.
The exact headline hasn't appeared. The architecture beneath it has.
THE 2026 SCORECARD
October 2024 concept → August 2026 status: Shipboard predictive maintenance → Active NAVSEA AI/ML machinery-health program.
Onboard decision support → Expanding through edge AI and fleet experimentation.
Autonomous maritime scouts → Entered sustained fleet operations under Operation Southern Spear, 20 Saildrones.
Distributed autonomous fleet → Becoming formal force architecture. CNO calling it the "spine of the hedge force."
AI-assisted shipyard modernization → $448M Ship OS initiative with Palantir across four public shipyards.
Intelligent maritime supply chain → Included in national shipbuilding policy.
Nuclear vessel as autonomous data center → Not publicly formalized. Architecture converging.
About 80 percent of the demonstration has now appeared publicly in recognizable operational form. The remaining 20 percent is the unifying layer — treating the nuclear vessel explicitly as a sovereign, mobile AI-compute platform rather than putting individual AI applications aboard it.
And that may be why the client lost his mind. He wasn't looking at someone who'd written clever articles after reading Navy press releases. He was looking at a civilian who stood in front of the Navy League in October 2024, assembled the power, compute, autonomy, maintenance, and shipbuilding pieces into one coherent architecture — and then watched the institution begin building those pieces over the following two years.
I said in April 2025 it wasn't about being right. It was about being ready. That's still true. But I'll add something now: it's also about the gap between foresight and institutional adoption. That gap is closing. Not because institutions got faster — because the technology got more obvious. The architecture I described wasn't speculative. It was inevitable. It just took two years for the procurement cycles to catch up to the physics.
The future did, in fact, catch up. The question now is what's next — and who's listening.
Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital.
Comments