Recently, Elon Musk opened X and typed nine words: "Path to Petawatts is Mass drivers on Moon."
No explanation. No context. No thread.
Just nine words that, if you know what they mean, describe the most audacious infrastructure project in human history and point directly to a publicly traded stock that most investors have never seriously considered.
Let's unpack what he actually said.
Here is the problem Musk is trying to solve, and it is a problem every investor in AI needs to understand.
AI runs on power.
Every ChatGPT query, every image generated, every algorithm trained burns electricity at a scale that is beginning to strain the entire U.S. power grid. The hyperscalers — Google, Microsoft, Amazon, Meta — are already fighting each other for access to nuclear plants, natural gas reserves, and transmission lines. Data center construction is being throttled not by capital or ambition but by a single constraint: there is not enough power on Earth to run the AI infrastructure they want to build.
Musk calls this the terawatt ceiling.
AI computing on Earth is capped — constrained by available land, power infrastructure, and cooling capacity. You cannot build your way out of it fast enough. The grid cannot keep up. And the projections for AI compute demand over the next decade make the current shortage look trivial.
So Musk looked up.
The Moon has no atmosphere. No weather. No competing cities fighting for power. It receives constant, unfiltered solar radiation across its surface. And its gravity is one-sixth of Earth's — meaning you can move enormous amounts of material on and off the surface for a fraction of the energy required here. The Moon, Musk has concluded, is the only place on Earth — or off it — where you can build AI infrastructure at truly civilizational scale.
The target: 500 to 1,000 terawatts of AI computing capacity. Petawatt scale. One thousand times what exists on Earth today.
A mass driver is exactly what it sounds like: an electromagnetic launch track built on the lunar surface that accelerates payloads to escape velocity using magnetic force — no rocket engines, no chemical propellant, no moving parts beyond the sled itself.
Think of it as a mile-long railgun built into the Moon's surface. A payload-carrying sled accelerates along the track until it reaches 2.38 kilometers per second — lunar escape velocity — at which point it releases and continues into space on its own momentum.
The Moon's low gravity and complete absence of atmosphere make this not just possible but extraordinarily efficient. Real-world mass driver designs run at 96–97% energetic efficiency.
Compare that to a rocket engine, which converts roughly 60% of its fuel into heat and exhaust.
"A mass driver is essentially a long electromagnetic track built on the lunar surface. Instead of burning rocket fuel to escape the Moon's gravity, a payload-carrying sled is accelerated using magnetic force until it reaches lunar escape velocity." — SpaceX Technical Briefing, March 2026
The economics are transformational.
A mass driver requires a massive upfront capital investment to build.
But once it is operational, the marginal cost of launching a payload into deep space drops effectively to zero. Every ton of material launched from the Moon — solar panels, computing hardware, construction material for orbital data centers — costs nothing beyond the electricity to run the magnets.
This isn't a new concept. The idea was first theorized in 1937 and was formally proposed by Professor Gerard O'Neill at Princeton in 1974 as the logical means for transporting lunar material to space habitats. What's new is that Elon Musk is the first person in history with the resources, the rockets, and the stated intention to actually build one.
The mass driver isn't a standalone project. It is the capstone of a larger architecture Musk has been assembling piece by piece for the past year.
Terafab — the chip complex announced at Giga Texas in March — is Phase One. Build the chips on Earth, prove the manufacturing process, and generate the revenue. But Terafab itself acknowledged a ceiling: even a fab complex approaching TSMC's scale cannot produce the compute infrastructure Musk's AI, robotics, and autonomous systems will eventually require. Earth simply doesn't have the power to run it all.
The Moon is Phase Two.
Manufacture AI computing infrastructure on the lunar surface using Terafab-derived chip technology, launch it into orbit via mass driver, and operate petawatt-scale AI from space — permanently free from Earth's energy constraints. SpaceX's Starship is the delivery vehicle that makes lunar construction possible, capable of landing over 100 metric tons per mission on the lunar surface.
The timeline is more concrete than most investors realize.
SpaceX is targeting June 2026 for its first full ship-to-ship orbital propellant transfer — a critical step for sustained lunar operations. NASA's Artemis program provides the broader framework. And Musk has already shifted SpaceX's internal priority from Mars to the Moon, describing a lunar city as humanity's immediate industrial priority over Mars colonization.
The Moon is no longer science fiction. It is a construction timeline.
You cannot buy SpaceX. Not yet.
SpaceX is the company at the center of everything Musk just described. The Starship that delivers cargo to the Moon. The Starlink constellation that provides communications infrastructure. The orbital refueling capability that makes sustained lunar operations possible. It is the most important space company ever built, and it is private.
But there is one publicly traded company that has spent the last five years building the infrastructure layer that makes commercial space viable.
That company is Rocket Lab.
Rocket Lab just reported the best financial year in its history.
Full-year 2025 revenue hit a record $602 million — up 38% year over year. Q4 revenue came in at $179.65 million, up 36%. The backlog grew 73% year over year to $1.85 billion. Q1 2026 guidance calls for $185 to $200 million in revenue — implying 57% year-over-year growth at the midpoint.
Those are not the numbers of a speculative space startup. They are the numbers of a company that has figured out how to generate real, recurring, contracted revenue from the space economy — and is accelerating.
Here is why Rocket Lab is interesting right now, specifically — not just as a long-term thesis, but as a stock with concrete near-term events.
The first is Neutron.
Rocket Lab's next-generation medium-lift reusable rocket filed its launch permit paperwork in April, targeting a debut window of July through December 2026. Neutron is the most anticipated milestone in commercial space this year outside of SpaceX itself.
When Neutron launches successfully, Rocket Lab graduates from a small-payload company into a medium-lift competitor capable of winning the kinds of constellation deployment contracts that currently go to SpaceX by default.
The permit filing alone moved the stock 9% in a single session.
The second is the contract momentum.
In the past 60 days alone, Rocket Lab won a $190 million defense deal, secured regulatory approval to acquire Mynaric for laser communications terminals, and passed $1 billion in total defense contracts for the year. The $816 million prime contract to build missile-warning satellites for the U.S. Space Development Agency — signed late last year — remains the foundation of a backlog that is growing faster than revenue itself.
The third is the CEO signal.
Peter Beck recently reduced his own salary — voluntarily — at a moment when the company's stock is near all-time highs.
Here's the full story, and it's more dramatic than a simple pay cut:
Effective March 30, 2026, Peter Beck voluntarily reduced his annual base salary to $1.00 — not a percentage cut, not a symbolic trim. One dollar. He also waived all bonus entitlements and cancelled 392,155 unvested RSUs, forfeiting shares he was already entitled to receive.
His prior total compensation was $20.11 million, with 97.1% coming from bonuses and equity.
He gave all of it up.
Why? The capital previously allocated to his compensation is being redirected directly to company priorities and strategic R&D initiatives. His stated reasoning: long-term shareholder value over short-term personal incentives.
The fourth near-term catalyst is the most important one — and it hasn't happened yet.
SpaceX confidentially filed for its IPO on April 1, 2026. The target valuation has already been revised upward above $2 trillion. Prediction markets on Polymarket currently assign a 93% probability that SpaceX is publicly traded before December 31, 2026.
When SpaceX lists, something structural happens to the entire space sector.
Institutional capital that has been sitting on the sidelines — unable to invest in space at scale because the only company that matters is private — suddenly has a $2 trillion anchor stock to build around. Index funds will be forced to buy it. ETFs will reprice. The entire category gets reframed from speculative to investable. And every company adjacent to SpaceX gets repriced in its shadow.
Rocket Lab already felt the first tremor.
When SpaceX IPO filing reports first surfaced, Rocket Lab's stock jumped immediately. That was the preview. The main event is a $2 trillion listing that introduces more retail and institutional capital to the commercial space sector in a single day than the entire category has seen in the last decade.
Rocket Lab is the only vertically integrated, publicly traded space company with real revenue, real contracts, and real infrastructure sitting in the direct path of that capital wave.