Have you ever watched a stock sprint out of the gate, then trip over its own hype? That is pretty much the mood around SpaceX right now. The company went public in June after the largest initial offering many investors can remember, priced at $135 a share, raced toward $225.64, and then gave a lot of that move back. It closed the latest session at $145.47. Ugly? A little. Fatal? I do not think so.
Why The SpaceX Comeback Story Is Getting Louder Again
A fresh initiation from a major research desk just put a buy rating on the name and a $200 target. That is about 37% above the latest close. The Street, broadly speaking, is not hiding under the desk either. Most coverage still leans constructive, and the average target in the latest compilation implied even more room than that single $200 call. The interesting part is not the target itself. It is the reason behind it.
People still talk about rockets first. Fair enough. Rockets are photogenic. But the near-term money story, according to the new note, is much less cinematic. It is terrestrial AI compute leasing. Warehouses of power. Contracts with frontier labs. Capacity measured in gigawatts. That is the part that made me sit up.
The IPO Hangover Was Predictable
Record listings attract record attention. They also attract record expectations. SpaceX priced at $135. The tape then did what crowded tapes do. It overshot. Then reality showed up with a clipboard. Valuation debates started. Timing debates started. Some holders who only wanted the first-week pop headed for the exit.
I have found that mega-listings often need a second act. The first act is theater. The second act is whether the operating story can outrun the multiple. SpaceX is now in that second act. The pullback from $225.64 to the mid-$140s is not a verdict on the company. It is a reset of the crowd.
Does that mean the stock is cheap in some absolute sense? Not automatically. Growth names after a historic float rarely look cheap on last year’s numbers. They look cheap, or not, against the next three years of cash generation. That is the frame the new research is using.
AI Compute Leasing Is The Near-Term Engine
Here is the claim that matters. SpaceX’s terrestrial AI compute leasing, with customers said to include large platform and lab names, is expected to become the fastest-growing revenue line and, by early 2027, the majority of overall company revenue. That is a bold rotation. Launch used to be the identity. Compute is becoming the cash register.
The biggest near-term driver of revenue is leasing terrestrial AI compute capacity to frontier labs and other companies given the current AI supply and demand imbalance.
That imbalance is not a secret. Training runs keep getting larger. Inference is no longer a side dish. Power is scarce. Chips are scarce. Buildings that can actually host both are scarcer still. If you can stand up gigawatts and rent them to people who cannot wait, you do not need a poetry contest to explain the pricing power.
The note sketches a steep ramp. Capacity is seen rising from about 2.1 gigawatts in 2026 to 6 gigawatts by the end of 2027. One gigawatt is a billion watts. That is not a hobby rack in a closet. That is industrial scale. The same analysis estimates AI compute leasing revenue around $8.1 billion in the fourth quarter of 2026, exiting the year with roughly $41 billion of annual recurring revenue. Those are huge figures. They deserve skepticism. They also deserve attention.
Four customers are described as already leasing or starting to lease capacity in 2026. Two of the names in circulation are a leading lab and a major search-and-cloud parent. Even if you discount the exact dollars, the customer mix tells you who is willing to write checks when cluster time is tight.
Why Power Became A Space Story
It sounds odd at first. Why would a launch company own the AI landlord business? Then you look at the constraints. Frontier models need dense power, dense cooling, and sites that can expand without a ten-year permitting novel. Groups that already think in terms of energy, logistics, and large civil works have a head start. SpaceX has spent years solving ugly infrastructure problems. That muscle transfers.
In my experience, markets underestimate businesses that sit next to a bottleneck. Everyone can see the model labs. Fewer people price the warehouse that feeds them. If terrestrial leasing really becomes the majority of revenue by the first quarter of 2027, the stock stops being a pure “rocket multiple” and starts trading like a hybrid of infrastructure and growth software. That mix can support a higher floor after a messy IPO.
- Demand from frontier labs is still running ahead of available clusters.
- Capacity is measured in gigawatts, not in pretty slide decks.
- Recurring lease revenue can smooth the lumpiness of launch contracts.
- Customer concentration is a risk, but the counterparties are deep-pocketed.
Starship Still Matters, Just On A Longer Clock
None of this means the hardware story vanished. A Starship vehicle reached orbit on Monday, and that is not a footnote. Heavy lift changes the menu. More complex government work becomes realistic. Lunar architectures. Eventual Mars concepts. Cargo that used to be fantasy starts looking like a schedule item.
The research view is that Starship opens the door for Launch and Development revenue to overtake traditional Launch Services around the first quarter of 2030. That is a long fuse. Investors who only want next quarter’s beat will not care. Investors who think in platforms might.
I keep coming back to a simple split. Compute leasing is the 2026-2027 argument. Starship is the 2028-2030 argument. Confusing those two clocks is how people get whipsawed. The stock can work on the first clock even if the second clock slips. That is unusual for a space name, and it is probably why coverage turned constructive again after the fade.
How The Street Is Lined Up
Consensus is not unanimous. It never is after a debut this large. Still, the latest tally showed 31 of 40 ratings in the buy or strong-buy camp. The average target implied about 57% upside from recent levels. That is not a quiet room. It is a room that has already debated the IPO pop and decided the operating ramp still looks intact.
A $200 single-name target sitting below the Street average is almost comforting. It suggests the new work is not the most aggressive sheet on the table. It is a base-case buy, not a moon-or-bust poster. Perhaps the most interesting aspect is how little of the thesis now depends on a perfect launch cadence next month.
| Piece of the story | Timing | Why it moves the stock |
| AI compute leasing | 2026 into early 2027 | Fastest growth line, possible majority of revenue |
| Gigawatt capacity | 2.1 GW in 2026, 6 GW by end-2027 | Physical proof that leases can scale |
| Starship missions | Late decade | Higher-value government and exploration work |
| IPO afterglow | Already fading | Clears space for fundamentals to matter again |
The Valuation Debate Will Not Go Away
Let’s be adults about this. A company that just completed a historic listing will live under a microscope. Every slip in launch timing will be a headline. Every delay in a data hall will be a headline. Every whisper about customer concentration will be a headline. That is the tax you pay for being famous.
The bull case needs those $8.1 billion quarterly leasing figures to look directionally right, even if they are not exact. It also needs the 6 gigawatt path to remain credible. Miss both, and $200 becomes a souvenir. Hit both, and the mid-$140s look like the hangover low rather than the start of a long grind.
Bears will say the market is stuffing two stories into one ticker and overpaying for both. Fair hit. Bulls will say the ticker is one of the few ways to own scarce power plus scarce lift in the same vehicle. Also a fair hit. I lean toward the second camp, with a seatbelt on.
What “Scooping Up Shares” Actually Means
Research notes love the phrase. Scoop. Accumulate. Buy the dip. Cute. In real life it means deciding whether $145.47 is a level where the next two years of compute contracts can justify the wait. It does not mean mortgaging the house because a rocket looked pretty on Monday.
Position size still matters more than poetry. A name this volatile can give you a 15% gift and take it back before lunch. If the AI leasing line is the true driver, the tape will start reacting to capacity updates and contract language, not just launch montages. That would be a healthier market, even if it is a noisier one.
- Separate the 2026 compute ramp from the 2030 exploration story.
- Watch gigawatt additions the way you once watched launch success rates.
- Treat customer concentration as a live risk, not a trivia item.
- Assume IPO-era holders will keep selling strength for a while.
- Only add if the operating evidence, not the logo, is improving.
Risks That Can Still Wreck The Rebound
Power projects slip. Permitting slips. Cooling systems fail in public. Labs can build their own halls and leave landlords with empty megawatts. Launch anomalies can still tax management attention. A broader risk-off tape can punish high-duration growth even when the local story is fine. None of that disappeared because one desk initiated coverage.
There is also narrative risk. If investors decide SpaceX is “just another AI landlord,” the scarcity premium tied to orbital capability could shrink. If they decide it is “just a rocket company with a side hustle,” the multiple attached to recurring compute revenue could shrink. The stock needs the market to accept the hybrid. Hybrids are awkward until they work.
I would rather see messy-but-real capacity additions than a perfect story with no warehouses. Markets eventually pay for electrons on the floor, not adjectives on a slide.
How This Fits A Growth Portfolio
For investors who already own platform and semiconductor names, SpaceX is not a clone. It is a bet that physical scarcity sits underneath the model boom. That can diversify a book that is otherwise all chips and software. It can also double the same cycle if AI capex slows. Correlation is not zero. Do not pretend it is.
The name also behaves like a celebrity stock. Newsflow is constant. Social chatter is constant. That can be useful if you like liquidity. It can be poison if you need silence. I’ve found that the investors who do best in names like this decide their time horizon on day one and then refuse to renegotiate with every clip.
Simple working split: Near term: terrestrial AI leases and GW buildout Medium term: mix shift toward recurring compute Long term: Starship-enabled government and exploration work
Reading The Next Few Quarters Without Getting Dizzy
Ignore the daily argument about whether $145 is “the” bottom. Bottoms are labeled after the fact. Watch whether leased megawatts show up on schedule. Watch whether the fourth-quarter 2026 leasing run-rate conversation stays in the same zip code as $8.1 billion. Watch whether Starship flights stay in the “useful test” bucket rather than the “expensive setback” bucket.
If those three stay constructive, the rebound thesis has oxygen. If two of the three wobble, the $200 target starts to look decorative. That is a cleaner checklist than most space stocks ever give you.
Growth in artificial intelligence and space can travel together, but they do not travel at the same speed. Price the faster one first.
A Personal Read On The Setup
I like businesses that surprise people by becoming boringly useful. Launch will never be boring. Compute halls can be. Renting scarce power to labs that cannot wait is a dull, beautiful business if the contracts hold. That dull layer is what makes the post-IPO fade look, to me, more like digestion than disaster.
Would I call it risk-free? Of course not. No honest person would. Would I call the mid-$140s automatically attractive after a trip to $225? Only if you believe the gigawatt plan and the 2027 mix shift. I do think that plan is the right thing to underwrite. I also think a lot of owners still have not emotionally left the IPO week behind.
So here is the unglamorous conclusion. The comeback is not a fireworks show. It is a landlord story wearing a flight suit. If that sentence bothers you, this ticker will keep bothering you. If it makes sense, the recent pullback is the first moment the stock has looked like a company again instead of a spectacle.
What To Remember When The Headlines Get Loud
Monday’s orbital flight will own the clips. The research initiation will own the morning notes. Neither one is the whole company. The whole company is a race to stand up power, keep labs paying, and still advance a vehicle that can do work no other stack can do. That is a lot of moving parts. It is also a clearer map than the one investors had on listing day.
SpaceX does not need another night at $225 to prove the listing was real. It needs a couple of quiet quarters where capacity and contracts look adult. If those arrive, the rebound will not feel like a rumor. It will feel like math. And math, after a circus opening, is usually the more durable kind of comeback.
Keep the time horizons straight. Keep the risks on the table. And if you do add, add because the compute story is becoming measurable, not because the logo still thrills a crowded room.
