Elon Needs Help
Elon Musk is going to need a lot of help.
Just recently on SpaceX's first-ever public earnings call, Musk laid out plans to deploy 10 gigawatts of compute by the end of 2027 — roughly the same energy usage as powering 8 million U.S. homes.
To build this capacity, Musk will need to acquire more than 3 million of the most in-demand processors on earth in just 12 months.
That's a scale no single company can build alone. Not even the guy behind reusable rockets and mainstream EVs.
Back in June, I came up with something I called “the Elon 7.”
These are the seven companies best positioned to profit from bottlenecks Elon can't solve himself.
But there were four more names that deserved to be on that list. So the Elon 7 became the Elon 11.
The reasons for that expansion split into two separate problems: building enough compute, and getting it to people.
Here’s what I mean by that…
Elon Can’t Build It All Alone
As much as Elon wants to build it all himself, Tesla and his other companies can't do it alone. Nowhere is that clearer than with chips.
Tesla designs its own AI chips, called AI5 and AI6, to run its self-driving software and its Optimus robots. But Tesla has no factory to build them.
That job belongs to Taiwan Semiconductor Manufacturing Company (TSM), the first new member of the Elon 11.
Tesla is trying to build its own chip manufacturing facility, an in-house project called Terafab. But catching TSM will take another year or two at a minimum. Until then, Tesla needs TSM to turn its designs into real silicon.
Producing the chips is only half the problem. Someone still has to turn those chips into a working data center fast — and at a scale most companies have never attempted.
Two names keep showing up when Musk's companies need that done: Super Micro Computer (SMCI) and Dell Technologies (DELL).
Together, SMCI and Dell built the racks and cooling for one of the largest AI computing clusters ever assembled, a roughly 100,000-chip buildout for xAI.
Elon may want to build it all in-house — Terafab is proof he's trying — but when the timeline is measured in months, he still needs outside crews who already know how to move that fast.
The fourth name solves a different kind of problem, one that has nothing to do with selling into Musk's spending.
Even the biggest datacenter buildout is worthless if it never reaches an actual person. AST SpaceMobile (ASTS) is built to close that gap.
AST’s satellites connect directly to a standard smartphone (the same one already in your pocket) with no dish or special equipment required. The company has demonstrated speeds up to 1 gigabit per second straight to a regular handset.
ASTS is the piece that turns Musk’s big datacenter buildout into something you can actually use.
If you’re wondering why Musk would need ASTS when he already has Starlink, the answer comes down to what happens on the receiving end.
Starlink was built around dedicated terminals. ASTS was built to connect directly to the cellular hardware already inside phones — and potentially vehicles, robots and other autonomous machines.
That matters as Musk pushes further into self-driving cars and Optimus. Those machines need to stay connected wherever they go, including beyond the reach of terrestrial networks.
In other words, ASTS wouldn’t replace Starlink. It could fill a gap Starlink was never originally designed to solve.
Where These Four Fit In
The August SpaceX earnings call is the whole story in miniature.
Musk can raise the money, design the chips, and launch the satellites. But he cannot fabricate every wafer, build every rack, or reach every phone by himself.
That gap breaks down into exactly two problems: building enough compute (the production side) and getting it to people (the delivery side).
TSM, SMCI, and Dell solve production. ASTS solves delivery.
Each one earns its place the same way the original Elon 7 did, by sitting at a chokepoint Musk's companies cannot build around.
None of that undoes the original seven. It just means the list keeps growing as the bottlenecks do.
