Elon Musk has spent years building companies around big, long-term bets. But two of his biggest businesses are now moving into an increasingly similar territory: artificial intelligence.
Tesla is betting on AI through autonomous driving, humanoid robots, custom chips and robotaxis.
SpaceX is taking a different route, building massive computing infrastructure, expanding Starlink and integrating AI businesses such as xAI and Grok into its ecosystem.
Both are chasing the same broad opportunity, but they are at very different stages.
Right now, SpaceX appears to have the more developed AI infrastructure story. Tesla, however, may have the bigger physical AI opportunity if Optimus and autonomy eventually scale.
That difference is becoming increasingly important for investors.
Tesla’s AI Bet Is Getting More Expensive
Tesla’s latest numbers show just how much money is being poured into the company’s future.
The company reported $28.24 billion in Q2 revenue, while its operating margin fell to just 1.4%. At the same time, capital expenditure jumped more than 140% year over year to $5.79 billion, pushing free cash flow into negative territory.
Management expects heavy spending to continue for the next two to three years.
The money is going toward several ambitious projects:
- Robotaxi
- Optimus humanoid robots
- AI computing infrastructure
- AI5 and future AI6 chips
- TerraFab, Tesla’s planned semiconductor facility
- Cybercab
This is the key issue with Tesla today.
The company isn’t simply investing in one new product. It is trying to build an entire AI ecosystem at the same time.
And investors are being asked to believe that these investments will eventually generate returns large enough to justify today’s valuation.
The Robotaxi Story Has Progress, But the Gap Is Still Large
Tesla’s autonomous driving ambitions remain at the center of its bull case.
Robotaxi operations have expanded into seven U.S. markets, while Tesla says its fleet has accumulated roughly 380,000 miles of unsupervised operation.
FSD subscriptions have also continued to grow, reaching roughly 1.48 million active subscriptions, up 56% year over year.
Those are meaningful numbers.
But the competition is moving too.
Waymo is reportedly completing around 500,000 fully autonomous rides per week, highlighting how much more operating experience it currently has in commercial autonomous transportation.
That doesn’t mean Tesla cannot catch up.
Tesla’s advantage is its enormous vehicle fleet and the amount of driving data generated through its vehicles. If its approach eventually works at scale, the economics could be very different from a smaller, highly mapped autonomous fleet.
But that “if” matters.
Tesla’s valuation is increasingly dependent on execution rather than simply potential.
Then There Is Optimus
Optimus could be Tesla’s biggest long-term opportunity.
Musk has suggested that Tesla could eventually produce millions of humanoid robots each year, with an aspirational target of around 10 million units annually.
That would completely change the scale of Tesla’s business.
But the company isn’t there yet.
The market is still waiting for evidence that Optimus can move from demonstrations and early production plans into a genuinely scalable commercial product.
The source material points to prediction-market odds of only around 3% for a 2026 Optimus release, underscoring how skeptical the market remains about the timing.
This is an important distinction:
The opportunity can be enormous without the timeline being realistic.
That has been a recurring theme around several Tesla projects.
SpaceX Is Playing the AI Game Differently
SpaceX’s AI strategy is much more focused on infrastructure.
After its integration with xAI, the company has been building out massive data-center capacity to support AI workloads.
And unlike Tesla, SpaceX is already generating revenue from this infrastructure.
That changes the conversation.
SpaceX reportedly had 1.4 gigawatts of installed computing capacity by the end of June, up from 1 gigawatt in March and 0.4 gigawatts a year earlier.
Musk expects that figure to exceed 2 gigawatts by the end of 2026, with the possibility of reaching something closer to 10 gigawatts by the end of 2027.
The spending required is enormous.
SpaceX spent $15.8 billion on AI computing infrastructure in Q2 alone, representing the overwhelming majority of its quarterly capital expenditure.
The company expects another roughly $37 billion of capital expenditure over the next two quarters, according to the source material.
That is a staggering amount of money.
But there is a reason SpaceX is willing to spend it.
SpaceX Already Has Customers Waiting
This is where the SpaceX story gets particularly interesting.
The company isn’t simply building AI infrastructure and hoping demand eventually arrives.
It has reportedly signed major agreements tied to Nvidia-powered computing capacity.
One agreement with Anthropic covers approximately 325,000 Nvidia GPUs, while an agreement with Google covers another 110,000 GPUs.
There are also additional cloud service contracts being signed.
Oppenheimer has argued that newer Nvidia Rubin hardware could potentially have a payback period of around one year under favorable cloud economics.
If that holds, SpaceX’s aggressive spending starts to look less like pure speculation and more like an infrastructure expansion backed by real demand.
But there is still a major risk.
The Nvidia Dependency
Musk has made an unusually strong commitment to Nvidia.
He said SpaceX had decided to build its AI infrastructure exclusively on Nvidia hardware, specifically citing the Vera Rubin architecture.
For Nvidia, that is obviously a major vote of confidence.
For SpaceX, however, it creates concentration risk.
Nvidia is already dealing with supply constraints, and SpaceX is trying to build computing capacity at a breathtaking pace.
That creates two obvious problems:
- Price: SpaceX loses some negotiating leverage by relying heavily on one supplier.
- Availability: Even if SpaceX has the money, it still needs Nvidia to deliver the hardware on time.
SpaceX’s planned Terafab semiconductor facility could eventually provide a hedge against some of these constraints, but the project is still developing and does not yet have definitive agreements in place.
So the AI opportunity is real, but the execution challenge is just as real.
Musk’s $3.5 Trillion Forecast Raises Eyebrows
Musk recently suggested that SpaceX could generate approximately $3.5 trillion in annual revenue by 2033.
That is an extraordinary figure.
SpaceX generated around $12.5 billion in revenue during the first six months of 2026, according to the source material.
To reach $3.5 trillion by 2033, the company would need to maintain an extraordinary growth rate for years.
For perspective, the projection would be several times larger than the combined annual revenue of Amazon and Walmart based on the figures cited in the source.
That doesn’t automatically mean Musk’s forecast is impossible.
But it does mean investors should treat it as an extremely aggressive long-term vision rather than conventional company guidance.
Musk Has Been Directionally Right, But Timing Has Been the Problem
This is probably the most important lesson from Tesla’s history.
Musk has repeatedly identified technologies that eventually became important.
The problem has often been the timeline.
Tesla’s autonomous driving promises, robotaxi ambitions, Cybertruck production targets, Tesla Semi rollout and Optimus expectations have all taken longer than originally suggested.
That doesn’t invalidate the underlying technology.
It does matter for investors.
A great technology arriving five years later than expected can produce a very different investment outcome from that same technology arriving next year.
And that is exactly why timing matters so much for Tesla and SpaceX today.
The $30 Trillion AI Question
Musk has also suggested that AI could eventually add $20 trillion to $30 trillion annually to the global economy.
The broader direction of that argument is difficult to dismiss.
AI adoption is accelerating, companies are spending heavily on computing infrastructure and businesses are increasingly integrating AI into everyday workflows.
Nvidia’s numbers illustrate the scale of the infrastructure boom.
The company reported roughly $96 billion in quarterly revenue, with data-center revenue reaching around $89 billion.
Alphabet is also spending aggressively, with quarterly capital expenditure approaching $45 billion while Google Cloud continues to grow rapidly.
The demand is clearly there.
The bigger question is how quickly that investment translates into economy-wide productivity gains.
That process may take considerably longer than Musk’s most aggressive forecasts suggest.
Technology can spread quickly.
Organizations usually don’t.
Tesla Has the Bigger Physical AI Opportunity
This is where the Tesla bull case becomes interesting again.
SpaceX may currently have the stronger AI infrastructure business, but Tesla has something SpaceX does not have to the same extent:
Millions of potential physical AI endpoints.
Vehicles.
Factories.
Robots.
Energy systems.
Robotaxis.
If Tesla can successfully combine its software, chips, autonomous driving systems and humanoid robots, it could create a very different type of AI business.
Imagine a world where the company isn’t primarily selling cars.
Instead, Tesla could be selling access to an enormous network of autonomous machines.
That is the long-term vision investors are paying for.
The problem is that much of that value is still ahead of the company.
The Valuation Is the Real Debate
Tesla’s valuation makes this especially important.
The stock has been trading at a trailing P/E of roughly 370 times earnings, while operating margins are only around 1.4% and free cash flow has turned negative.
At that valuation, investors aren’t paying for today’s Tesla alone.
They are paying for the Tesla that could exist if robotaxis, Optimus, AI chips and energy infrastructure all scale successfully.
That creates a very high bar.
Tesla can continue making progress and still disappoint investors if that progress is slower than expected.
This is why the next few quarters matter.
What Investors Should Watch
Rather than focusing on every new Musk prediction, investors may be better served by watching a handful of measurable milestones.
For Tesla:
- Robotaxi expansion and autonomous miles
- FSD adoption and subscription growth
- Optimus production progress
- TerraFab construction and AI5 chip production
- Automotive margins
- Free cash flow
- Capital expenditure
For SpaceX:
- Growth in installed AI computing capacity
- Nvidia GPU deliveries
- New cloud contracts
- AI infrastructure revenue
- Starlink growth
- Terafab progress
- Capital expenditure and cash requirements
These numbers will tell investors much more than another ambitious forecast.
So Who Is Better Positioned to Win the AI Race?
For the near-term AI infrastructure opportunity, SpaceX has the stronger position.
It already has computing capacity, customers and revenue tied to that infrastructure. Its integration with xAI also gives the company a way to participate in AI software and compute rather than simply owning hardware.
Tesla’s opportunity is different.
It is betting that AI will eventually move beyond data centers and into the physical world through autonomous vehicles, robotaxis and humanoid robots.
If that happens at scale, Tesla’s addressable market could be enormous.
But Tesla has more execution risk to overcome.
SpaceX is building the infrastructure that AI needs today. Tesla is betting on what AI-enabled machines could become tomorrow.
That distinction may be the most important thing for investors to keep in mind.
The Bigger Musk Ecosystem
There is also an increasingly interesting connection between the two companies.
Tesla already recorded a roughly $1 billion mark-to-market gain on its SpaceX holdings, meaning Tesla shareholders are indirectly exposed to part of the SpaceX story.
There has also been market speculation around deeper ties between the companies, although the strongest prediction-market outcome for a formal merger-related announcement by the end of 2027 was priced at only around 46.5% in the source material.
For now, investors should focus less on merger speculation and more on execution.
Because both companies are spending enormous amounts of money on the same broad AI revolution.
The difference is that SpaceX is already selling the infrastructure, while Tesla is still trying to prove that its most ambitious AI products can scale.
And that could make 2027 a very important year for both.
The AI prize may be huge. The real question is who can turn the vision into cash flow first.