Tesla still holds unproven physical-AI optionality

Tesla still holds unproven physical-AI optionality. SpaceX already shows Starlink and AI revenue. If the companies combined at today’s market caps, Tesla holders would keep about 40%. A better deal for Tesla holders comes only after Robotaxi and Optimus create value first.

How $2 billion became a SpaceX stake

On January 16, 2026, Tesla agreed to invest about $2 billion in xAI Holdings Series E preferred stock. The filing described a minority equity investment. Tesla had already recognized about $430.1 million of Megapack revenue from xAI in 2025. A physical-AI company was selling power infrastructure to a digital-AI company and then trying to own a slice of it.

That structure did not last. According to Tesla’s filing, xAI Holdings was acquired by SpaceX on February 2, 2026. Tesla’s right to buy Series E preferred converted into a right to buy SpaceX Class A common stock. Tesla completed the $2 billion purchase on March 12. The first-quarter 10-Q recorded the stake as less than 1%.

The official language does not prove the investment was designed as a Tesla–SpaceX merger. The result is still clear. The companies now share commercial work, technical collaboration, and capital. The economic friction of a later combination is lower than before. That is not a completed merger.

Cars and rockets are the wrong frame

Tesla owns energy, FSD, Robotaxi, and Optimus. SpaceX and xAI own large-scale compute, Grok, Starlink, launch, and orbital infrastructure. One side has power and real-world data. The other has models, connectivity, and space infrastructure.

Regulation and governance set the speed

Robotaxi is not only a technology problem. City permits, insurance, remote intervention, and accident liability set the expansion rate. In July 2026 Reuters reported that Tesla’s tone on robotaxis had become more cautious and that the rollout was slower than hoped. Cumulative mileage is still early.

A merger scenario makes related-party terms more sensitive. Musk is an executive and a large shareholder across several companies. Reuters reported on August 13, 2026 that he disclosed a 48.4% SpaceX stake. Exchange ratios, asset transfers, and compensation stock decide existing holders’ share. A good business fit can still be diluted by governance.

A merger at today’s prices leaves Tesla holders with about 42%

In a stock-for-stock deal, existing holders’ stake is relative value. Tesla holders would own Tesla’s value divided by Tesla plus SpaceX. At current market caps, that is about 42%. SpaceX is larger. Tesla is already above $1 trillion, but the scale still tilts toward SpaceX.

Do not treat 2030 as a simple extension of today’s market caps. Split the businesses. The figures below are a base-case scenario, not price targets.

Under those assumptions, the combined firm is worth about $6.55 trillion. That is more than 10 percentage points better for Tesla holders than a deal today. Almost all of the difference comes from whether Robotaxi and Optimus are proven.

Treat 2035 as a range. The same work’s bear, base, and bull cases put the combined firm at $7T, $13T, and $20T. The $20T case is a super-bull in which almost every option works. Do not use it as a midpoint. If Tesla holders own 55% after a 2030 combination and Tesla’s share count is simplified to about 4 billion, today’s TSLA share back-solves to about $960 at $7T, $1,790 at $13T, and $2,750 at $20T. Later stock-based pay, Musk compensation, new issuance, and merger dilution are not included. A 15–20% cumulative dilution would cut those figures by the same amount.

Watch intervention rates and uptime, not demos

Tesla’s long-term value cannot be measured by car volumes alone. 2027 is a technology test: unpaid-to-paid unsupervised miles should rise quickly, Optimus should prove useful work inside factories at thousands to tens of thousands of units, and energy should reach about 60–80 GWh. 2028 is a business-model test: about 1 billion Robotaxi miles a year, the start of external Optimus sales around 50,000 units, and energy above 100 GWh. If Robotaxi reaches 10 billion miles, Optimus 500,000–1 million units, and energy 200 GWh by 2030, the company is in a platform phase.

A $2 trillion Optimus value is not a few robot sales. At 30x EBIT, $2 trillion implies about $67 billion of operating profit. At a 25% operating margin, that requires about $267 billion of revenue. At a $30,000 average selling price, that is about 8.9 million units a year. Even at $50,000 it is about 5.3 million.

A realistic $2 trillion model therefore needs recurring software, AI, and service revenue. That is not the industry’s current base case. Goldman Sachs Research’s 2024 base case put the 2035 global humanoid market at $38 billion and 1.4 million units a year. Tesla selling more than 5 million units a year by itself would require the market to be redefined. That is a bull case, not a consensus path.

The real addressable market is labor, not the robot catalog. At a $30,000 price, $6,000 of five-year depreciation, $3,000 of maintenance, $1,000 of power, and $5,000 of AI/software, annual cost is about $15,000. Against a $60,000 fully loaded human worker, the theoretical gap is $45,000 a year. Actual payback is narrower. It depends on whether the robot does the job, how often it stops, and how often it fails. A 99% task-success rate can still be too low in a factory. One thousand tasks a day means about ten failures. The useful Optimus metrics are task success, human intervention, uptime, and customer payback, not demo videos.

Tesla is a wait name; SpaceX is a watch name

As a business, Tesla has more physical-AI optionality and SpaceX has more proven growth. As a price, Tesla’s market cap is smaller than SpaceX today. As timing, 2027–2028 operating data matter more than merger headlines.

Tesla looks closer to a wait name. If the Robotaxi and Optimus logic remains intact, the optionality can look cheap on a relative-value basis. The convexity sitting on top of cars and energy still carries execution risk. If the logic breaks, cheap-looking relative value becomes a trap.

SpaceX looks closer to a watch name. Starlink and AI revenue are starting to show up in numbers. That strength is already in the market cap. This is the place to buy a more proven platform. The remaining asymmetry of unproven options still sits more with Tesla.

Use market-cap ratios, not share-price ratios

Share counts will change with issuance and compensation. The working ratio is Tesla market cap divided by SpaceX market cap. Below 0.70, Tesla looks very cheap on relative value. 0.70–0.90 is Tesla-favored. 0.90–1.05 is gap-narrowing. 1.05–1.20 is neutral. 1.20–1.40 favors SpaceX. Above 1.40, Tesla carries a larger premium. The August 14 close of 0.73 sits in the Tesla-favored zone.

Do not apply the rule mechanically to price alone. Put an execution gate in front of it. A low ratio is not a reason to add Tesla if Robotaxi and Optimus execution is breaking. Review this quarterly, not daily.

The best case is not a merger now

The best path is Tesla’s physical AI creating value first, then combining with SpaceX connectivity, AI, and space infrastructure at a fairer relative value. Only then do Tesla holders enter the combined platform on less unfavorable terms. Reuters’ report on SpaceX’s second-quarter debut results cited about $7.8 billion of revenue, nearly double the prior year, Starlink more than half of sales, and AI revenue up 250% year over year. Those details should be rechecked against company filings and the article text. Directionally, SpaceX is already moving from a launch company toward connectivity, AI, and space infrastructure.

The one-empire story is too strong

Narrative bias is the main risk. If the story that everything eventually combines feels too clean, unfinished value starts to look like current ownership. Long-time Tesla watchers also tend to anchor on old price targets. The importance of Musk’s ecosystem is not the same as a cheap current price.

The opposite error is also real. If SpaceX’s proven growth is all a reader sees, Tesla’s remaining optionality can be sold too early. A 0.73 ratio is not a reason to rebalance today. Moving size on one week of headlines is closer to FOMO.

Questions to keep asking

• Are Tesla and SpaceX being split into energy, physical AI, connectivity, compute, and space infrastructure rather than cars and rockets?
• Does the math include Tesla holders keeping only about 40% in a deal at today’s market caps?
• For Robotaxi, are cumulative miles, unsupervised paid share, intervention rates, and per-mile economics being tracked?
• For Optimus, are task success, uptime, external revenue, and payback being tracked instead of unit hype?
• For SpaceX, is Starlink and AI revenue starting to return the capital spend in cash?
• Are the market-cap ratio and execution data being reviewed together each quarter?
• If a merger headline appears, will size stay still until the exchange ratio and dilution are known?

A merger is not a confirmed event

It is only a strategic possibility. Robotaxi can work technically and still expand more slowly than hoped because of regulation, insurance, and unit economics. Optimus can remain a limited industrial robot rather than a general humanoid. In that case, the TAM and the valuation fall together.

SpaceX AI compute and orbital compute need huge capital spending. High growth is not the same as high free cash flow. Related-party terms and governance can distort the exchange ratio. This is not a bet that Musk builds one empire. It is an investment that has to keep testing each business’s economics and relative value.

Korean version: 한국어 버전 읽기

This article is a framework for separating company quality, price, and timing. It is not an instruction to buy or sell any security.

Telegram: https://t.me/signalandflow