How to Buy
          Active Investing

          Why Tesla And SpaceX Are Converging, And Why A Combination Could Make Both Stronger

          11 November 2026

          14 Min Read

          Key Takeaways

          SpaceX’s AI compute business is scaling at a pace no peer matches. New capacity pays back in under a year, and management is targeting roughly 13–23 incremental GW by the end of 2027, versus the 3.9 GW AWS added in all of 2025.

          Tesla’s autonomy flywheel is accelerating. FSD subscriptions grew 67% year over year, and FSD reached 75% of new North American deliveries, above comparable software and in-car subscription benchmarks despite a shorter trial and a higher price.

          Both companies are investing from strength. Tesla cut 2026 capex to about $20 billion, funded from cash flow, while SpaceX guided Q1 2027 revenue of $33–37 billion against roughly $22 billion expected.

          A combination is under formal review. Independent special committees at both companies are considering it. No terms are agreed, and Tesla would need approval from a majority of all shares.

          We believe it would be a substantial win for both shareholder bases. The businesses are already entangled, share a CEO and his managerial attention, and Tesla’s robotaxi cash flow could help fund SpaceX’s compute build-out, which could cost more than $400 billion for 10 GW.

          The math works across a range of terms. At every premium we examined, from 0% to 50%, both shareholder bases come out ahead. An equal split of the added upside implies SpaceX holders paying roughly 24% over Tesla’s price.

          We expect the two companies to come to terms, while recognizing that the shareholder vote is a high bar and nothing is decided.

          SpaceX reported on October 20 and Tesla on October 21, and read together the two quarters point the same way. SpaceX is scaling AI compute at a pace few can match, while Tesla is building the autonomy, robotics and energy businesses that will need that compute, and the two companies have now formed special committees to consider combining. This note brings our views on both quarters, our latest Tesla modeling and the case for a combination together in one document.

           

          SpaceX Q3: AI Compute Is Now The Engine

          Revenue was $14.6 billion, up 161%, with adjusted EBITDA of $8.8 billion and net income of $4.4 billion. The Super Intelligence (SI) segment generated $7.6 billion, including $0.6 billion from the Cursor acquisition, and is now larger than Connectivity at $5.3 billion. Excluding Cursor, segment revenue still grew 192% sequentially, and segment adjusted EBITDA of about $5.0 billion implies a margin of roughly two-thirds. One data point came in light: Starlink consumer subscribers of 13.6 million were below the roughly 14.2 million consensus, although ARPU rose to $69.

          • The unit economics are exceptional. New terrestrial compute deployments pay back in under a year at current pricing. Market pricing is running above last quarter’s $30–$50 billion per gigawatt, against roughly $40 billion per gigawatt to deploy NVIDIA Vera Rubin systems before surrounding data center costs. Musk said a six-month payback next year is “not impossible.”
          • SpaceX is now operating a large AI cloud. Management aims to split compute roughly 50/50 between its own products and outside customers, on short-term contracts that are renewing at higher prices. Musk said hyperscalers and others are asking to rent far more than SpaceX can offer, and that Grok Bot, with 800,000 paid weekly users (more than double the prior month), may end up using most of its own compute.
          • The build-out plan keeps growing. A quarter ago, Musk said cumulative compute could be closer to 10 GW than 5 GW by the end of 2027. This quarter he said he is confident of adding 10 GW and reaching roughly 13–23 GW in total. Nameplate capacity was 1.7 GW at quarter-end, with 2.5–3 GW expected by year-end. For perspective, AWS added 3.9 GW in 2025, and the largest neocloud plans to cross 8 GW by 2030.
          • The financial outlook is being pulled forward. Short-term contracts renewing at higher prices have accelerated the forecast. Q4 capex is guided to $35–40 billion, Q1 2027 revenue guidance of $33–37 billion compares with the roughly $22 billion analysts had modeled, and after the call Musk said $1 trillion of revenue in 2028 is “no longer impossible,” earlier than the 2030 he cited a quarter ago and 2031 at the time of the IPO.
          • Starship lowers the cost curve. Flight 15, targeted for November, will try to catch both booster and ship, a first for an orbital launch vehicle, and Musk expects weekly flights in the first half of 2027 and two a week in the second half. According to our research, Falcon 9 has cut launch costs from $15,600 per kilogram in 2008 to just under $1,000, and at roughly $300 per kilogram we think orbital data centers can compete with terrestrial ones. SpaceX’s Starmind satellites are slated to begin flying at the end of 2027.
          • Market worries about AI cloud look overblown. Shares of neocloud providers have sold off on questions about long-term demand, borrowing costs and customer credit quality. We view these concerns as largely overblown and reminiscent of the bubble fears of late 2025, which faded as Anthropic’s rapid revenue ramp reignited optimism.

           

          Tesla Q3: Autonomy Accelerates Beneath A Steady Headline

          Revenue was $28.0 billion, flat year over year, and non-GAAP EPS of $0.48 was ahead of the roughly $0.44–0.45 expected. Gross margin excluding credits improved to 18.1%, helped by higher deliveries in Asia-Pacific. GAAP operating income fell 56% to $718 million as Tesla stepped up spending on AI compute, autonomy and Optimus. The merger was not discussed on the call.

          • FSD demand is accelerating. Active subscriptions reached 1.74 million, up 67% from 56% growth last quarter and ahead of the roughly 1.57 million consensus. The share of new North American deliveries with an FSD subscription rose from about 55% to more than 75% in a single quarter.
          • Robotaxi is scaling. Cumulative driverless miles passed 3.3 million, and the Cybercab fleet in Austin has grown tenfold in the month since launch. Customer-owned Model Ys are expected to join the fleet from around Q2 2027. Management called the ramp technology-driven, and expects it to be noticeable in revenue in Q2 2027 and a big deal by the second half.
          • Ambitions extend beyond robotaxis. The CFO said Tesla wants to “own… transportation of any kind on earth.” As robotaxi prices fall below $1 per mile, we expect Tesla to work with The Boring Company to move some trips underground, and it may pursue electric aviation, while Starship could cut long-haul international flights to under 30 minutes.
          • Autos and energy have room to grow. Management is aiming for roughly 5 million vehicles a year around 2030. Storage gross margin improved to 33.1% from 20.4%, Megapacks are close to sold out for 2027, and demand from data centers is exceptional.
          • Capital is deployed efficiently. 2026 capex guidance was cut to about $20 billion from more than $25 billion, funded from cash flow, with $30 billion of debt capacity available and no draws planned this year.
          • Optimus is moving into production. Most of a million-unit-a-year line (about 20,000 units a week) has landed in Fremont, with at least a year to reach that rate. Integrating Tesla’s AI5 chip, targeted for April, brings an order-of-magnitude increase in operations per second.

          The FSD conversion rate is striking next to other subscription businesses: about 18% of software trials convert to paid without a card, and 49% with one, while SiriusXM converted about 39% of new-car owners and GM Super Cruise about 40%. Both offered longer trials (three to twelve months, and three years) at lower prices ($11–21 and about $40 a month, against $99 for FSD with a one-month trial). Tesla outpacing them on shorter terms and a higher price suggests customers see FSD as more essential than optional. Conservatively, our modeling excludes personal FSD subscription revenue, because we believe autonomous ride-hail will dwarf it.

           

          How ARK’s Tesla Modeling Lines Up With The Results

          ARK’s latest Tesla modeling, completed ahead of these results, anticipated two of the themes that surfaced in the earnings:

          • Robotaxi is the core of the value. We believe robotaxis alone will create substantial value for Tesla shareholders over the next five years, and could account for roughly 80% of Tesla’s enterprise value in 2031. Tesla launched in June 2025, the most probable year in our prior work, and we estimate its cost per mile could run roughly 40% below Waymo’s next-generation vehicle at equal utilization at scale.
          • Optimus carries more weight. We now assume Optimus scales to roughly 1 million units and about 14% of enterprise value in 2031. A million-unit line taking shape in Fremont fits that direction, and our assumptions are more conservative than management’s unit guidance, Elon has previously said 100M – 1B in the same year.

           

          The Case For Convergence

          On the SpaceX call, Musk said a combination would be a good idea, with committees of independent directors and himself recused. He also cautioned that it should not be assumed to be a certainty, that no terms have been agreed, and that Tesla needs approval from a majority of all shareholders, which he called a high hurdle. Within that framing, we see six reasons the logic holds:

          • The businesses are already entangled. Tesla and SpaceX are collaborating on Terafab and the next-generation Roadster, and both have large commitments to GPU-accelerated data center build-outs and global-scale solar manufacturing. Starlink antennas are going into Cybercabs, Grok powers Tesla’s in-vehicle experience, and Musk’s stated payload for the first uncrewed Mars missions includes Cybertrucks and Optimus bots.
          • Managerial attention is shared. Perhaps most important, both companies share Musk’s attention. A more complete integration would let both reach their potential with less friction.
          • Our modeling has Optimus needing roughly 4 GW by 2031. Tesla is targeting about 0.4 GW this year, while SpaceX is heading toward 13–23 GW by the end of 2027.
          • Deploying 10 GW of NVIDIA Vera Rubin systems could cost more than $400 billion before data center costs, which would require both debt and equity financing. Our modeling suggests Tesla’s Cybercab and robotaxi network could prove extraordinarily cash-flow generative, and SpaceX will have plenty of opportunities to put that cash to good use.
          • Chips and power. Terafab has a long-term cost above $100 billion, with volume production targeted for Q3 2028. Musk says SpaceX’s binding constraint is power and cooling, and Tesla’s Megapack business already supplies it ($221 million in Q3).
          • Each needs what the other has. At a high level, Starlink’s potential is dwarfed by Tesla’s robotaxis, which in turn will be swamped by the orbital and terrestrial compute opportunity. Optimus is potentially larger still, but we believe it will need SpaceX’s superintelligence to scale.

           

          How Merger Terms Allocate Shareholder Value

          Using our 2031 models and October 23 closing prices, we examined how the terms of a share swap would divide the incremental value between the two shareholder bases. With no premium to either company, Tesla shareholders would see a 3.8-fold increase in returns on each invested dollar, and SpaceX shareholders 7.2 times. Splitting the added upside equally implies SpaceX paying about 24% over Tesla’s closing price, which adds roughly 5.9 cash-on-cash turns for both. Even at a 50% premium, SpaceX holders gain 4.8 additional turns.

          In the equal-split case, Tesla shareholders would own about 42% of the combined company’s projected diluted market cap in 2031, with the balance of the economics leaning toward SpaceX: roughly 62% of revenue and 69% of earnings before interest and taxes, with robotaxi cash flow an important enabler along the way. Ultimately, the boards and shareholders will determine fair value, but our research points to a highly accretive transaction for both groups.

           

          What To Watch: Signs A Combination Is Advancing, And Signs It Is Stalling

          Area Signs a combination is advancing Signs it is stalling
          Process Committees report and terms are announced ahead of Tesla’s December 3 annual meeting No agreement on terms, or the process stalls or slips well into 2027
          Valuation An exchange ratio that Tesla holders accept without a punitive premium A premium so large that Tesla holders reject it, or the vote falls short of the majority needed
          Strategic overlap Terafab milestones met, with test chip in May or June; more joint projects Terafab or Optimus delays, or the companies pursuing separate paths
          China and defense Tesla’s China operations ring-fenced without friction Regulatory or national-security objections tied to SpaceX’s defense work

           

          The Balanced View And The Bottom Line

          Nothing is decided, and some sell-side analysts see limited conventional financial synergies, so the case rests on integrating the two companies’ AI technologies. Tesla’s proxy for its December 3 meeting does not mention a merger, and Tesla is investing ahead of revenue, which is weighing on near-term earnings. On the SpaceX side, short-term contracts at spot-like pricing are a strength while compute is scarce, but they leave revenue exposed if pricing softens. Even so, we expect the two companies to come to terms.

          We are also proactively reviewing how likely a combination is and what it would mean for the portfolios we manage. Prediction markets see a meaningful chance, with Kalshi contracts implying roughly 60% odds of a merger before May 2027 and about 77% before 2028.1 Our view draws on public information and our own research into why the businesses fit together, rather than on anything beyond management’s public statements.

          Bottom line. SpaceX reported what an AI infrastructure build-out looks like, and Tesla reported what the businesses that depend on it look like. Whether or not a combination happens, we think the two stories are converging.

          References

          1

          Kalshi, as of October 21, 2026, as reported. Prediction market odds change frequently. ARK’s venture strategy holds an investment in Kalshi.

          Related Posts

          You are leaving europe.ark-funds.com

          By clicking below you acknowledge that you are navigating away from europe.ark-funds and will be connected to ark-funds.com. ARK Investment Management LLC manages both web domains. Please take note of ARK’s privacy policy, terms of use, and disclosures that may vary between sites.

          Cancel Proceed
          ======