Musk Lifts Tesla Voting Power to 20% Days After the $1.7 Trillion SpaceX-xAI IPO

By exercising 304 million options from his 2018 award five days after the newly merged SpaceX-xAI began trading, Musk is quietly assembling the control stack needed to bolt Tesla onto a nascent $4 trillion machine.

The debate over whether Tesla and SpaceX would ever combine is effectively settled, because two of the three pieces already have. The AI brain and the rocket company merged, went public, and became the financial and computational foundation for everything Musk touches. What remains is the hardest piece — not because Tesla is technically difficult to fold in, but because it is the one asset owned by outside shareholders who sue, litigate, and demand a fair price. Every move Musk has made since the IPO reads as scaffolding for that final, contentious combination.

Key Takeaways

• February 2026 saw xAI and SpaceX merge in an all-stock deal valued near $1.25 trillion, fusing the model and the rockets into one entity.
• On June 12, 2026, that combined company listed on NASDAQ under SPCX at roughly $1.7 trillion — the largest IPO in history — surging to about $2.4 trillion within a week.
• Five days later, Musk exercised 304 million shares from his 2018 award at $23 apiece, lifting his Tesla voting power from around 13% to 20%.
• Paying the strike and tax required selling 17 million shares for about $7 billion, with the remaining stock locked until January 2028.
• SpaceX priced its first investment-grade bond in June 2026, upsizing from $20 billion to $25 billion against a ~$90 billion order book, oversubscribed roughly 3.5 times.
• Committed outside compute revenue on the former Colossus cluster now tops $80 billion through 2029, anchored by Anthropic at ~$45 billion and Google at ~$920 million per month.
• Terafab, the three-way chip megafab near Austin, has seen cost estimates balloon from $25 billion in March to as much as $120 billion by May.
• November 2025's shareholder-approved pay package — 12 tranches, ~424 million shares, up to ~$1 trillion — walks Musk toward roughly 25% voting control.

---

The IPO That Rearranged the Board

The sequence matters more than any single number. xAI and SpaceX combined privately, then the merged company hit the NASDAQ at a valuation that instantly became the biggest public debut ever recorded. Within a week the market had bid it up by hundreds of billions, validating the thesis that a rocket company carrying a frontier AI lab is worth more than the sum of its parts.

That listing did something structural: it created a public, deep-pocketed, credibly-rated entity that can serve as the balance sheet and compute engine for the entire ecosystem. Tesla is now the outlier — the only major piece still standing outside the fused machine, and the only one whose combination will be fought over in open court.

Reading the Five-Day Gap

SpaceX started trading on June 12. Musk moved on his Tesla options on June 17. That is not a coincidence in search of a narrative; it is a narrative in search of an obvious explanation. Exercising the 2018 award converted roughly 304 million options into restricted stock carrying immediate voting rights, jumping his stake from about 13% to 20% and putting him near 700 million shares — a fifth of the company.

The mechanics are unglamorous and revealing. Options aren't free: you owe the strike price and the tax bill, which is why 17 million shares were sold for around $7 billion. The rest stays restricted until early 2028. What Musk bought with that $7 billion was not liquidity — it was votes, and votes are the currency this merger runs on.

The Long March to 25%

The June exercise is a single step in a march that began in earnest in November 2025, when Tesla shareholders approved a new pay package with over 75% support. Twelve tranches, roughly 424 million shares, up to a trillion dollars over a decade, gated by milestones that read like science fiction: an $8.5 trillion market cap, 20 million vehicles, a million Optimus robots deployed.

As those tranches vest, Musk's ownership climbs toward about 25% voting control. That figure is deliberate. He named it publicly in January 2024 — enough to be influential, not so much that he can't be overturned. The pay package and the options exercise are two instruments tuned to the same target.

Negotiating With Himself

On the SpaceX side, super-voting shares give Musk something like 82–85% of the vote. Approaching 25% on the Tesla side, he sits on both sides of any deal between the two — decisive control over one party, a commanding block in the other. He is, functionally, negotiating across the table from his own reflection.

Here is the counterintuitive part: more votes help you win the shareholder vote and simultaneously make the resulting deal harder to defend. When a controlling insider sits on both sides, the legal tests tighten rather than relax. Passing the vote and surviving the inevitable lawsuit are two entirely different problems — which is precisely why the quiet pieces merged first and the litigated one was left for last.

Why Tesla Comes Last — The Delaware Problem

The SolarCity acquisition of 2016 is the template. Musk was on both sides of that $2.6 billion deal too, and it survived a full Delaware trial only because there was a genuinely independent board, a minority-shareholder vote that excluded Musk's own shares, and a defensible price. That is a brutal gauntlet.

The xAI-SpaceX merger simply declined to run it — staying private and incorporating in Nevada, where an insider essentially has to be shown to have committed known misconduct, versus Delaware's much steeper "entirely fair" standard. Tesla enjoys no such shelter. It is public, heavily owned by outsiders, and constantly in litigation — shareholders are already fighting over Tesla's $2 billion investment into xAI from January 2026. The merger order came down to friction: the frictionless pieces went first.

SpaceX as the Financial Foundation

The IPO handed Musk a cleaner cap table and a balance sheet. In June 2026 SpaceX priced its first investment-grade bond, targeting $20 billion, upsizing to $25 billion against an order book near $90 billion — roughly 3.5 times oversubscribed. Five tranches, coupons from about 5.35% to 6.65%, maturities out to 2056, costing around $1.5 billion a year in interest, underwritten by the largest banks on Earth and rated investment grade across the board.

The proceeds refinance the expensive debt left over from the 2022 Twitter acquisition and the costly 2025 xAI loans, with SpaceX disclosing a cash pile near $100 billion. The whole ecosystem's debt now routes through SpaceX and gets refinanced at cheap public-market rates. Any Tesla combination leans on this foundation.

The Compute Landlord and the Silicon

SpaceX inherited Colossus — the supercomputer that trained Grok — and flipped it into a multi-tenant commercial cloud. Reflection AI signed on for $150 million a month, about $6.3 billion through 2029, layered atop Google at roughly $920 million monthly and Anthropic at around $45 billion. Committed outside compute revenue now exceeds $80 billion through 2029, a backlog that both services the new bond and makes SpaceX a credible AI infrastructure partner for the exact things Tesla needs: autonomy and Optimus.

Then there is silicon. Terafab, the megafab near Austin, is a three-way joint venture across Tesla, SpaceX, and xAI — and the fact that it must be a three-way JV is itself proof the corporate walls are already in the way. Estimates climbed from $25 billion in March to as much as $120 billion by May. Tesla killed its Dojo supercomputer in August 2025, declaring all paths converge on AI6, then quietly revived the program months later. The AI5 inference chip taped out in April 2026, claimed to run roughly 10 times cheaper than NVIDIA hardware; Grok already ships in millions of cars and inside Optimus Gen 3.

Producers and Spenders of Intelligence

Strip away the ticker symbols and there are two kinds of companies here. Tesla acts on intelligence — every car and robot is an edge node converting a finished model into physical motion, a steering input, a left turn through traffic. SpaceX and xAI produce it — training the model, generating inference in orbit, beaming it down via Starlink, and driving down the cost of putting compute anywhere with Starship.

Right now those two costs live in separate companies, and every unit of intelligence crossing from producer to consumer gets taxed: IP licensing, related-party reviews, duplicate overhead, lawyers on both sides. Fuse them and that friction approaches zero, while the fixed cost of training the frontier model — the single largest cost in all of AI — amortizes across robotaxis, millions of robots, satellites, and paying cloud tenants. It is the Apple playbook at its logical extreme: own the chip, the OS, and the device, so every layer subsidizes the next — except the device is a car, a humanoid, and a satellite.

Inevitable Is Not the Same as Fair

Put the pieces together and you get a roughly $4 trillion entity whose combined profit would demand a heroic multiple to justify today's valuation. Tesla's GAAP profit is positive but small relative to the whole. That sets up a genuinely tense fight among investors — many of whom own both companies — over the relative valuation each side uses to buy the other. Higher relative value favors Tesla holders; lower favors SpaceX holders.

Given Musk's voting power and the deference his execution earns, the price will likely land near what he decides is right. That is the uncomfortable core: the deal can be entirely inevitable and still be a bad trade for one side of the table. The combination is getting done — there are too many interlocking pieces for it not to. The open questions are what Musk decides, how Tesla's minority holders take it, and whether it survives the lawsuits that are as certain as the merger itself.