Search SAM.gov for a novation request. Search EDGAR for a Form S-4. Search the FTC's early termination notices for a filing pair. Nothing. The Tesla–SpaceX merger, as of this writing, exists only as a proposal — no term sheet, no exchange ratio, no merger agreement, no HSR submission. That absence is the most informative data point in the entire story, because it tells you which gates the people floating the idea have already priced and which ones they have never opened.
Here is the anomaly. Nearly every commentary written on this proposal spends its budget on Clayton Act §7 — whether the combination substantially lessens competition in satellite broadband, in enterprise energy management, in autonomous-vehicle telematics. Almost none of them mention 32 C.F.R. Part 117. That is the National Industrial Security Program Operating Manual, and it governs whether a cleared defense contractor may sit inside an ownership structure with a large non-U.S. passive shareholder base. SpaceX is not a bystander in federal procurement. It is a prime contractor on crew transport, on national security launch, on Space Development Agency tranches. A change of control runs through that manual before it runs through any antitrust courtroom — and unlike a court, the manual produces no public opinion. It produces a determination. Sometimes a classified annex. Sometimes a condition the parties cannot renegotiate.
Context: what is actually being combined
The proposal's mechanics are simple, and that is exactly why they are dangerous. Tesla is a Delaware corporation listed on Nasdaq, with quarterly disclosure obligations, a Form S-4 requirement for any stock-funded acquisition, and Item 404 related-party disclosure obligations under Regulation S-K. SpaceX is a private company with a capped table, federal contracts, and no obligation to explain itself to anyone. One person is chief executive of both. Any structure that combines them — stock swap, holding company, asset contribution, dual-class umbrella — has to reconcile three different accountability regimes inside a single cap table.
The legal chassis is not complicated, though it is frequently misdescribed. The substantive review standard is Clayton §7, 15 U.S.C. §18: whether the effect may be substantially to lessen competition, or to tend to create a monopoly. The procedural gate is HSR, Clayton §7A, 15 U.S.C. §18a — file, wait, do not close. Conduct after closing — access terms, API pricing, exclusivity — falls under Sherman §1 and §2 and FTC Act §5. The phrase "exclusionary behavior" that circulates in the commentary is not a statute. It is vertical foreclosure theory, and it is a §7 argument about input denial, not an independent cause of action.
For two years, commentary hardened around a single sentence: regulators historically view vertical mergers as pro-competitive. That sentence was accurate under the 2020 Vertical Merger Guidelines. The FTC withdrew those guidelines on September 15, 2021; the DOJ did not join the withdrawal, so for a period the two agencies operated on incompatible standards. The 2023 Merger Guidelines unified them in the other direction. Anyone still quoting the 2020 consensus is describing a framework that has been decommissioned. This is the first systematic error in the public analysis: the risk was priced against a rulebook that no longer exists.
The clause nobody reads
15 U.S.C. §19 — Clayton §8 — prohibits interlocking directors and officers between corporations that are competitors, subject to annually indexed de minimis safe harbors. It does not require a market definition exercise. It does not require a showing of effect. It does not wait for a merger to close. It applies today.
Musk holds the chief executive title at both entities. If any plausible relevant market treats Tesla and SpaceX as competitors — satellite terminals embedded in vehicles, enterprise energy storage and management, on-site power for launch infrastructure, and increasingly compute capacity — §8 is a standalone violation that exists before anyone files anything. DOJ's §8 enforcement posture has been visibly more active since 2022, producing forced board resignations across several public companies. The remedy is trivial to state and expensive to execute: someone surrenders an operational role. In a transaction whose entire thesis is that a single operator should run both, §8 is not a technicality. It is a structural contradiction written into the statute that the commentary skips.
Two cases, two return values
The controlling analogues point in opposite directions, which is why the outcome distribution on this deal is wide rather than binary.
FTC v. Illumina/Grail. The FTC challenged a vertical acquisition of an early-stage competitor by the dominant sequencing incumbent. Illumina litigated. The Fifth Circuit declined to stay the administrative process in 2023, and Illumina abandoned the transaction and divested Grail. The theory that survived was not price effects. It was foreclosure of an innovation path — rivals could not develop competing assays without access to the incumbent's platform.
United States v. AT&T/Time Warner. DOJ advanced a pure vertical theory and lost at the district court in 2018, lost again at the D.C. Circuit in 2019, and the court accepted arbitration-style behavioral commitments as a sufficient remedy.
The differentiator is not ideology. It is whether the input is a critical, non-substitutable chokepoint. Starlink's low-Earth-orbit broadband position is structurally closer to Illumina than to a content library. If a relevant market is drawn around vehicle connectivity for autonomy — where latency, coverage, and uplink density are the binding constraints — Starlink is not one supplier among several. It is the input. And Guideline 6 of the 2023 framework asks precisely the foreclosure question that the 2020 framework treated as a rounding error. Guideline 7, on entrenching or extending a dominant position, does the same for the second-order argument: vertical integration as a mechanism for exporting an existing advantage into an adjacent market.
The procedural wall
If a filing lands after February 10, 2025, it lands under the revised HSR rules. The expansion is not cosmetic: transaction rationale documents, narrative descriptions of business-line overlap, five years of prior acquisitions, and — this is the part that matters here — employee and labor-market information. Preparation time moves from weeks to months before the initial waiting period even begins. The practical effect is a procedural barrier that functions as a substantive one for any deal carrying a financing clock.
And the labor-market disclosure is not neutral for Musk-affiliated entities, which carry a documented history of employment litigation. The filing itself opens an investigative window that has nothing to do with satellite broadband share. Procedural cost is a filter, not a fee. Filters select for deals with unlimited time.
The reverts that fire first
Now the uncovered part. SpaceX holds federal contracts. A change of control triggers FAR 42.12 novation — the government must recognize a successor in interest, and it may refuse. It triggers FAR 9.5 organizational conflict of interest review, because a merged entity that sells launch services and also supplies the payload infrastructure sits on both sides of a procurement. And it triggers NISPOM facility clearance review, including foreign ownership, control, or influence analysis.
FOCI is where this deal breaks in a way no antitrust remedy patches. Tesla's shareholder base is dominated by index and passive vehicles with substantial non-U.S. beneficial ownership. FOCI mitigation for a cleared contractor typically takes one of four forms: board resolution, proxy agreement, voting trust, or special security agreement. Every one of them restricts the parent's ability to exercise control over the cleared subsidiary.
Read that against the merger thesis. The stated value of combining the two companies is operational integration — shared capital allocation, shared engineering roadmap, shared data pipes. FOCI mitigation is, by construction, a legal instrument that denies the parent operational control. You cannot satisfy a security review with a voting trust and simultaneously satisfy the integration thesis with consolidated control. The two remedies are mutually exclusive, and they are administered by agencies that do not negotiate with each other.
When I built a verifiable inference oracle for off-chain model execution, the hard problem was never proving the computation. It was proving the prover had authority to run it. Authority is a separate state machine from correctness, and it can revert the transaction before correctness is ever evaluated. This deal has that shape. Part 117 is an authority check sitting earlier in the call stack than §7's correctness check.
Four oracles, one composite
Jurisdiction is not a single gate. If EU turnover thresholds are met, the Commission reviews. The General Court's 2022 judgment in Illumina/Grail (T-227/21) annulled the Commission's expansive Article 22 referral policy, so non-notifiable deals can no longer be pulled up as a matter of course — but a notifiable deal remains notifiable. The UK CMA asserts jurisdiction on a share-of-supply test with no turnover floor and has repeatedly taken cases below EU and U.S. thresholds. China's SAMR, under the 2022 amended Anti-Monopoly Law, holds penalty exposure of up to 10% of prior-year turnover for substantive violations and, under the 2024 revised filing thresholds, retains authority to investigate below-threshold transactions with exclusionary effects.
Four approval oracles. Each returns a boolean. The transaction executes only if all return true. Latency is additive, not parallel, because remedies issued in one jurisdiction constrain the commitments available in another. A twelve-to-eighteen-month composite timeline is the base case, and every month of it carries financing cost, management distraction, and — in the current market — a widening merger-arbitrage spread that prices precisely this kind of indeterminate tail.
The blind spot
The consensus error is not that people are pricing antitrust. It is that they are pricing only antitrust, and pricing it against a 2020 framework. Two exposures sit outside that model entirely: the interlock prohibition, which is already live, and the security review, which is classified. One of these is cheap to cure and politically expensive. The other is expensive to cure and politically invisible.
There is a deeper inversion. In a vertical case, the efficiency story and the foreclosure story are the same sentence read at different speeds. "Integration lets vehicles use satellite capacity more efficiently" and "integration lets the merged entity degrade rivals' access to that same capacity" are not competing claims. They are one claim. That is the structural reason vertical cases resist balancing, and the reason the 2023 framework replaced balancing with screens.
And behavioral commitments are not protocol invariants. When the AT&T court blessed arbitration-style remedies, it created a promise enforceable only through a subsequent enforcement action, years later, at the discretion of an agency whose priorities have changed. A commitment about future capacity access has no escrow, no automatic penalty, no slashing condition. Audits are opinions, not guarantees — and consent decrees are opinions with a docket number.
Takeaway
Watch two artifacts, not headlines. First, whether a role separation occurs that cures §8 before any filing exists — that is the cheapest signal that someone has actually run the analysis. Second, whether a novation request appears on SAM.gov. If neither surfaces within two quarters, the proposal was never a transaction. It was a narrative instrument.
The rational structure, if this is real, is not a merger at all. It is asset-level contracts — capacity agreements, supply agreements, joint development agreements — which clear every gate with less latency, no interlock exposure, and no classified review.
The question is not whether this merger can be approved. The question is whether anyone involved has executed the state machine to completion, or whether they are still quoting a guideline that was withdrawn in September 2021.