SpaceX carried NASA's whole $4.3bn Roman telescope on a single Falcon Heavy

Roman flew with a fuel port, or rather with the assumption that something will eventually arrive to use it: the telescope is designed to be refuelled, and the framing NASA offered on launch day was that a robotic tanker becoming available in the coming decade could extend its life ^[13]. That is an option written into a $4.3bn asset ^[1] with no counterparty, no price and no delivery date, which is a strange thing to find on the books of a government science program and a very ordinary thing to find in a venture portfolio. Build the port, assume a servicing market shows up. If it does not, the mass and volume given over to it were spent for nothing.

One launch measures no market share, and I am not going to pretend otherwise. What the flight record establishes is narrower and, for underwriting purposes, enough: the account of the launch names SpaceX and Falcon Heavy and no competing vehicle in the class ^[18], and the moment the payload separated from the upper stage ^[6] is the moment the risk on the entire appropriation transferred. Two readings price differently. If several qualified vehicles could have flown this mass to that point and NASA chose one, the dependency is a procurement preference and reversible. If only one could, then the launch line on the next flagship is quoted by a party with no reference competitor, and every dollar of restraint in that quote is a courtesy rather than a market outcome.

The science throughput deserves the same reading as a vendor spec sheet. Senior project scientist Julie McEnery's comparison, that a month of Milky Way observing by Roman would take Hubble a century ^[9], works out to roughly 1,200 Hubble-months per Roman month ^[15], while the wide-field infrared camera is specified at 1,000 times Hubble's speed at matching sensitivity ^[10]. The 20 percent gap ^[16] is nothing to an astronomer and everything to anyone who has priced a throughput claim before. Run the same rate the other way and Hubble's whole 36 years of orbit compresses to something near four and a half months of Roman survey time ^[19], which is the number that tells you why the agency accepted the concentration risk in the first place.

This is probably wrong, but I read the fuel port as the more consequential term than the rocket. A launch dependency is a single transaction with a known loss case; a servicing dependency is recurring revenue for whoever owns the tanker, and it arrives attached to an asset the government has already declared it wants to keep alive. The thesis fails if the tanker never gets funded, in which case Roman is simply a well-built instrument that runs out of propellant on its original schedule, and it fails differently if a second provider flies comparable mass to the same region before the servicing market exists, which would show the concentration was a timing artifact rather than a structure. The money is already spent; the number I want next is whatever a tanker costs.