Two spacecraft built to beam power reached orbit on 1 October 2026. Neither has beamed anything yet,
\nevery number attached to them is a target rather than a measurement, and when the first results arrive
\nthere is no published method by which anyone outside the two companies can check them.
\nThe launch itself is well attested. Via Satellite reported on 1 October 2026 that SpaceX's
\nTransporter-18 lifted from Vandenberg at 11:32 a.m. PT with 130 payloads aboard, with deployment
\nconcluding at 3:41 p.m. ET. Two of those payloads are power beamers. Star Catcher's Protostar is a
\nprototype power node intended to beam power to another free-flying spacecraft. Cowboy Space's
\nReason-1 is intended to send power the other way, from low Earth orbit down to a receiver on the
\nground.
\nThe figures circulating for Reason-1 come from trade press, not from the company. Payload reported on
\n1 October 2026, under Victoria Woodburn's byline, a goal of 30 to 100 W delivered into a spot of 10
\nto 20 m at a ground facility, with a first beaming attempt possible as soon as late October, and
\ndescribed the spacecraft as carrying the "first commercial kilowatt-class laser in space designed to
\ndeliver energy to Earth". Via Satellite, same day, says only that Reason-1 is designed to prove out
\nusing a high-power laser to move energy from orbit to a ground receiver. The same laser is reported
\nto return in the first half of 2027 on Reason-2 as a data link rather than a power link.
\nThose delivery figures are targets. Treat them as documented statements of intent and nothing more.
\nThey are also the one pair of numbers in this story worth holding onto, because they are quoted at
\nthe two ends that matter. A kilowatt-class source is a transmitter rating. Thirty to a hundred watts
\nis what is meant to land. If the source is taken at a round kilowatt, the implied end-to-end delivery
\nis roughly 3 to 10 per cent, and that is my arithmetic on two reported targets rather than anybody's
\npublished budget, so treat it as an estimate. It is not a criticism. A first flight that delivers a
\nfew per cent of its source power to a spot on the ground would be a significant result. It is only a
\nproblem when the kilowatt travels alone in a headline and the watts stay behind.
\nStar Catcher has published a first-party statement, which is more than the other side of this story
\nhas. Its release of 28 September 2026 says Protostar will conduct a "landmark power beaming
\ndemonstration between two untethered spacecraft, a feat that has never been achieved". That release
\nnames no power level, no wavelength, no separation distance and no customer for the demonstration
\nitself. Payload's description adds that Protostar will deploy a cubesat and "beam measurable power to
\nits off-the-shelf solar panels", and names Starcloud and Aethero as publicised customers of the wider
\ngrid the company is selling. The company's own updates index carries a November 2025 item about a
\nrecord-setting ground test at Kennedy Space Center. I could not get that page to render its figures
\nto a fetch on 3 October 2026, so I am not restating them here.
\nOn the Cowboy Space side the first-party record is close to empty. On 3 October 2026 the company's
\nown site described its Stampede constellation as satellites harnessing solar power in low Earth orbit
\nto run on-orbit GPU data centres, with no mention of Reason-1, no mention of beaming power to Earth,
\nand no watt figure anywhere. Its news index returned a placeholder: "Nothing to report yet. Check
\nback soon." The company does appear to have described the mission on its own X account, but that post
\ncould not be retrieved for this piece, so everything above rests on trade press. Readers should weigh
\nit accordingly.
\nOne correction belongs here rather than in a footnote. The register this project keeps recorded, on
\n5 September 2026, that this company had left power beaming for orbital data centres, and called it
\nthe first programme in that table to leave the field rather than enter it. That was wrong. The
\nhardware flew. The data-centre pitch turned out to be where the beaming work is pointed, not a
\nreplacement for it.
\nSpace-to-Earth beaming already has a precedent, and the precedent is instructive about numbers.
\nCaltech announced on 1 June 2023 that its MAPLE experiment on SSPD-1 had transmitted power wirelessly
\nin space and beamed detectable power to Earth, picked up on 22 May 2023 by a receiver on the roof of the
\nGordon and Betty Moore Laboratory of Engineering in Pasadena. The release describes the received
\nsignal as arriving at the expected time and frequency with the predicted frequency shift. It gives no
\nfigure in watts for what arrived. Caltech's own wrap-up of the mission, published in January 2024,
\nsays MAPLE demonstrated its ability to direct a beam to Earth and calls that a first. It also
\npublishes no delivered-power number.
\nThat is not a criticism of Caltech either. Detecting a beam and delivering usable power are different
\nexperiments, and the team said which one it had done. But it means the entire published history of
\nspace-to-Earth power beaming currently consists of one detection with no watt figure attached. The
\nnext entry in that history is likely to be written in the next few weeks, by a company that operates
\nboth the transmitter and the receiver, and reports the result itself.
\nHere is where the standards position gets uncomfortable, and it is the part of this story that is
\ncheckable today rather than in late October.
\nThe ITU Radio Regulations, Edition of 2024, define radio waves at No. 1.5 as "Electromagnetic waves
\nof frequencies arbitrarily lower than 3 000 GHz, propagated in space without artificial guide". Three
\nthousand gigahertz is a wavelength of 100 micrometres. A practical high-power infrared laser runs
\nnear one micrometre, which is about two orders of magnitude above the ceiling the Regulations set for
\nthemselves. An optical downlink is therefore not a radio emission in the regulatory sense, and the
\nmachinery that would otherwise apply does not reach it.
\nThe exposure standards go the same way. ICNIRP's current radiofrequency guidelines, published in
\n2020, cover 100 kHz to 300 GHz. IEEE Std C95.1-2019, an active standard published on 4 October 2019,
\ncovers 0 Hz to 300 GHz. An infrared beam sits above both. What covers it instead is ICNIRP's laser
\nguidelines, published in 2013, which set exposure limits for wavelengths between 180 nm and 1 000
\nmicrometres and for exposure durations from 100 femtoseconds to 30 kiloseconds. Those guidelines are
\nexplicit that detailed measurement procedures and calculation methods are beyond their scope, and
\npoint to a literature from 1980 to 2004 for the methods instead.
\nSo a ground spot of 10 to 20 metres, illuminated from orbit at kilowatt-class source power, is
\ngoverned by an exposure limit written for laboratory and industrial optics, with the measurement
\nmethod deliberately left outside the document. One standard does carry measurement conditions, and it
\nis the wrong one for this purpose: IEC 60825-1, edition 3.0 of 2014, classifies laser products by
\nhazard across 180 nm to 1 mm. It tells you what class a device is. It does not tell you how to
\nreport the power a beam from orbit puts into a receiver on the ground. On the radio side, the ITU's only publication on
\nbeamed wireless power, SM.2392-2 of June 2026, is a Report, which recommends nothing and binds
\nnothing, and the only Recommendation in the area, SM.2151-0 of September 2022, is scoped to
\nmobile and portable devices and sensor networks. Neither reaches a spacecraft, and neither is about
\noptics at all.
\nThis project is implicated in exactly the same gap. The standards registry behind the BeamDesk
\nconsole pins ICNIRP 2020 and C95.1-2019 for exposure, and both stop at 300 GHz. Its optical entries
\ncover beam propagation, not human exposure and not measurement. A console that can tell you the power
\nflux density under a microwave beam cannot tell you the equivalent under an infrared one, and saying
\nso is cheaper than being found out.
\nFour things would make the late-October attempt, or Protostar's, into evidence rather than an
\nannouncement. The watts quoted at the receiver, not at the source. The aperture and spot size that
\nproduced them, so the figure can be recomputed. The integration time, because a peak and a sustained
\naverage are different claims. And a receiver measurement someone other than the transmitting company
\ncan stand behind.
\nNone of that requires a new standards body. It requires a published method and a witness. The
\nawkwardness is that the first of those does not exist yet for optical power delivered from orbit, and
\nthe second is nobody's obligation. Until both arrive, the honest description of any number that comes
\nback from these two spacecraft is that it is a company-reported figure from a company-operated
\nreceiver, and that is how it will be recorded here.
\nA delivered-power figure measured at the receiver, published with aperture, spot size and integration
\ntime, and witnessed by a party that does not operate the transmitter, would make my central complaint
\nobsolete for that mission. I would record it as a measurement and say so.
\nA chartered measurement project would change the structural half. If IEC TC 76, ICNIRP or an IEEE
\nworking group opened work on measuring and assessing high-power optical beams delivered from space,
\nthat would be the most consequential development in this area and I would report it as good news,
\nwithout hedging.
\nI would also revise the first-party criticism quickly if Cowboy Space publishes a mission page with a
\npower budget on it. The company having said nothing on its own site on 3 October 2026 is a fact about
\n3 October, not about the company's intent, and a single page would move the Reason-1 figures from
\ntrade press to first-party overnight.
\n