Three objects
2017, 2019, 2025.
1I/ʻOumuamua was found in October 2017, three days after perihelion and already outbound. It showed no coma, an extreme apparent axis ratio, and an unexplained non-gravitational acceleration, and it was gone before anyone could do more than argue about it.
2I/Borisov followed in August 2019, roughly three months before perihelion: an unambiguous comet, unusually rich in carbon monoxide.
3I/ATLAS was discovered by the ATLAS survey telescope at Rio Hurtado, Chile on 1 July 2025, at 4.4 AU on an inbound trajectory, four months before a perihelion at 1.358 AU. Archival searches extended the orbital arc back to 22 May. It was found in Sagittarius, close to the galactic plane, in exactly the crowded region most surveys avoid.
What 3I/ATLAS gave us
From a modest, largely ground-based campaign.
| Property | Value | Implication |
|---|---|---|
| Eccentricity | ~6.14 | The most hyperbolic of the three |
| Hyperbolic excess velocity | 57.98 km/s | Sets the rendezvous requirement |
| Inclination | ~175°, retrograde | Arrived against the direction of planetary motion |
| Spectrum | Red, featureless D-type | Water ice grains in the coma via the 2.0 µm band |
| Nucleus radius | ~5.6 km (Rubin); <2.8 km (Hubble) | The two bounds disagree |
| Galactic kinematics | Thick disk | Kinematic age 3–11 billion years |
| Semi-heavy water | ~30× solar system comets | Formation at extremely low temperature |
| Technosignature scan | None, 1–9 GHz | Seven-hour Allen Telescope Array scan, begun within a day |
The thick-disk kinematics imply an origin in the era of intense star formation astronomers call cosmic noon, and the semi-heavy water ratio implies formation in a cold, isolated region — quite possibly predating its own host star.
Stop and look at what that list is. It is the composition, formation temperature, isotopic history and approximate age of a planetesimal from another star system, possibly six billion years older than the Sun, obtained with existing telescopes at a marginal cost that rounds to zero against any spacecraft program. Nothing else in this section delivers a comparable result before the 22nd century.
The campaign continued past perihelion, which was 29 October 2025. Subaru observed the object on 7 January 2026 and measured the carbon dioxide to water ratio in the coma to be notably lower than pre-perihelion space-telescope measurements, implying that the composition of the nucleus interior differs from its exterior — heating during the solar passage drove gases from different layers. That result was published in The Astronomical Journal on 22 April 2026, and it is the kind of depth-resolved compositional measurement nobody expected to get from an interstellar visitor.
Seven orders of magnitude
The energy argument, stated plainly.
Reaching Proxima at 0.2c costs 515 gigawatt-hours per kilogram. Intercepting an interstellar object inside 5 AU costs a velocity change on the order of 15 kilometers per second, which is 31 kilowatt-hours per kilogram.
Interstellar material is available at seven orders of magnitude less energy per kilogram of delivered instrument than interstellar destinations.
And it is not a lesser measurement. A flyby at Proxima returns remote sensing of a planet from a gram of instruments moving at 60,000 kilometers per second. An interstellar object rendezvous returns in-situ mass spectrometry, isotopic ratios, mineralogy and potentially a physical sample, from a spacecraft that can take its time.
Flyby, rendezvous, sample
Completely ordinary mass ratios.
A flyby at 58 kilometers per second relative velocity lasts minutes and is close to useless. Rendezvous means matching the hyperbolic excess velocity, which for 3I/ATLAS is 58 kilometers per second.
| Mission | Velocity change | Mass ratio at 50 km/s exhaust | Verdict |
|---|---|---|---|
| Flyby | None | 1 | Minutes of useful observation |
| Rendezvous | 58 km/s | 3.2 | Ordinary |
| Sample return | 58 km/s, then departure | ~10 | Ordinary |
| For comparison: 0.1c interstellar flyby | 30,000 km/s | 10²⁶¹ | Impossible |
Matching velocities with a piece of another star system is within the capability of propulsion this architecture already places in the 2040s. Interstellar sample return is achievable without an interstellar mission. It requires only that something from another star comes close enough, and something does.
The binding constraint is not propulsion. It is notice.
The scarce resource is warning time
And the trend is monotonic.
Detection is not the scarce resource. Lead time is, and the trend across the three known objects is the entire story.
| Object | Discovered | Relative to perihelion | Consequence |
|---|---|---|---|
| 1I/ʻOumuamua | Oct 2017 | 3 days after, outbound | Unreachable by anything; characterized only in the argument afterwards |
| 2I/Borisov | Aug 2019 | ~3 months before | Extended observing campaign possible; no intercept |
| 3I/ATLAS | Jul 2025 | 4 months before, at 4.4 AU inbound | Full multi-observatory campaign through perihelion; intercept studies published within weeks |
Rubin’s depth and cadence push the typical detection distance outward. An object found at 5 to 10 AU inbound converts a three-day warning into a one to three year warning, and a loiter vehicle needs roughly a year of notice to plan, depart and match a trajectory. Rubin is what supplies the year.
The other half of that is parked capability. A vehicle waiting in the inner system with a beam-fed departure available can take 20 to 50 kilometers per second on a few weeks’ notice, in whatever direction the object turns out to be coming from. Parked Δv is the binding constraint on this mission class, and a beam array is parked Δv — which converts an intercept from a lucky opportunity into a standing program.
Three objects is the whole sample, and the site will say so until a fourth is confirmed. Rubin recovered pre-discovery images of 3I/ATLAS in its own commissioning data, which is the mechanism by which the next one gets found earlier rather than later. A confirmed fourth object is the single update most likely to change the numbers on this page.