The European Space Agency's (ESA) Euclid space telescope has released "DR1-Foundation," a large dataset assembled from roughly 13 months of observations. More than 1 billion objects were captured in visible and near-infrared light, and the data volume reaches about 1.7 PB (roughly 1.7 million GB). Collected from a swath of sky equivalent to about 9,500 full moons, these observations form the backbone of a 3D map that will chart the large-scale structure of the universe.
The release is not a finished map, however, nor does it announce measurements that reveal the nature of dark energy or dark matter. What has reached researchers worldwide is the foundation for future precision science: rigorously calibrated images, spectra, and object catalogs. How will astronomers use this vast observational archive to bring out the invisible skeleton of the universe? Here is where the released data actually stands, and where its scientific limits lie.
1,900 square degrees and a billion objects: what the initial catalog contains
Released on November 12, 2026, DR1-Foundation covers more than 2,000 square degrees of sky across the Wide Survey, the Deep Fields, and auxiliary fields. The Wide Survey alone spans about 1,900 square degrees, a more than 30-fold increase in observed area over the earlier trial release (about 60 square degrees, a few days of observations). The data include more than 120,000 corrected and calibrated image frames from the visible-light camera (VIS) and the Near-Infrared Spectrometer and Photometer (NISP).
Carole Mundell, ESA's Director of Science, said the following on the occasion of the release:
"This major Euclid data release is the largest of its kind in space exploration, laying the foundation for future discoveries in a new realm of astronomy and demonstrating the excellence of European technology, science, and international collaboration."
According to the DR1-Foundation overview paper, the total number of cataloged objects exceeds 1 billion, of which more than 900 million were classified as galaxies by the analysis pipeline. The Wide Survey yielded 817,846,065 classified galaxies, the Deep Fields 49,393,051, and the auxiliary fields 34,427,023.
データを表で見る
| Classified galaxies (Number of objects) | |
|---|---|
| Wide Survey | 817,846,065 |
| Deep Fields | 49,393,051 |
| Auxiliary fields | 34,427,023 |
The figures show that the Wide Survey accounts for the bulk of detected galaxies, while the Deep Fields are observed at extremely high density.
These numbers need to be interpreted with care, though. For example, the more than 60 million quasar candidates in the dataset are an initial sample selected by color classification from photometric data. As the photometric processing paper states, the sample inevitably includes contamination from stars and unresolved faint galaxies, so it is not a definitive quasar catalog. What DR1-Foundation provides is a statistically validated, high-quality set of observations. It is important to be clear about the boundary: it is not a report that has pinned down the nature of dark energy or the expansion rate of the universe itself.
Splitting light into a rainbow: how spectroscopy turns position into depth
No matter how many photographs of the night sky are stacked, they remain flat, two-dimensional projections. To determine where clusters of galaxies sit in the universe and reconstruct the giant filamentary "cosmic web" in three dimensions, information along the line of sight, that is, the distance to each object, is essential.
Euclid tackles this problem with slitless spectroscopy. The grisms on board NISP (optical elements combining a prism and a diffraction grating) disperse an object's light by wavelength, spreading it into a spectrum, a "rainbow of light." DR1-Foundation contains more than 170 million validated red and blue grism spectra. ESA's published figure also confirms more than 160 million spectra.
What astronomers look for in these spectra is a specific emission line, hydrogen-alpha (), emitted by hydrogen gas in star-forming regions. Because the universe is expanding, light from distant galaxies is stretched in wavelength by the time it reaches Earth, shifting toward the red. Measuring the degree of this redshift allows a galaxy's distance to be worked out.
In parallel, the 880 million objects in the 1,900-square-degree Wide Survey were assigned "photometric redshifts," statistical distance estimates derived from differences in color across multiple wavelength bands. By combining precise spectroscopic measurements with comprehensive photometric estimates for huge numbers of objects, images of the flat night sky emerge as three-dimensional structure.
Euclid's strength lies in capturing a wide field of view at once from space, free from atmospheric blurring and extinction. After launching in July 2023, it began full-scale science observations on February 14, 2024, and has now accumulated about 13 months of observing time. Even so, the Wide Survey area released here is only about 14% of the total survey area Euclid plans to cover over its entire mission. The foundation of the 3D map has been laid, but large blank areas remain before it is complete.
15,000 lensing candidates: citizen science and machine learning sort out the distortions
A methodological milestone in DR1-Foundation is the cataloging of strong gravitational lens candidates. Gravitational lensing is the phenomenon in which light from a distant galaxy is bent by the enormous mass of a galaxy or galaxy cluster in front of it, distorting or splitting the image or making it appear as a ring (an Einstein ring). It is a rare means of directly "weighing" the total mass of the foreground object, including not only its stars but also its invisible dark matter halo.
The probability that a foreground galaxy acts as a clean natural lens is extremely low, around one in several thousand or less. Ground-based telescopes miss subtle distortions because of atmospheric turbulence, and earlier space telescopes had survey fields too narrow, making it difficult to collect a sizable number of lens systems under uniform conditions.
The DR1 catalog paper led by Natalie Lines reports 15,072 high-confidence strong lens candidates from the 2,000 square degrees observed. These comprise 6,073 high-certainty "Grade A (sure lens)" candidates and 8,999 likely "Grade B (probable lens)" candidates. As the paper itself notes, the size of this high-confidence sample exceeds the total number of strong lenses humanity has discovered to date, in a single survey.
データを表で見る
| Candidates (Number of candidates) | |
|---|---|
| Grade A (sure) | 6,073 |
| Grade B (probable) | 8,999 |
| Grade C (possible) | 9,010 |
| Grade X (not a lens) | 2,580 |
This selection is not the result of running the pipeline exhaustively on all 900 million detected objects. Starting from about 68 million objects that passed a prior catalog cut, the team scored them with five machine learning models. Candidates were then narrowed down through visual classification by roughly 7,000 volunteers on the "Space Warps" project on the citizen science platform Zooniverse, followed by verification by 67 experts.
The catalog also includes 9,010 "Grade C (possible lens)" objects and 2,580 "Grade X" objects judged not to be lenses. The authors state explicitly that some subjectivity remains at the boundaries between grades. No individual object has been proven by spectroscopy to be a confirmed lens; these are strong candidates based on shape and features.
Automated lens modeling was run on the combined Grade A and B sample, yielding physical quantities such as lens redshifts and stellar masses. The median lens stellar mass is (16th to 84th percentile range ), the median photometric redshift is $0.68+0.33 / -0.29$), and the typical Einstein radius is $0.85$ arcseconds ().
The machine learning training and validation insights gained from the earlier Q1 survey (about 63.1 square degrees, 500 candidates detected) underpinned this scale-up. A lens population on the order of 10,000 was built using a single telescope and common criteria. As a result, strong gravitational lenses, once treated as individual rare objects found by chance, have become targets for large-scale population analysis aimed at uncovering the statistical properties of dark matter halos.
The road to June 2027, when cosmology's real test comes
To read the DR1-Foundation data properly, it helps to understand why the astronomical community split this release into two stages. The batch of scientific papers published alongside the release are all preprints posted to arXiv before submission for peer review. Until peer review and academic validation are complete, interpretation of individual details carries scholarly caveats.
The more fundamental reason was a technical problem with the optical model that arose during pipeline development. During data reprocessing in early 2026, it emerged that the model of the point spread function (PSF, the function describing how an optical system spreads a point source into an image) of the visible-light camera (VIS) had underestimated the actual PSF size.
This small error produces a critical systematic error in analyses of weak gravitational lensing (weak-lensing shear), which measures the slight stretching of object shapes. Redoing the shear analysis correctly requires enormous recomputation, and fewer than six months remained before the planned November 2026 release date, not enough to complete a full cosmological analysis.
For this reason, ESA and the Euclid Consortium released the observation images, spectra, and object catalogs that met technical standards first, as "DR1-Foundation." They decided to postpone the higher-level analysis results central to cosmology to the "full DR1," scheduled for June 16, 2027. Of the 19,588 standard VIS exposures planned for the Wide Survey, 17,143 met the quality criteria and are included here, a yield of about 87%. The foundation data released now will not be redistributed in 2027; later releases will add higher-level analysis products that complement them.
| Product / analysis | Status in DR1-Foundation (Nov. 2026) | Scale and method | Scientific role and caveats |
|---|---|---|---|
| Wide and deep-field images and basic catalogs | Released | About 1.7 PB, over 1 billion objects, over 120,000 calibrated VIS/NISP frames | Foundation of the cosmic map. Includes provisional pipeline classifications rather than confirmed identifications |
| Grism spectra | Released | Over 160 million (over 170 million validated spectra) | Primary data for 3D distance determination via emission-line searches |
| Photometric redshift catalog | Released | 880 million objects in the Wide Survey | Statistical distance estimates from multicolor photometry. Differs from direct measurement of individual emission lines |
| Strong lens candidate catalog | Released | 15,072 high-confidence (Grade A: 6,073; Grade B: 8,999) | Combines machine learning and citizen science. Candidate stage; spectroscopic confirmation not yet complete |
| Quasar candidate catalog | Released | Over 60 million (photometrically selected) | Parent population for early-universe studies. Initial sample including stellar and faint-galaxy contamination |
| Weak lensing (shear analysis) | Not yet released (postponed) | Included in the full DR1 due June 16, 2027 | Shape measurements after PSF model correction. Essential for directly measuring dark matter distribution |
| Cosmological correlation functions and parameter analysis | Not yet released (postponed) | Level 3 products to be delivered with the full DR1 | Joint analysis of galaxy clustering and shear. Essential for constraining dark energy |
As the table shows, the tools needed for Euclid's true goals, clarifying dark energy and testing general relativity on cosmological scales, will not be complete until the full DR1 in 2027.
The pieces for detecting the invisible mass that distorts the universe and tracing 10 billion years of cosmic expansion history are now laid out on the astronomers' board. Whether the picture of the universe assembled from them fits within the existing cosmological model () or presents signs of new physics will depend on the correlation-function values due in 2027.