The QS World University Rankings by Subject 2026 lists 201 institutions in Data Science and Artificial Intelligence, drawn from 37 countries and territories. The Massachusetts Institute of Technology holds the top position. For prospective students choosing a programme by subject rather than by overall university brand, those two facts frame the field: it is broad enough to span 37 national systems, yet its upper tier is narrow and heavily concentrated.
This article works through what the 2026 edition shows about where the subject is taught, how deep each country’s presence runs, and what separates the institutions at the very top from the rest of the ranked field.
How large is the field, and how is it layered?
The ranked list covers 201 institutions in total. The distribution by rank band shows a steep pyramid. Ten institutions sit in ranks 1 to 10, another ten in ranks 11 to 20, ten more in ranks 21 to 30, and twenty in ranks 31 to 50. That accounts for the top 50. Beyond that point, QS does not publish exact ranks for every institution: 50 institutions are placed in the published rank band 51 to 100, and 101 are placed in the published rank band 101 to 200.
The consequence for anyone reading the list is that the top 100 is fully resolved — 100 institutions fall within it — while the remaining 101 carry only a range. An institution in the 101–200 band cannot be described as being inside the top 150, nor as being outside it, because its range straddles that line. QS treats these as a separate group, and readers comparing institutions in this zone should treat the range, not a single number, as the unit of comparison.
A median rank cannot be computed for this subject. The institution sitting at the middle position has only a published range, so there is no single midpoint figure to quote. Any summary that claims a typical rank for the field would be inventing precision the data does not carry.
Which countries and territories appear most often?
The country leaderboard is led by the United States with 41 ranked institutions, followed by the United Kingdom with 22. China (Mainland) and Australia each have 13. India and Spain each have 8, while Canada, Italy and Malaysia each have 7, and Hong Kong SAR, China has 6.
| Country/Region | Ranked Institutions |
|---|---|
| United States of America | 41 |
| United Kingdom | 22 |
| China (Mainland) | 13 |
| Australia | 13 |
| India | 8 |
| Spain | 8 |
| Canada | 7 |
| Italy | 7 |
| Malaysia | 7 |
| Hong Kong SAR, China | 6 |
Two patterns are worth drawing out. First, the gap between the top two and the rest is wide: the United States alone accounts for a share of the 201-institution list that no other single system approaches. Second, the field is not a two-country story. The ten systems shown here together account for a large majority of the list, but the remaining entries are spread across the full set of 37 countries and territories, meaning smaller national systems do hold ranked positions even when they contribute only a handful of institutions each.
For a student weighing where to study, this matters in a practical way. A country with many ranked institutions offers a deeper local peer group, more adjacent research capacity, and more alternative programmes if a first choice does not work out. A country with a small number of ranked entries can still contain a strong option, but the surrounding ecosystem is thinner.
What do the top-tier scores actually show?
QS publishes an overall score only for a leading subset of ranked institutions. In this subject, 50 institutions have a published overall score and 151 do not. A missing score does not mean an institution performed poorly or was removed from the ranking — it remains ranked, and it simply sits outside the group for which QS publishes a score. Every statement below about scores therefore covers only those 50 institutions.
Among them, the distribution is as follows. Eleven institutions score 90 or higher. Thirty-five score 80 or higher. All 50 score 70 or higher. Broken into bands rather than cumulative counts, 11 institutions fall in the 90–100 band, 24 in the 80–89 band, and 15 in the 70–79 band.
The top of that distribution is tight. Across the top 10 institutions, published overall scores run from 98.0 at the highest to 90.4 at the lowest. Widening the frame shows how quickly the floor drops: across the top 20, the lowest published score is 83.3; across the top 30, it is 80.7; and across the top 50, it is 76.0.
The top 100 range is the same as the top 50 range — highest 98.0, lowest 76.0 — because scores are published for 50 institutions within the top 100, and those are the same 50 institutions. There are no additional published scores between rank 51 and rank 100 to extend the range downward. Reading the top 100 row as though it described 100 scored institutions would misstate what the data covers.
| Subject Rank | Institution | Country/Region | Academic Reputation | Employer Reputation | Citations | Overall Score |
|---|---|---|---|---|---|---|
| 1 | Massachusetts Institute of Technology (MIT) | United States of America | 100 | 100 | 93.4 | 98 |
| 2 | Stanford University | United States of America | 96.3 | 98.3 | 96.2 | 96.4 |
| 3 | National University of Singapore (NUS) | Singapore | 99.6 | 95.8 | 93.1 | 96.2 |
| 4 | Nanyang Technological University, Singapore (NTU Singapore) | Singapore | 95.6 | 92.4 | 94 | 94 |
| 5 | Carnegie Mellon University | United States of America | 99.2 | 87.4 | 94.4 | 93.9 |
| 6 | University of California, Berkeley (UCB) | United States of America | 91.9 | 93.1 | 98.6 | 93.2 |
| 6 | University of Oxford | United Kingdom | 92 | 95.8 | 96.8 | 93.2 |
| 8 | Harvard University | United States of America | 89.9 | 99.3 | 91.5 | 92.8 |
| 9 | University of Cambridge | United Kingdom | 91.9 | 95.6 | 88.6 | 91.4 |
| 10 | Tsinghua University | China (Mainland) | 87.5 | 88.9 | 91.2 | 90.4 |
| 11 | ETH Zurich | Switzerland | 90.3 | 90.9 | 91.7 | 90.2 |
| 12 | Peking University | China (Mainland) | 86.1 | 88.3 | 91.7 | 89 |
| 13 | University of Toronto | Canada | 84.6 | 86.6 | 89.7 | 86 |
| 14 | University of California, Los Angeles (UCLA) | United States of America | 81.8 | 87.2 | 89 | 85 |
| 15 | EPFL – École polytechnique fédérale de Lausanne | Switzerland | 86.7 | 86.2 | 85.1 | 84.9 |
| 15 | Imperial College London | United Kingdom | 85.8 | 82 | 88.9 | 84.9 |
| 17 | Princeton University | United States of America | 80.4 | 86.7 | 93.1 | 84.2 |
| 18 | The University of Hong Kong | Hong Kong SAR, China | 80.9 | 83.4 | 92.8 | 83.7 |
| 19 | University of Washington | United States of America | 82.2 | 80.8 | 92.6 | 83.6 |
| 20 | Yale University | United States of America | 85.2 | 86.7 | 82.7 | 83.3 |
What separates the head of the list from the rest?
The score compression at the top is the clearest signal in the dataset. The top 10 institutions span 98.0 down to 90.4 — a narrow window at the very top of the scale. By the time the frame reaches the top 50, the floor has fallen to 76.0. In other words, the difference between the best and worst published score among the top 10 is small compared with the difference between the top of the list and the bottom of the scored group.
That has a direct implication for how much weight a small rank difference deserves. Within the top 10, adjacent institutions are separated by margins that are difficult to interpret as meaningful differences in strength. Further down, where published scores are spread more widely, the gaps are larger — but they also cover only the 50 institutions for which a score exists, so the picture below that group is described by rank band rather than by score.
The other structural feature is geographic depth at the top. The United States holds the largest number of ranked institutions at 41, and the top-ranked institution is MIT. The United Kingdom follows with 22. China (Mainland) and Australia each contribute 13, and Hong Kong SAR, China contributes 6. These figures describe how many institutions each system places on the list, not how high any individual institution ranks; the two questions are separate, and the data here supports only the first directly.
What this means for choosing a programme
Three practical readings follow from the 2026 edition.
The list is deep but its scored tier is shallow. Of 201 ranked institutions, 50 carry a published overall score. If score comparison is central to a decision, the comparable set is that group, and choices below it have to be made on other grounds — research fit, curriculum, location, cost — rather than on a score that QS does not publish.
Rank bands, not exact ranks, govern the lower half. The 101 institutions in the published 101–200 band cannot be ordered against one another from this data. Treating a band position as a precise standing overstates what the ranking says.
Country counts measure breadth, not peak. A system with many ranked institutions offers more options; it does not by itself establish where the strongest single programme sits. The two questions need to be asked separately.
Data note
The figures in this article come from the QS World University Rankings by Subject 2026, published by Quacquarelli Symonds (QS), with a data reference date of 2026-04-03. The edition covers 60 subject tables and includes indicator-level scores for academic reputation, employer reputation, citations, H index and international research network.
The scope here is the complete ranked list for Data Science and Artificial Intelligence, covering 201 institutions across 37 countries and territories. Country-level counts reflect the number of ranked institutions attributed to each system.
Two boundaries deserve emphasis. First, QS publishes an overall score only for a leading subset of ranked institutions — 50 in this subject. The remaining 151 are ranked but unscored, and their absence from the score distribution carries no implication about their performance. All score bands and ranges in this article describe only the scored group. Second, QS publishes a rank range rather than an exact rank for some institutions. An institution counts as within the top N only when the upper bound of its range is at most N. Where a range straddles a threshold, the institution is counted neither inside nor outside that threshold, and is reported separately. Rank ranges are not used here for medians, averages or year-on-year movement, and no median rank is stated for this subject because the institution at the middle position has only a published range.