A PhD in artificial intelligence is the highest-level AI qualification you can earn. Recognized worldwide, an AI PhD will position you to apply for top technical and research roles at frontier AI labs, Big Tech, and educational institutions.
An AI PhD gives you the chance to dive deep into a specific area of interest, such as autonomous systems, AI cybersecurity, or computer vision. You’ll complete coursework and then develop a thesis proposal in collaboration with an advisor, before researching your findings and presenting them in front of a panel of AI experts.
In the best programs, you’ll have access to massive compute power to complete your research, working under the supervision of genuine industry-leading AI researchers. Top schools also have proven graduate outcomes and reliable access to funding in the form of tuition waivers and monthly stipends.
In this guide, we’ll take you through a list of the top AI PhD programs in the US. We’ll also outline what a typical program consists of, the career paths that will be open to you once you graduate, and how to decide which schools best suit your personal circumstances.
What is an AI PhD?
In this guide, an “AI PhD” refers to any doctoral degree where you can meaningfully specialize in artificial intelligence.
There are dedicated AI PhD programs in the US: Carnegie Mellon, Georgia Tech, and Arizona State are all notable examples where you officially graduate as a Doctor of Philosophy in AI or Machine Learning. However, many Computer Science doctoral programs enable you to conduct similarly extensive AI research.
A PhD is a research degree: You are required to specialize in, and contribute to, a particular subfield; in practical terms, your eventual dissertation won’t look any different regardless of whether your diploma says AI or Computer Science.
AI vs Computer Science PhD programs: The differences largely come down to the coursework you do. Typically completed before you start your research in earnest, CS PhD coursework usually has breadth requirements spanning a range of computing topics. By contrast, the breadth courses you take in a dedicated AI PhD will still be AI-related.
Here’s a side-by-side of the coursework breadth requirements in Georgia Tech’s Machine Learning PhD and Stanford’s Computer Science PhD:
| Coursework | Machine Learning PhD (Georgia Tech) | Computer Science PhD (Stanford) |
|---|---|---|
| Breadth area 1 | Mathematical Foundations | Formal Foundations |
| Breadth area 2 | Probabilistic and Statistical Methods in Machine Learning | Learning and Modeling |
| Breadth area 3 | ML Theory and Methods | Systems |
| Breadth area 4 | Optimization | People and Society |
In other Computer Science programs, like the one at MIT, AI counts as one of your breadth areas. You can take 1 or 2 AI courses to satisfy the breadth requirement, but you must also cover ground like Theoretical Computer Science or Systems in CS.
Do I need a Master’s degree to take an AI PhD?
Although a PhD is considered the highest level of graduate study, you don’t typically need to earn a Master’s degree before you can enroll in an AI PhD. Admissions are very competitive, but schools are usually more interested in portfolios that showcase research potential.
In fact, some AI PhD programs (e.g. CS at Princeton) incorporate “incidental” Master’s degrees, where you’ll earn an MS automatically by completing the PhD coursework stage and subsequent examination. In other cases, you can petition for a Master’s by completing either some extra courses (e.g. at the University of Texas at Austin) or an additional thesis (e.g. at Rice).
AI PhD Program Structure
The best way to understand AI PhDs is in a three-part structure:
- Coursework
- Thesis proposal
- Dissertation and defense
Outside of these 3 milestones, you’ll also typically need to complete a teaching requirement at some point in your AI PhD program. The timing varies, but you’ll generally have to spend 2 semesters as a Teaching Assistant.
Coursework
The coursework aspect of your AI PhD will likely be the most familiar: it’s closest in format to what you’ll be used to from Bachelor’s/Master’s level. You’ll take a series of core and elective courses designed to deepen your knowledge of AI and computer science.
For the core courses, you’ll generally choose from a “menu.” Most programs have multiple core areas, and you have to pick one course from some or all of these areas in order to fulfill a breadth requirement.
You’re then free to pick your electives from any area. In a Computer Science program (as opposed to a dedicated AI PhD), this is where you’ll start to specialize in AI. Here’s an example from MIT:
The PhD structure is not always entirely linear, and some schools encourage you to take on a light course load so that you can start research straight away alongside your coursework. For example, UMass Amherst recommends taking no more than 1 course per semester. However, in most programs, you’ll have to complete all of the coursework requirements before you advance to PhD candidacy.
In programs with an integrated Master’s degree, successfully completing all coursework is sometimes the point where you earn your MS. That’s the case at Georgia Tech, for example.
Thesis proposal
Generally, your thesis proposal is what you need to complete to officially advance to PhD candidacy. If you haven’t chosen a member of faculty as an advisor already, you’ll do so at this point, working alongside them to research your proposal.
The proposal usually takes the form of a written document, followed by an oral examination in front of a panel of experts in the field of AI you’re researching. By conducting preliminary research into your dissertation topic and presenting it adequately, you prove your readiness to do PhD-level AI study.
Some programs choose to assess your PhD candidacy without directly tying it into your thesis proposal. At Georgia Tech and Cornell, you’ll do a literature review instead (known as the “A Exam” at Cornell); this is still designed to be useful background for your eventual dissertation, but your formal proposal is only required after you’ve advanced to candidate status.
Certain programs with embedded Master’s degrees will award the MS at this point, with the proposal effectively doubling as a Master’s thesis (e.g. University of Chicago). Other programs, like MIT, require you to complete a separate Master’s thesis.
Dissertation
The dissertation is the culminating project of your PhD. It’s an original piece of research that advances the field of AI study in some way.
Building on your proposal, you will carry out your research, typically focused on a highly specific area of AI. For example, here are some recent PhD dissertation titles from Princeton:
- Advancing Responsible AI with Human-Centered Evaluation
- Offline Processing for Machine Learning-Based Video Systems
- Generative Universal Models for Speech
- Language Models with Retrieval Augmentation and Conditional Computation
Once you’ve written your dissertation, you will need to defend it in front of a committee of AI specialists. You’ll give a presentation on your findings, and field questions from the committee: this is known as a defense.
Passing your dissertation defense is the final step to earning your PhD.
How long does an AI PhD take?
The vast majority of AI PhD programs expect you to enroll full-time. Even so, the time it takes to earn the degree varies significantly: anywhere between 4 and 7 years is within the usual range. Most schools place some kind of upper limit on the length of the degree (e.g. you must graduate within 8 years at San Diego).
For PhD programs that include an incidental Master’s degree, applicants who already have a relevant Master’s may be able to graduate more quickly. At the University of Illinois Urbana-Champaign, students who start with an MS have a maximum of 6 years to complete the program, 1 year fewer than students entering with only a Bachelor’s degree.
Top AI PhD Program 2026
You can find our full top 25 AI PhD ranking later on in this guide. But before we get there, it’s worth doing a deep dive into our #1 program: the PhD in Machine Learning at Carnegie Mellon University.
Carnegie Mellon: Anatomy of an Elite AI PhD Program
Carnegie Mellon is one of the few schools in the US that actually offers a PhD in AI/Machine Learning, as opposed to a Computer Science program where you specialize through your choice of courses and eventual thesis topic. A program name alone doesn’t make one AI PhD better than another. But in CMU’s case, the Machine Learning title is indicative of a highly specialized, elite AI program.
You’ll spend your first two years taking advanced AI-specific graduate courses. There are only two core courses: Advanced Introduction to Machine Learning, and Intermediate Statistics. Beyond the core curriculum, you’ll choose one course each from a “menu” of Theory, Methods, and Practical options. You’ll also pick one extra course from any section, plus any PhD-level statistics or School of Computer Science course.
Here’s an example of each type of course:
- Theory: Foundations of Autonomous Decision Making under Uncertainty
- Methods: Generative AI
- Practice: ML with Large Datasets
Starting in your second year, you’ll also need to complete two semesters as a teaching assistant in ML courses. You may choose to take additional advanced electives in your third year, but you’ll be expected to submit a formal thesis proposal by the end of the year.
You’ll then work on your thesis for the remainder of the PhD, which typically lasts 4-5 years in total. You actually need to choose an advisor within a month of enrolling, but you have the option to switch later on: pick an advisor who aligns with the research you want to conduct. CMU boasts world-leading core AI faculty, including Tianqi Chen, creator of the renowned machine learning library XGBoost.
Here are some recent thesis titles:
Carnegie Mellon doesn’t publish exact admissions figures, but describes the AI PhD program as very competitive. On the other hand, CMU doesn’t impose too many formal admissions requirements: you don’t need to submit GRE scores, and you don’t need to have majored in Computer Science at undergraduate level, although it’s recommended that you have a high level of comfort with math.
If you do manage to get on the program, you’ll benefit from substantial financial support. Carnegie Mellon covers full tuition and provides an additional stipend to its PhD students, continuing throughout the program as long as you maintain satisfactory academic progress. If you receive an outside fellowship, CMU will supplement it to bring it in line with the full level of support (or indeed higher).
CMU Machine Learning PhD graduates are also all but guaranteed excellent graduate outcomes. 100% of graduates who completed the First Destination survey within the last 5 years reported permanent employment.
Google is the top employer of graduates from the Carnegie Mellon AI PhD program. Only around 20% of graduates reported starting salaries, but among those that did, the average was over $170,000.
Here’s the full data for CMU Machine Learning PhD graduates from the last 5 years:
Google, Amazon, Anthropic, and Microsoft all hired graduates from the CMU Machine Learning PhD Class of 2025. Other graduates went to work for AI startups like GenBio AI, while still others went into elite academia at Harvard and Carnegie Mellon itself.
AI PhD Graduate Outcomes
An AI PhD qualifies you for some of the most senior AI roles at top companies and academic institutions.
For example, here’s an AI Engineer role at TikTok that requires you either to hold a PhD or to be working toward one. It’s a critical role building the infrastructure for the app’s For You recommendation system:
A lot of common AI PhD career paths are research-focused. These were some of the most common job titles from the last 5 years among AI PhD programs that disclose graduate outcomes data:
- Research Scientist
- Applied Scientist
- Assistant Professor
- Postdoctoral Researcher
- Software Engineer
Research Scientist
Becoming a research scientist effectively allows you to continue working on cutting-edge AI projects, but with an emphasis dictated by the needs of your employer. You may well still continue to collaborate with academic institutions and publish academic papers, through the lens of advancing AI technology internally.
Here’s an example of a 2026 Research Scientist opening at Google:
Average salary: $140,910 (BLS)
Assistant Professor
Assistant professor is your most likely path into academia straight out of an AI PhD. It’s a pretty direct continuation of your degree: you’ll continue to carry out academic research for the university, while also teaching classes.
Assistant professors aren’t tenured, but they tend to be “tenure-track” positions, working toward a full professorship eventually. Here’s an example of a recent vacancy at the University of Colorado Boulder:
Average salary $146,475 (Payscale)
Postdoctoral Researcher
Taking up a role as a postdoctoral researcher is an alternative pathway into full-time academia. Postdoc positions are usually temporary, offering a chance to continue your research, and build up your academic profile through publications.
Typically, a postdoc is considered a “bridge” role to help you move onto assistant professorships and professorships in future. But some candidates may simply prefer postdoctoral research posts. They don’t usually come with teaching obligations, and they are often attached to prestigious AI research facilities.
Here’s an example of a rolling postdoctoral fellowship program at the Stanford AI Laboratory:
Average salary: $73,590 (Comparably)
Software Engineer
You’re not wedded to a research-focused role when you graduate with an AI PhD. The degree showcases a deep knowledge of AI tools and processes, so many employers will be keen to hire you as an engineer.
And at the cutting edge, the boundary between an engineer and a researcher may blur anyway. You’ll still be working to discover novel solutions to AI problems, but with the primary aim of improving a product or system.
Here’s an example of an AI software engineer role at Google requiring a PhD:
Average salary: $153,445 (Indeed)
Methodology
We’re about to break down the top 25 AI PhD programs in the US. But before we do, here’s a quick word on our methodology:
- We gave top weighting to research output and AI/ML reputation. Schools with a consistent reputation for putting out a high volume of regularly-cited AI research were given major credit.
- Funding and resources were also given a high weighting. Programs were rewarded for full tuition waivers, generous stipends, and support that continues through the summer. Schools with significant AI research spending and compute resources also scored well.
- We gave graduate outcomes the next-highest importance: first destinations, starting salaries etc.
- Schools also received credit for faculty size and breadth. Active research in multiple AI subfields is a good sign that you’ll be able to pursue your specific area of interest on your PhD.
- Finally, a small amount of credit was given for ease of admissions. All of the top AI PhD programs are highly competitive, so this had a limited effect in practice.
Top 25 AI PhD Programs
1
Carnegie Mellon University
Pittsburgh, Pennsylvania
Overview
- PhD in Machine Learning
- Best for: Dedicated ML program, graduate outcomes
About the program
Carnegie Mellon is the only overall top-20 national school with a dedicated AI PhD program. Compared with Computer Science programs, the biggest difference comes during the first stage of the degree: instead of a series of graduate-level breadth courses spanning the whole CS curriculum, you’ll be doing targeted AI study from day 1.
Once you move onto thesis work, you’ll benefit from the guidance of some of the most respected AI faculty in the country. Times Higher Education ranks CMU 4th in the world for its Computer Science research quality, giving it a score of 99.8/100.
As well as fully funding tuition for the PhD program, Carnegie Mellon also provides all students with a stipend throughout the course of their studies. And graduate outcomes data speaks for itself: the most recent class of graduates went to the likes of Google, Amazon, and Anthropic, or to research roles within CMU or at other elite schools.
2
Stanford University
Stanford, California
Overview
- PhD in Computer Science
- Best for: Research focus
About the program
Stanford is a great choice if you’re pursuing a PhD with a view to a permanent career in academia. The Computer Science program is highly research-focused, with very few specific course requirements. Instead, you can shape the program around your interests, working with your advisor to map out an appropriate course of study. As such, you can maintain a strong AI focus.
Research is at the heart of the Stanford Computer Science PhD program from the beginning. You’ll need to take at least 3 research units every quarter. In the first year, you’ll do 3 different research rotations, helping you to finalize your permanent advisor by the middle of Spring term. Assuming you’re focusing on AI, then you may have the chance to work with faculty at the world-renowned Stanford AI Laboratory (SAIL). Faculty research areas include Human-Centered AI, Robotics, Reinforcement Learning, and Computer Vision.
Accepted applicants receive full funding throughout the PhD program, provided they keep up a satisfactory academic level. Unlike some other programs, funding is guaranteed throughout the year, including the summer semester. An annual stipend of $52,560 is one of the highest in the country, although it drops down the rankings once you factor in Silicon Valley living costs.
3
MIT
Cambridge, Massachusetts
Overview
- PhD in Electrical Engineering and Computer Science
- Best for: Elite AI research lab
About the program
The Department of Electrical Engineering and Computer Science (EECS) at MIT has a large doctoral program, with around 700 graduate students. Interestingly, there is no terminal EECS Master’s degree. Instead, you’ll earn a Master’s degree as part of the PhD program. By the end of your first year, you’ll need to have submitted a Master’s thesis proposal, which must then be completed in your second year.
Before you can move onto your PhD proposal, you’ll need to complete a Technical Qualification Evaluation (TQE) and Research Qualifying Examination (RQE). The TQE is effectively the core curriculum: you’ll have to take 4 courses, 2 of which can come from the AI group. Course options include Deep Learning, Shape Analysis, and Robotic Manipulation. The RQE is a research project, graded on a pass/fail basis.
Although you can’t apply directly to the Computer Science and Artificial Intelligence Laboratory, the EECS PhD gives you the chance to conduct your research at CSAIL. Recent focus areas have included fair and reliable machine learning systems, privacy in AI systems, and cognitive robotics. CSAIL is the largest lab at MIT, and it has an annual research budget of $65 million.
Full tuition plus a stipend is guaranteed for all first year students, and continues throughout the PhD program as long as you’re making adequate progress. Funding may take the form of a Research Assistant post or a fellowship/departmental grant. After the first year, you may also be funded via a Teaching Assistantship.
Be aware that MIT also offers a Computational Science and Engineering (CSE) PhD. Confusingly, this is not the same thing as an Electrical Engineering and Computer Science PhD. The CSE program is also a viable option for doctoral AI study, but it takes a more interdisciplinary approach, looking at how cutting-edge computing can be applied in various STEM fields.
4
UC Berkeley
Berkeley, California
Overview
- PhD in Computer Science (AI major)
- Best for: Well-paid RA/TA posts, strong AI focus
About the program
With a PhD in Computer Science at Berkeley, you’ll get the opportunity to work at BAIR, the Berkeley Artificial Intelligence Research Lab. More than 300 graduate students and postdoctoral researchers are based at the lab, working to advance fields like multi-modal deep learning and responsible AI. Current PhD topics include Data-Driven Control and Multi-Agent Systems, Human-Robot Collaboration and Safe Motion Planning Algorithms, and Machine Learning, Drug Discovery and Protein Design.
As well as your research, you’ll also take 24 credits of coursework (roughly 8 courses): 12 of those credits will form your major in AI. Courses in the AI major include Deep Reinforcement Learning, Natural Language Processing, and Advanced Robotics. John Schulman, co-founder of OpenAI and leader of the team that created ChatGPT, is among a long list of notable Berkeley CS PhD alumni.
Berkeley has a stated goal of providing all doctoral students with the financial stability to do their best work. Your tuition and fees are guaranteed to be covered throughout the PhD, and remuneration for academic employment or research is among the most generous in the country.
5
University of Washington
Seattle, Washington
Overview
- PhD in Computer Science and Engineering
- Best for: Integrated Master’s degree
About the program
The University of Washington PhD in Computer Science and Engineering offers a strong blend of coursework and research, with an integrated Master’s degree. You’ll take 5 breadth courses as part of your Qualifying Evaluation (Quals): ML/AI is 1 of 4 breadth groups, and you’ll need to choose courses from at least 3 groups. Assuming you pass Quals at a Master’s or PhD level, you’ll earn a Master’s degree at this stage in the process.
After you’ve got your Master’s, you’ll move onto research-focused PhD work. The General Exam consists of research for a detailed thesis proposal, which you’ll then present in front of a doctoral committee. You’ll typically then take at least a year to complete your dissertation and thesis defense. Recent AI-focused dissertations have covered AI-assisted video data management, reinforcement learning, and approaches to curating and scaling LLM training data.
The UW program is extremely competitive: the School of Computer Science and Engineering accepts 50-60 new PhD students each year from approximately 3,500 applicants. If you are accepted, funding is guaranteed for 3 years, provided you make satisfactory progress. Funding typically continues beyond 3 years to completion of the PhD. You’ll generally earn funding through either a Research Assistantship or a Teaching Assistantship.
6
Georgia Institute of Technology
Atlanta, Georgia
Overview
- PhD in Machine Learning
- Best for: Immediate AI/ML focus (no generic CS coursework)
About the program
As a dedicated Machine Learning PhD, the program at Georgia Tech does a good job of maintaining an AI focus from the outset. The core curriculum is split into 4 areas: Mathematical Foundations of Machine Learning, Probabilistic and Statistical Methods in Machine Learning, Optimization, and Machine Learning: Theory and Methods. You’ll take 1 course from each of these areas, and then 5 electives from a much wider list of ML breadth courses.
Once you’ve completed the coursework section of the PhD, you’ll move onto the qualifying examination, a 1-semester literature review and oral examination. Assuming you pass, you’ll move onto a research proposal and dissertation. At this stage, you’ll also complete a 2-course doctoral minor entirely outside of your field of research.
Interestingly, you can apply to the Machine Learning PhD at Georgia Tech through 1 of 9 different “home schools.” The courses on the PhD are drawn from various different departments, and your application will start at the school of your choice, e.g. Mathematics or Computer Science. It’s important to note that your application goes through 2 tiers of approval: it’s technically possible that you could be accepted by the home school but not by the ML faculty, which would result in admission to the general PhD program (e.g. Mathematics, if that was the school through which you applied).
When it comes to funding, 80% of Georgia Tech doctoral students are employed as TAs or RAs, which comes with a tuition waiver. You can also apply for merit-based fellowships covering partial or full tuition and fees, often with additional stipends.
7
Cornell University
Ithaca/New York, New York
Overview
- PhD in Computer Science
- Best for: Structured coursework, breadth of study
About the program
Cornell has one of the more structured Computer Science PhD programs among top schools, designed to produce researchers who have developed genuine breadth and depth of knowledge. You’ll first have to fulfill a competency requirement equivalent to a high undergraduate level in 4 areas: AI, Programming Languages, Systems, and Theory.
After that, you’ll take graduate-level courses in at least 3 of 5 breadth areas (again including AI). Unlike breadth requirements at other schools, you’ll also need to pick courses that span 3 different research styles: theoretical (mathematical proofs), systems (improving performance/reliability), and applied (using computers to solve interdisciplinary problems).
After your first year, you’ll pick a depth area and form a faculty research committee that will ultimately take you through to your PhD dissertation. You’ll have the chance to move from Ithaca to New York at this point, if you choose to work with faculty based at Cornell Tech.
Cornell is one of the schools that structures its PhD as a joint Master’s/Doctorate, so you’ll technically earn a Master’s after you complete your “A Exam.” The A Exam is a literature review in your chosen area of research, testing your preparedness for the eventual B Exam: a defense of your final dissertation.
CS PhD students at Cornell are guaranteed full funding throughout their degree, as long as they maintain good academic standing. You’ll receive funding through a combination of teaching assistantships, research assistantships, and fellowships.
8
University of Illinois Urbana-Champaign
Urbana, Illinois
Overview
- PhD in Computer Science
- Best for: Accelerated route for Master’s degree holders
About the program
The Computer Science PhD at the University of Illinois Urbana-Champaign follows a broadly standard formula, but with an accelerated route if you already hold a relevant Master’s degree. If you’ve completed an approved MS in Computer Science, you’ll only need to complete 64 credit hours instead of the full 96, skipping some of the 500-level CS coursework.
UIUC organizes its computing research around 11 broad areas, one of which is artificial intelligence. The department has developed a strong reputation in computer vision, machine listening, and natural language processing, among other things, earning multiple awards for its research. 75 faculty members have published AI papers in the last 5 years, the second-highest rate in the country.
As a PhD candidate at Urbana-Champaign, you’ll get the chance to contribute to the impressive AI research. Specialization begins early: you’ll take your 4th-semester qualifying exam specifically in AI, if that is your chosen research area.
CS PhD students at Illinois are guaranteed a funded TA/RA appointment for the first 5 years of their enrollment, provided they remain in good academic standing. These appointments include a full tuition waiver, partial fee waiver, and a stipend. However, be aware that the average time to complete the PhD is 6 years, which may mean a year or more where funding is not guaranteed.
9
Princeton University
Princeton, New Jersey
Overview
- PhD in Computer Science
- Best for: Elite faculty and facilities
About the program
The Computer Science PhD at Princeton begins with the breadth requirement: a total of 6 graduate-level courses, with at least one from each of the Artificial Intelligence, Systems, and Theory areas of study. Courses in the AI area include Advanced Computer Vision, Neural Rendering, and Foundations of Deep Learning.
After you’ve fulfilled the breadth requirement, you’ll move onto the General Exam, where you’ll give a research seminar on the area where you have begun to specialize. Once you pass, you’ll qualify for an incidental Master’s degree. In order to get the PhD, you’ll need to complete preliminary and full Final Public Oral exams, where you will present your proposal and your dissertation respectively.
You’ll do your research alongside elite AI faculty. The Associate Chair of the Computer Science department is Ryan Adams, a former Google Brain employee who specializes in AI/ML: he is currently directly advising 12 PhD students. All in all, hundreds of Princeton faculty are involved in AI research. The AI Lab has 300 Nvidia H100 GPUs, so you’ll have ample compute to carry out research at scale.
Princeton fully covers tuition for the standard 5-year term of the CS PhD program, and also pays the highest stipend of any CS program on our list. If you haven’t finished your degree after 5 years, you may be eligible for Dissertation Completion Funding. First-year students all receive a fellowship, while subsequent funding typically comes from TA/RA posts. You’ll need to complete at least 2 terms of teaching to graduate.
10
University of Michigan
Ann Arbor, Michigan
Overview
- PhD in Computer Science and Engineering
- Best for: Internship possibility
About the program
The University of Michigan follows the usual 3-part PhD structure: breadth study, preliminary exam, thesis. However, unlike some other programs, you have the option to complete a first-year Spring/Summer internship, as long as it is directly relevant to the research you’ll do for your preliminary exam. The internship is an alternative to the standard route of independent research in collaboration with a faculty member.
The breadth study consists of courses from 3 out of 5 breadth areas: AI, Human-Centered Computing, Software, Hardware, and Theory. You’ll also take 2 electives, during which you can begin to specialize in AI. There are more than 20 AI-related courses available at graduate (500+) level. Once you move onto thesis preparation, you’ll be advised by 1 of 65+ faculty members currently working on AI research.
Michigan guarantees funding to all students for the anticipated duration of their studies: 5 years normally, or 4 years for students who already have an MS. If you don’t already have a Master’s degree, then you’ll earn it on the way to getting the PhD.
11
University of Texas at Austin
Austin, Texas
Overview
- PhD in Computer Science
- Best for: Early research and publication opportunities
About the program
The University of Texas at Austin PhD stands out for its “early research immersion” model. You’ll join a research pod in your focus area (e.g. AI) straight away, which comes with the possibility of co-authoring papers and presenting at conferences. The research-focused structure also means you’ll be extremely well-prepared for your eventual dissertation, and for a subsequent career in academia.
Alongside your research, you’ll take 3 breadth and 3 depth courses. Breadth courses must cover each of Theory, Systems, and Applications, while the depth courses are your chance to begin specializing in your AI research area. You can also choose to take 2 further Minor courses, in order to earn a Master’s degree along the way.
Most UT Austin PhD students receive financial support for at least 5 years, although it is technically dependent on both academic performance and the availability of funds. Beyond 5 years, you may continue to receive funding if you remain in good standing. The funding package covers tuition, health insurance, and a stipend.
12
Harvard University
Cambridge, Massachusetts
Overview
- PhD in Computer Science
- Best for: Overall institutional prestige, funding
About the program
Harvard is a top-5 global university. It goes without saying that a PhD in Computer Science from a leading Ivy League school will be incredibly well-respected by potential employers. As a result, admissions are fiercely competitive: across all School of Engineering and Applied Sciences PhD and Master’s programs, the acceptance rate is roughly 3.8%, and the average undergraduate GPA is 3.8 on a 4.0 scale.
When it comes to specific artificial intelligence credentials, Harvard lags a little way behind some of the other top US schools, hence its positioning on our list. 29 members of faculty have been published on AI topics in the past 5 years, compared to over 100 at Carnegie Mellon. However, you can certainly shape the program to have an AI/ML emphasis, particularly once you’ve completed your course requirements and begun working toward a dissertation. Taken in your second year, the Qualifying Exam actually needs to be more specialized than “AI”: you’ll already be working within an AI subfield, like computer vision or natural language processing.
And like any PhD program, your eventual dissertation research will be highly specialized. If you’ve found a faculty member who’s a good fit for you, there’s no doubt that you can produce some extremely high-level AI research at Harvard. Recent projects include AI for conservation and anti-poaching efforts, and machine learning for real-world sequential decision-making problems.
Funding is a real plus of the Harvard program. You’ll receive full funding until you graduate, covering tuition, fees, a generous stipend, and transportation and dental subsidies. You’ll usually have a full fellowship for the first year so you can focus on coursework, before taking up TA/RA posts in subsequent years.
13
New York University
New York, New York (+ option of Shanghai, China)
Overview
- PhD in Data Science
- Best for: Internship, chance to study abroad
About the program
Offering a PhD in Data Science as opposed to Computer Science or AI/ML, New York University is home to one of the more unique programs on our list. The crossover between AI and data science is significant: you’ll take Machine Learning as one of your required courses, plus electives in topics like Natural Language Processing, Large Language Models, and Deep Learning. You can also take Practical Training as an elective, providing the chance to complete an academically relevant internship.
Once you begin your research, you’ll be working alongside faculty who specialize in areas such as Responsible AI and Machine Learning and Perception. You’ll also have the option to work with faculty at NYU Shanghai, moving there full-time after you have completed your coursework.
All PhD students are guaranteed a full tuition waiver and a stipend for up to 5 years, in the Fall and Spring semesters. You’ll also get a one-time award for initial expenses. And unlike the majority of other PhD programs, funding typically comes entirely via fellowships, although you will still need to fulfill a teaching requirement in order to graduate.
14
Columbia University
New York, New York
Overview
- PhD in Computer Science
- Best for: Research focus
About the program
Columbia’s Computer Science PhD puts a major emphasis on research. Most students arrange their research advisor before enrollment; you’ll work with your advisor from day 1 of the program, effectively starting as a research intern and progressing toward junior colleague status. Even during the coursework stage of the program, you’ll devote at least 50% of your time to research.
Compared to the most specialized schools, Columbia offers a relatively modest pool of AI-dedicated faculty. However, you should be able to find a researcher who aligns with your area of interest, enabling you to carry out high-level research of your own. Current PhD students working on AI projects at Columbia have earned prestigious external fellowships for their research from the likes of Amazon and Google.
Most students receive full funding through research assistantships, consisting of tuition exemption and a monthly stipend. However, you are still strongly encouraged to apply for external fellowships: as well as supplementing your funding, the process of applying helps to hone your grant application skills. Columbia also subsidizes low-cost apartments near campus.
15
UC San Diego
San Diego, California
Overview
- PhD in Computer Science
- Best for: Strength of AI faculty
About the program
The PhD at UC San Diego really comes into its own once you begin your research, but the coursework does a good job of covering all the usual bases. You’ll select 4 courses of breadth, each drawn from a different general area of computer science. You’ll then take a further 5 depth electives (or 6, if you also want to gain an MS), which is where you’ll begin to specialize in AI.
When it comes to research, San Diego boasts an impressive number of AI experts within their faculty. You should be able to find a suitable advisor, who will support you from the first milestone of the Research Exam through to the final dissertation. An example of a PhD dissertation topic in 2026 is Efficient and Resilient Neural Networks for On-Chip Inference. The dissertation advisor was Ryan Kastner, author of 4 books and more than 200 technical articles.
Your tuition and fees will be covered during the PhD at UC San Diego, and you’ll receive a monthly stipend. Support has a hard final limit of 7 years, and you must finish the PhD within 8 years (although 4-6 years is typical).
16
University of Southern California
Los Angeles, California
Overview
- PhD in Computer Science
- Best for: Pathway to academia
About the program
The University of Southern California says that its Computer Science PhD is designed to prepare you for a career in research. But before you can write and defend your own paper for the Qualifying Examination, you have to complete the coursework requirements: this includes 5 electives, 4 of which can be in your proposed area of research (i.e. artificial intelligence).
When it comes to conducting research, 50 members of faculty are listed as part of the AI/ML research group. Past and present AI research projects at USC include efforts to combat human trafficking, protect endangered species, and improve the diagnosis and treatment of diseases such as Alzheimer’s.
Funding for the PhD is not technically guaranteed, although USC says it tries to fully fund all of its doctoral students. The funding covers tuition, health benefits, and a monthly stipend.
17
University of Maryland
College Park, Maryland (+ option of study in Germany)
Overview
- PhD in Computer Science
- Best for: Joint program with Max Planck Institute
About the program
You’ll begin to shape an AI focus at the coursework stage of the University of Maryland PhD program: artificial intelligence is one of 4 areas you’ll need to cover across a total of 6 graduate-level computer science courses. You’ll be able to take courses in Machine Learning, Multi-Agent Systems, Neural Computation, and more. After you complete the coursework element, you’ll move onto a dissertation proposal and defense, working with a member of faculty whose research best aligns with your AI interests.
Just under 40 faculty members are currently associated with AI research. You’ll also get the chance to enroll in a joint PhD program with a Max Planck Institute in Germany, where you’ll have one advisor from UMD and one from MPI. If you’re accepted into this program, you’ll spend at least a third of your time with each advisor. You’ll graduate with both the standard Maryland PhD and an MPI certificate of doctoral research.
PhD funding is guaranteed for 9.5 months, but you’ll need to find external assistantships or internships to secure funding through the summer (unless you’re on the MPI program, in which case you’ll spend summers in Germany). Stipends are also on the lower end, at $30,800-$33,100.
18
University of Pennsylvania
Philadelphia, Pennsylvania
Overview
- PhD in Computer and Information Science
- Best for: Graduate outcomes, stipend matching cost of living
About the program
The PhD in Computer and Information Science at the University of Pennsylvania stands out as one of the few programs that systematically publishes graduate outcomes. Over the past 5 years, 91% of CS PhD graduates were in full-time employment within 6 months of graduation, with a median annual salary of $160,000.
Such a high graduate success rate is driven not only by UPenn’s Ivy League status, but also by an impressive CS PhD program with a strong AI route. You can take one “Learning and Modeling” course for your breadth requirement, plus up to 4 for your depth requirement: options include Machine Perception, Biomedical Image Analysis, and Natural Language Processing.
You’ll do your research alongside well-respected AI faculty. Recent dissertation topics have included trustworthy machine learning and applying language models to patient health records. Your research will be fully funded, and the full-year stipend is 1 of just 4 nationwide where the amount you’ll receive surpasses the annual cost of living + fees, according to CS Stipend Rankings.
19
University of Massachusetts Amherst
Amherst, Massachusetts
Overview
- PhD in Computer Science
- Best for: Early specialization, curriculum personalization
About the program
UMass Amherst tells prospective students that they can “immerse themselves in machine learning and AI” on the Computer Science PhD program. You’re encouraged to specialize in your area of interest early: you’re advised to take a maximum of 1 core course per semester, so as not to delay your research. To enable this, you’re able to conditionally pass the Qualifying Exam before you’ve completed all of your required courses.
When you do work through the core curriculum, you’ll actually be working toward a joint MS/PhD, unless you already have a Master’s degree. There are significant opportunities for AI study. You’ll need to take at least 1 course to fulfill a specific AI requirement, and can choose 3 more to meet the general core requirement. You’ll also need to produce a Master’s project: it’s called a “synthesis,” and involves combining your core area of research with a secondary area (e.g. machine learning + systems security).
The candidacy section of the PhD follows the usual formula: forming a committee, making and defending your proposal, and finally writing and defending your dissertation. You’re guaranteed funding for the first 5 years of the program, although summer funding is only guaranteed in your first year.
20
UCLA
Los Angeles, California
Overview
- PhD in Computer Science
- Best for: Interdisciplinary study, slightly less competitive admissions versus other top schools
About the program
The PhD in Computer Science at UCLA is divided into 8 subdisciplines. You’ll pick one of them (AI) as your major field, in which you will take 4 courses. You’ll also do one course in each of two minor fields: options include computational systems biology, computer networks, and graphics and vision. So if you know from the outset that you want to conduct research that straddles AI and another subdiscipline, the program structure may suit you.
Many of the UCLA Computer Science faculty are conducting high-level AI research. One team of researchers (including the department chair) was recently awarded a $250,000 grant as part of the Moonshots Initiative, working on an AI system that can reason like a mathematician.
Admissions to the program are certainly competitive, but marginally less so than other top programs. The latest published acceptance rate is a little over 5%, with successful applicants averaging a 3.75 undergraduate GPA. If you’re accepted, funding continues throughout the program as long as you maintain satisfactory progress.
21
University of Wisconsin-Madison
Madison, Wisconsin
Overview
- PhD in Computer Sciences
- Best for: Breadth of study
About the program
If you’re considering a PhD in AI, then clearly you’re looking to specialize. But a breadth of study across the computer science spectrum can help to inform your eventual research, and the University of Wisconsin-Madison ensures that you cover a lot of bases in your coursework. Lots of PhD programs mandate 3 breadth courses, but Wisconsin asks for 4: 1 course each from 4 different bands. You’ll also need to complete either a doctoral minor or a graduate/professional certificate.
Otherwise, the UW-Madison PhD follows a fairly standard formula, with intensive research work alongside faculty once you’ve completed the coursework. The school is committed to AI research: it launched the College of Computing and Artificial Intelligence this summer, the first new college in over 40 years. The college houses more than 100 faculty, and the school’s overall research expenditure is in the top 5 nationally.
It’s important to be aware that funding is not fully guaranteed. 91% of PhD students on the program received full funding in Fall 2025, down 3% from the 5-year median. For those who do receive funding, the average monthly stipend is $3,337, which is on the lower side. Graduate outcomes are strong: over the last 5 full years of data, a quarter of program graduates have joined Google or Meta. 30% have gone into academia, the majority of whom have become tenure-track faculty.
22
Johns Hopkins University
Baltimore, Maryland
Overview
- PhD in Computer Science
- Best for: Funding, AI applications in healthcare
About the program
AI isn’t one of the named core areas in the PhD coursework at Johns Hopkins: the program is split into Applications, Reasoning, Software, Systems and Theory. However, there are AI-pertinent courses available in all of these areas, and you’ll also choose 4 free electives from CS or related departments.
Once you get going on your own research, there is ample opportunity to specialize in AI alongside expert faculty, especially if you’re interested in AI applications in healthcare. Johns Hopkins is widely considered one of the best medical schools in the world: examples of interdisciplinary projects include improving diagnostics, harnessing wearables data, and advancing precision medicine.
You’ll also benefit from one of the best PhD funding packages in the country while at JHU. You’ll get a stipend, full tuition coverage, medical and dental insurance, and paid health insurance premiums for any eligible dependents. Not only that, but CS Stipend Rankings places Johns Hopkins #1 nationwide for the most generous stipends, once the cost of living is factored in.
23
Arizona State University
Tempe, Arizona
Overview
- PhD in Artificial Intelligence
- Best for: AI focus
About the program
The Arizona State University PhD in AI is brand new: the first cohort joins in Fall 2026. However, ASU still warrants a place on this list given the school’s long-standing commitment to AI. The new program draws on the existing expertise of 60+ faculty members to offer one of the few dedicated AI doctorates in the country.
The AI focus really shines through in the coursework. As well as a core course in AI, you’ll pick your electives from 3 AI streams: statistical learning, embodied AI, and reasoning and decision-making. From there, you’ll move onto AI research alongside faculty.
As a new program, it’s impossible to judge what admissions will look like in practice. But the baseline requirements are lower than most similar programs: the minimum undergraduate/Master’s GPA is 3.0, lower than the 3.5 GPA that ASU requests for the CS PhD program. One major downside is that funding is not guaranteed, although the program aims to support all incoming students.
24
Purdue University
West Lafayette, Indiana
Overview
- PhD in Computer Science
- Best for: School-assisted summer internships
About the program
As of 2026, Purdue has joined many of the other PhD Computer Science programs on our list in adding AI as a core course requirement. You’ll need to take at least one of Artificial Intelligence, Data Mining, Statistical Machine Learning, and Foundations of Deep Learning. Alongside the new AI focus, you’ll also need to take at least one course from Theory and Systems categories.
The increased emphasis on AI is a natural step, given that Purdue has 40+ faculty members conducting AI/ML research. Once you’ve found a supervisor, you’ll sit a preliminary examination in your chosen research area, before going on to prepare and defend your dissertation.
Funding at Purdue is not guaranteed. Even if you do qualify, the initial stipend amount is quite low: the latest quoted figure is $23,389, or $31,390.50 if you’re approved for support through the summer as well. However, the stipend does increase as you pass milestones in the PhD. You are also free to find paid summer internships, which a Corporate Partners Program helps facilitate.
25
Rice University
Houston, Texas
Overview
- PhD in Computer Science
- Best for: Funding
About the program
Rice earns a place on this list in part because of its generous Computer Science PhD funding. All students are supported for up to 6 years, assuming they make satisfactory progress, and they receive an annual stipend of over $41,500. Payments continue through the summer if you’re working on research projects. Factoring in local living costs, the Rice stipend is top-10 nationally according to CS Stipend Rankings.
As for the PhD program itself, Rice provides a good route to an AI-focused doctoral degree. You’ll need to take at least 1 AI course to satisfy the breadth requirement, and you can take 3 more as electives. You’ll choose your electives in consultation with your research advisor, so they should help to shape your eventual area of research.
Once you’ve completed your coursework, you can petition for an MS degree at any time, which requires an additional thesis and defense. Otherwise, you’ll begin to work on your proposal and subsequent dissertation. Other schools have more AI faculty, but Rice boasts some well-respected names. For example, Professor Moshe Vardi has authored or co-authored more than 700 papers, earning over 67,000 citations.
Choosing the Right AI PhD
Perhaps more than any other AI degree, the “right” PhD program depends to a large extent on your personal priorities. You need to think about which environment is likely to best support your research goals.
Our ranking serves as a helpful starting point. But when you’re looking to finalize your shortlist, there are certain factors you should be researching in detail.
Advisors
When you’re looking at a Bachelor’s or Master’s degree in AI, you don’t generally need to go into too much detail on faculty. When it comes to your PhD, faculty is the single most important thing to consider.
It’s helpful to remember that the AI faculty won’t primarily be your teachers. They will be your colleagues and research partners; you will have the best experience if you find a great fit between your area of interest and those of your eventual advisor.
That’s why some of the most research-focused PhDs, e.g. Columbia, get you to connect with your advisor before you even begin the program. It’s a relationship that will shape your entire degree.
Of course, it’s unrealistic to become intimately familiar with the research of every AI faculty member across the hundreds of US schools with Computer Science PhDs. So you might want to save this granular level of research for when you’ve narrowed down your options to a few front-runners.
Alternatively, if you know you’re intending to conduct research in a very niche AI subfield, it may be better to start by building a list of relevant researchers. You can then go on to compare the PhD programs in which they serve as potential advisors.
Funding
At first glance, it can look like funding is broadly equivalent across all of the elite AI PhD programs. For instance, almost all of the top schools offer full tuition waivers and monthly stipends. But when you dig into the details, you start to find some important differences that could influence your final decision.
These are some of the questions you’ll want to ask yourself:
- Is funding 100% guaranteed for all PhD students?
- Does funding cover both tuition and fees?
- Is there a risk that funding expires after a certain number of years?
- How much is the monthly stipend?
- Does the stipend continue through summer?
- How does the stipend compare to the local cost of living?
- Do I get any additional benefits, e.g. health insurance?
CS Stipend Rankings is a valuable resource for answering some of these questions, in particular the stipend amounts, how they compare to cost of living, and whether they are awarded for 9 or 12 months. However, it’s always worth trying to get the most up-to-date information directly from the school in question.
There is also technically the option of self-funding a PhD. The self-funded model is mostly contained to untraditional, online AI PhD programs, e.g. at the University of the Cumberlands. You may exceptionally want to consider this route if you cannot commit to full-time study, or if you don’t have the necessary academic background for admission to a traditional PhD.
Facilities
Depending on your specific area of interest, you may require significant technical resources in order to make meaningful contributions to AI research. Most of the top schools have impressive AI labs, but it’s still worth checking that you’ll be equipped with the tools you need to complete your PhD work to a high level.
For instance, Princeton has invested in 300 Nvidia GPUs. Sanjeev Arora, a Professor in Computer Science at Princeton, put it in simple terms: “If you don’t have the compute, you cannot do things at scale and you can’t join the conversation.”