If your talent strategy assumes that AI hiring will be solved by the standard CS pipeline expanding to meet demand, the 2024-25 data shows the pipeline is reshaping in ways that contradict the assumption. Computer science enrollment at US four-year universities fell 11% between 2024 and 2025. In the same period, master's graduates in AI software-related fields grew 17%. The aggregate pipeline is not shrinking; it is bifurcating, and the part that matters for AI hiring is the part that is growing.

AI postsecondary graduates in the US by degree level, 2014–24:

The structural data separates AI-relevant majors into two categories. AI software-related majors include Artificial Intelligence, Computer Programming, and Computational and Applied Mathematics. AI hardware-related majors include Electrical and Electronics Engineering, Condensed Matter and Materials Physics, and Industrial Engineering. The bifurcation is visible across the categories.

AI software-related degrees: associate's 23,440 (2024) up from a near-zero baseline in 2014. Bachelor's 120,950, up from roughly 19,000. Master's 94,920, up from roughly 6,000. PhD 6,030. Master's degrees specifically grew 82% between 2022 and 2024 and 17% in the most recent year alone. AI hardware-related degrees show a different trajectory: bachelor's degrees peaked in 2020 and have declined 13% since.

The CS enrollment decline that gets the headlines is real. It is also incomplete as a description of what is happening in the pipeline. The 11% drop is in CS as a specific major. The broader category of Computer and Information Sciences and Support Services did not fall by 11%. AI-related majors within that broader category, those identified by the January 2025 White House AI Talent Report, grew. The undergraduate is choosing differently within an expanding catalogue of options, not abandoning the field.

Why does the framing matter for talent strategy? Three structural implications follow.

The first: the recruiting funnel from "CS graduate" to "AI hire" is not the right funnel for 2026 talent planning. The undergraduate who would have graduated with a CS degree five years ago may now graduate with an AI-specific major, an Applied Mathematics degree, a Data Science programme, or a hybrid CS-domain degree (CS-biology, CS-economics, etc.). Talent strategy that filters on the CS-major credential will under-source qualified candidates from the new pipeline shape. The filter needs to widen to include the proliferating AI-adjacent majors.

The second: the master's-level bulge is the part of the pipeline most directly tied to advanced AI roles, and it is growing fastest. 17% year-over-year growth in AI software-related master's degrees is steeper than any other degree-level trajectory in the data. For organisations hiring research-track or applied-ML roles, the master's cohort is now the most numerous and the most directly trained population. But, and this matters, the majority of those master's graduates are non-residents. The talent supply at this layer of the pipeline depends heavily on international student inflows that are now under regulatory pressure from US visa policy changes. The pipeline shape is therefore sensitive to immigration policy in ways the prior CS-undergraduate pipeline was not.

The third: the slow-down in CS undergraduate enrollment is partly a response to perceived labour market signals. The early-career employment cliff in software development is documented in the same data: software developer employment in the 22-25 cohort fell roughly 20% from its 2022 peak. Students are responding to a job market that has shifted. The implication for talent strategy: the cohort entering AI roles in 2026-2028 will be smaller at the undergraduate-CS level than the cohort that entered in 2020-2022. The gap is being partly filled by AI-specific majors with different curriculum content. Hiring frameworks calibrated on CS-undergraduate fundamentals will need to adjust to a population with different training emphases.

The trajectory: the bifurcation will continue through 2026-2027. AI-specific majors are being added at additional universities each year. The Classification of Instructional Programs (CIP) code for "Artificial Intelligence and Robotics" (11.0102) has existed since 2016 but has been used by very few schools. As more institutions adopt explicit AI majors, the AI-specific graduate counts will grow further and the CS-major counts will shrink further. The aggregate is approximately flat or modestly growing; the composition is changing.

What does this mean for talent strategy planning in 2026? Three implications.

The first: recruiting frameworks should be evaluated on the breadth of source majors they accept. Frameworks that filter strictly on "Computer Science" exclude the population of AI-specific graduates that is now the fastest-growing cohort. The minimum source-major list should explicitly include AI software-related majors per the AI Talent Report's classification.

The second: the immigration-policy dependency on the master's-level pipeline is now a planning variable. If US federal policy continues to restrict international student inflows, the AI master's pipeline will narrow within 24-36 months. Organisations that rely heavily on this pipeline should be planning for either substitution (US-domestic master's expansion) or geographic diversification of hiring (Canadian, UK, European, Asian master's programmes). The 67% non-resident share is a single point of failure for the current trajectory.

The third: the curriculum content of AI-specific majors is not standardised. Universities adopting AI majors are doing so faster than the curricular consensus is forming. The 2026-2027 hires from these programmes will have varied training depending on the institution. Hiring frameworks should be calibrated for this variance: capability assessment at hire becomes more important than credential filtering when the credential ↔ skill mapping is not yet stable.

For workforce strategy teams setting AI talent direction, the planning anchor needs to shift from "CS pipeline" to "AI talent pipeline," explicitly including AI software-related majors, AI hardware-related majors, hybrid CS-domain programmes, and the master's-level cohort. The 11% CS drop and 17% AI master's growth are both real; treating them as competing narratives misses the structural pattern. The pipeline is reshaping. Talent strategy that reshapes with it will out-perform talent strategy that anchors on the 2018-2022 pipeline structure.

Sources

  • Primary: Stanford AI Index 2026, Chapter 7 (Education) 7.2 — hai.stanford.edu/ai-index/2026
  • US degree-graduate data: National Center for Education Statistics IPEDS, 2024 — graduate counts by major and degree level
  • AI-relevant majors classification: January 2025 White House AI Talent Report — AI software-related and AI hardware-related major taxonomy
  • CIP code reference: Classification of Instructional Programs (CIP) code 11.0102 — Artificial Intelligence and Robotics