If your global AI talent strategy is anchored on the United States retaining its lead in producing ICT graduates, the 2022-23 OECD data shows the lead is being eroded faster than the visible numbers suggest, and the data itself excludes the countries that would change the comparison most dramatically. The US produced 122,814 ICT bachelor's graduates in 2023, more than any other country in the OECD sample. The growth rate, however, was modest (4.6% year-over-year). Turkey grew its bachelor's output 30% in the same period. Brazil grew 30%. At the PhD level, Mexico grew 76%. And the OECD dataset does not include India, China, or most of Africa.

ICT bachelor's graduates by country, 2022–23:

The headline framing, "US leads in ICT graduates," is technically correct and strategically misleading. Three structural data points complicate the framing.

The first: the growth rate gap. Across the four degree levels (short-cycle, bachelor's, master's, PhD), the US grew its graduate output more slowly than several non-US producers in 2022-23. At the bachelor's level: US +4.6%, Turkey +30%, Brazil +30%. At the short-cycle level: Turkey +27%. At the PhD level: Mexico +76% (off a smaller base, 254 graduates, but the growth rate matters for trajectory). The only level where the US led on growth was master's, where the US grew 55%, and per the IPEDS data discussed elsewhere, much of that master's growth is non-resident graduates whose presence depends on US visa policy.

ICT master's graduates by country, 2022–23:

The second: the dataset omission. The OECD sample covers OECD member countries plus several non-OECD partners. It does not include India, China, or most of Africa. The omission is consequential because these countries have made significant investments in computing education and make up a significant proportion of the global population. India and China are the most-populous countries in the world and are widely understood to produce very large cohorts of ICT graduates. Their absence from the OECD-based comparison means the visible "US leads" framing reflects only the countries the OECD tracks, not the global producer landscape. The US lead within the OECD sample is real. The US lead globally is unmeasured and likely substantially smaller.

The third: the reporting lag. The OECD data shows 2022-23 graduate cohorts. Those students enrolled in 2018-2020 in most cases (bachelor's takes ~4 years, master's takes ~2 years, PhD takes ~5+ years). The data is therefore a snapshot of pipeline decisions made 4-7 years ago. The current pipeline, students enrolling in 2024-25, will appear in the OECD data in 2028-2030. The lead position in the 2022-23 data does not necessarily reflect the relative pipeline position today.

ICT PhD graduates by country, 2022–23:

Three structural implications follow for global AI talent strategy.

The first: the planning anchor "the US has a structural lead in technical talent" needs revision. The OECD data supports "the US currently produces the largest number of ICT graduates in the OECD sample," a narrower and more defensible claim. The broader strategic narrative that follows from the narrower claim is different: the US position is not a structural moat but a measurable advantage in a specific cohort definition. The advantage can erode through faster growth elsewhere even if the absolute US output remains constant.

The second: the geographic diversification opportunity is larger than the visible OECD data suggests. If India and China are producing comparable or larger ICT graduate cohorts (which the data does not directly show but which population scale and reported investment make likely), then global talent strategy that includes Indian and Chinese hiring channels accesses a supply that the OECD-anchored planning view ignores. Organisations that have built recruiting infrastructure in India and China (or via Indian and Chinese diaspora networks) have access to talent pools that don't appear on the standard "OECD ICT graduate" comparison.

The third: the high-growth-rate countries deserve specific attention. Turkey, Brazil, and Mexico are producing ICT graduates at growth rates that exceed the US. The 5-year forward projection: if the growth rates persist, the absolute US-Turkey gap at the bachelor's level closes from ~115,000 (2023) to a much smaller number by 2028. The growth rate trajectory is the leading indicator; the absolute output is the lagging indicator. Strategic plans built on the lagging indicator will be reading a 5-year-stale signal.

The Mexico PhD growth (76% YoY) is worth pausing on. The absolute numbers are small: 254 PhDs in 2023 vs the US 2,874. A 76% growth rate compounded for five years would put Mexico at roughly 3,800 PhDs, exceeding the current US output. Whether the growth rate persists at that level is genuinely uncertain, as small-base growth rates are highly volatile. But the structural signal, a country specifically investing in PhD-level ICT capacity, is now visible in the data and should be treated as a forward-looking variable rather than dismissed as noise.

What does this mean for global AI talent strategy planning in 2026?

The first move: rebuild the planning view to include the omitted countries. The OECD-only view is convenient because the data is standardised. It is also misleading. Strategic plans should explicitly model India and China as significant talent supply countries even where the formal data is not directly comparable. Proxy data such as STEM graduate counts, AI publication counts, and engineering school output can substitute where direct ICT graduate counts are not available.

The second move: prioritise the high-growth countries for recruiting infrastructure development. Building recruiting channels in Turkey, Brazil, Mexico, India, and China is high-cost in the near term but accesses pipelines whose growth trajectories are steeper than the US pipeline. The forward-looking talent supply favours these channels.

The third move: build planning sensitivity around the growth-rate variable rather than the absolute-position variable. The 5-year forward state of the global ICT graduate landscape depends primarily on growth rates over the next 36-60 months. Strategic plans anchored on the 2023 absolute position will be reading the trailing indicator. Plans that monitor and incorporate the growth-rate signal will be reading the leading indicator.

The trajectory: the global ICT graduate landscape is becoming more multi-polar over the 2023-2030 window. The US will likely retain a large absolute output. The relative US share will decline as faster-growing producers expand. The OECD-tracked countries will produce a measurable lead-erosion signal; the omitted countries (India, China, large parts of Africa) will produce an unmeasured lead-erosion signal that is likely larger. Global talent strategy that builds for multi-polar supply will be aligned with the trajectory the data describes. Strategy that assumes a continued single-pole US lead is reading the headline and missing the structural shift underneath.

Sources

  • Primary: Stanford AI Index 2026, Chapter 7 (Education) 7.2 — hai.stanford.edu/ai-index/2026
  • Cross-country ICT data: OECD, 2025 — number of enrolled students, graduates and new entrants by field of education; ISCED classification framework
  • Notable omission acknowledged: OECD dataset excludes India, China, and most of Africa