What are the most in-demand AI skills?
Python is the most in-demand AI skill: it appeared in 258,674 U.S. AI job postings in 2025, according to Lightcast data published in Stanford's AI Index Report 2026. Computer science was a close second at 257,127 postings, followed by scalability (197,744), automation (190,758) and workflow management (186,325). The rest of the top 10 are data analysis, SQL, project management, data science and Amazon Web Services.
Figure 1, The Skill Stretch, sets each of those skills against its 2013–15 level. The longest stretches are not in classic data-science skills. Amazon Web Services grew 1,358%, workflow management 818% and scalability 733%. SQL grew the least, at 132%. Employers hiring for AI now want people who can run systems at scale, not only people who can build models.
How many job postings require AI skills?
In 2025, 2.56% of all U.S. job postings required AI skills, according to Lightcast's figures in the AI Index. Lightcast, which rounds the number to 2.5%, says that share is up 55% from 2024 and nearly 300% from a decade earlier. By skill cluster, the broad "artificial intelligence" cluster appeared in 1.70% of all postings and machine learning in 0.99%. A posting can fall into several clusters at once.
Which AI skills are growing fastest?
Agentic AI is the fastest-growing AI skill: postings that named it rose from 151 in 2024 to 16,541 in 2025, a 10,854% increase. Postings naming "AI agents" rose from 1,310 to 15,217, and LangGraph, an agent orchestration framework, rose from 194 to 4,294. Lightcast puts the whole AI agent cluster at 0.23% of all U.S. postings in 2025, up from 0.06%, which already ranks it ahead of natural language processing (0.22%).
Generative AI skills kept growing too. Postings naming generative AI itself rose 111% to 138,188, and large language modeling doubled to 38,526. Retrieval augmented generation grew 337% and prompt engineering 261%. Chatbot-era skills slipped by comparison: ChatGPT postings still grew 160%, but ChatGPT's share of AI agent skill mentions fell from 25.26% to 16.43%.
Which industries hire the most AI talent?
The information sector hires the most AI talent relative to its size: 13.22% of its job postings required AI skills in 2025, up from 7.82% in 2024. That is about twice the share in professional, scientific and technical services (6.49%) and well ahead of finance and insurance (5.33%) and manufacturing (4.66%). Every one of the 15 sectors in the AI Index rose from 2024, including sectors with low adoption such as transportation and warehousing (1.26%).
Which states and countries have the most AI jobs?
California has the most AI jobs, with 170,881 AI job postings in 2025, or 17.18% of the U.S. total. Texas (80,547) and New York (66,029) follow, and the three states together account for roughly a third of U.S. AI postings. Across countries, Singapore has the highest share of postings requiring AI skills (4.69%), followed by Hong Kong (3.48%) and Luxembourg (3.43%). The United States, at 2.56%, ranks 9th of the 22 countries and territories Lightcast tracks.
Methodology & caveats
All figures come from Lightcast's database of online job postings, as published in Chapter 4 of Stanford HAI's AI Index Report 2026, with chart values read from the report's figures (4.4.1 to 4.4.10). Data covers calendar year 2025, compared with 2024 or with a 2013–15 baseline period as the report does. Lightcast classifies skills with its own open skills taxonomy, and a single posting can list several skills, so skill counts overlap and cannot be added up. The report notes that historical posting counts may differ from earlier editions because of data revisions.
This graphic began from a Visual Capitalist chart of an earlier edition of the same series; every number here was re-checked against the 2026 edition. Card labels shorten some Lightcast skill names and NAICS sector names; the full names are in each card's note. The state total of 994,372 postings is derived by adding the 51 state and DC counts; it reproduces the report's published state shares. The report and Lightcast's summary differ slightly in rounding (U.S. share 2.56% in the chart, 2.6% in the report text, 2.5% in Lightcast's post; Singapore 4.69% in the chart, written as 4.769% in Lightcast's post), and the report cites both 17.2% and 17.9% for California's share in different charts. Where they differ, this graphic uses the chart values.
Sources
- [1]Stanford HAI, AI Index (data: Lightcast). AI Index Report 2026, Chapter 4: Economy, section 4.4 Jobs (Figures 4.4.1–4.4.10, Lightcast job postings, 2025)https://hai.stanford.edu/assets/files/ai_index_report_2026_chapter_4_economy.pdf
- [2]Lightcast. Four Takeaways from the 2026 Stanford AI Indexhttps://lightcast.io/resources/blog/stanford-ai-2026


Automatic.co