Who Builds Data Centers in the US?

To answer who builds data centers, this data book covers four parts of the work behind large-scale computing: building and commissioning data centers, operating them, manufacturing the chips inside them and writing the software that runs on them. Each group is identified from job titles.

Caveat: A job title doesn’t always tell you what someone does day to day. The method section explains where that limits the analysis.

Job postings show what employers are looking for, not who they end up hiring. This section compares the number of US postings in each group in September 2024, September 2025 and September 2026.

The headline measure is the change in monthly postings between two single months, shown for the full two years and for each year separately. Total US postings in the data were lower in September 2026 than in September 2024, so each group's share of all US postings, shown as a secondary measure in the table, rose faster than its raw count.

Most of the increase in data center postings came in the first year. Electrical and power postings rose 190% in year one and 14% in year two, cooling and mechanical postings 189% and 2%, and compute software postings 85% and 2%. Semiconductor and technician postings grew least over the two years. Comparisons between single months are sensitive to a small number of postings.

Two sources of evidence on experience are compared: the minimum experience employers request in job postings, and the experience recorded on the profiles of US professionals who currently hold these roles.

The two measures diverge for data center technicians. Among the 25,946 technician postings that state a minimum, 21,676 (84%) ask for three years or less, while only 2,488 of the 12,887 US technician profiles with recorded experience (19%) show four years or less. The measures are not identical, because postings state requirements and profiles record careers, so the comparison indicates direction only.

Draup’s data shows that existing Data Center talent has higher median years of experience than what is being asked in new job postings. The high demand for this talent pool is possibly bringing down the years of experience required for these roles. Organizations may also be looking for adjacent talent pools with high experience to make up for the shortfall in the highly experienced data center talent. Postings were counted by US metro area and divided between data center titles, which cover construction and operation, and compute software titles. The two groups cluster in different places.

Each metro is flagged for data center construction and shows the change in its monthly data center postings, excluding technician roles, between September 2024 and September 2026. Of the metros studied, 15 have data centers under construction, 5 have projects planned or proposed and 2 had no under-construction project in the sources reviewed. Metros with small counts start from low bases, so their percentage changes are volatile. The evidence behind each flag is listed in the table at the end of this section.

Of the ten largest employers of electrical, cooling and mechanical data center roles, five are engineering, construction, real estate services or equipment companies. None of the ten largest employers of compute software roles falls into those categories. That list consists mainly of chipmakers and large technology companies, together with a job board and a bank.

We looked at which roles are growing fastest, how much experience employers ask for, where data center construction projects are concentrated, and which employers are hiring. The data shows demand rising fastest in the trades that most people don’t associate with AI. Demand rose in every metro, even where construction has not started. For talent leaders and economic developers, this is the workforce that decides how fast AI infrastructure gets built. Understanding where it is, how experienced it is, and who is competing for it is the first step in planning for it.

Postings, experience, locations and employers. Figures are drawn from Draup's job-posting table in its data warehouse, restricted to postings located in the United States from September 2024 to September 2026, and were queried on October 16, 2026. Job titles were matched on the normalized title field. Counts are postings, not openings or hires, and may include reposts. Growth is the change in monthly postings between September 2024 and September 2026, with September 2025 as the midpoint. Shares of all US postings are a secondary measure.

Experience requirements. The share asking for three years or less is calculated from postings that state a minimum, so it includes postings that ask for three or more years.

Workforce profiles and experience. Figures come from Draup's professional profile snapshot, restricted to US profiles carrying the named role, queried on October 16, 2026. The share of data center technician profiles with four years or less is calculated from profiles with recorded experience and was queried on October 22, 2026. A profile may carry more than one role.

Talent size. Estimates come from Draup Talent Intelligence for the United States, queried on October 26, 2026, and combine Draup's professional profiles and job postings with Bureau of Labor Statistics and International Labour Organization data. Relative standard errors are 5% to 10%. The estimates were refreshed between January and October 2026, depending on the role.

Construction flags. Flags are based on Draup news signals in the Center Setup and Expansion category, captured between January and October 2026 and searched by metro, and on CBRE's North America Data Center Trends H1 2026 report, which gives megawatts under construction for its primary markets. Where Draup signals had no coverage, as in Des Moines and Omaha, public project trackers (Baxtel and Epoch AI) were used and are named in each metro's note. The flags rest on news reporting and have not been checked against permits.

Metro demand growth. For each metro, monthly data center postings in September 2026 were compared with September 2024, excluding technician titles because bulk postings distort them, using the same warehouse table and US filter as the other demand figures. Metros with small counts, such as Abilene, show very large percentage changes.

Limitations. Job titles are an imperfect guide to the work. Electricians hired directly by contractors may not carry "data center" in the title, so electrical roles are probably undercounted. The compute software group includes some titles that describe chip design. Total US postings varied from month to month, which is why shares of all postings are shown alongside raw counts. Growth rates for smaller metros rest on counts in the tens to low hundreds a month and move easily. Talent size is modeled, and profile data may over-represent experienced workers and under-represent skilled trades. Compensation, hiring difficulty and workforce supply by metro are not covered in this edition.

© 2026 Draup. All rights reserved.