Latest / Can't Find Job? AI Is Quietly Replacing Millions of Workers / U.S. Data-Hosting Employment Fell in 2026: What the Payroll Numbers Actually Measure

Transcript

U.S. Data-Hosting Employment Fell in 2026: What the Payroll Numbers Actually Measure · Can't Find Job? AI Is Quietly Replacing Millions of Workers

• 0:00
  SPEAKER_00: U.S.
• 0:00
  SPEAKER_00: data hosting employment fell in 2026.
• 0:03
  SPEAKER_00: What the payroll numbers actually measure by Ifyak Namdi, Human Potential Advocate.
• 0:09
  SPEAKER_00: The first U.S.
• 0:10
  SPEAKER_00: payroll estimate for August 2026 showed a sharp monthly drop in jobs tied to computing infrastructure, data processing, and web hosting.
• 0:19
  SPEAKER_00: It put employment at 453,300, down 7,700 from July and 26,100 from August 2025.
• 0:30
  SPEAKER_00: That is a real warning for job seekers, but it is not a head count of people working inside physical data centers.
• 0:37
  SPEAKER_00: And the monthly drop was later revised down.
• 0:40
  SPEAKER_00: Those two facts change how to read the headline.
• 0:43
  SPEAKER_00: The numbers point to a weaker job count in one broad industry group.
• 0:48
  SPEAKER_00: They do not prove that AI caused layoffs or that data center investment has stopped creating work.
• 0:55
  SPEAKER_00: What the August payroll numbers showed, the Bureau of Labor Statistics, BLS, released its August 2026 payroll report on September 4, 2026.
• 1:06
  SPEAKER_00: In that first report, seasonally adjusted employment in computing infrastructure providers, data processing, web hosting, and related services was 453,300.
• 1:18
  SPEAKER_00: The table showed 461,000 for July and 479,400 for August 2025.
• 1:26
  SPEAKER_00: For the full table, please open this article on can'tfindjob.com.
• 1:31
  SPEAKER_00: Changes are calculated from the September 4th report seasonally adjusted figures.
• 1:36
  SPEAKER_00: The year-over-year declines are about 5.4% for computing and hosting, 3.0% for publishing, and 4.0% for the broader information sector.
• 1:48
  SPEAKER_00: BLS reported that the information sector lost 23,000 jobs in August, after averaging losses of about 8,000 a month over the prior year.
• 1:57
  SPEAKER_00: The first estimate showed losses across several parts of the sector: 7,700 in computing and hosting, 6,700 in publishing, and 5,000 in broadcasting and content providers.
• 2:10
  SPEAKER_00: That comparison matters.
• 2:12
  SPEAKER_00: Publishing lost nearly as many jobs as computing and hosting in August.
• 2:16
  SPEAKER_00: Over the year, its decline was slightly larger.
• 2:20
  SPEAKER_00: The first report showed a broad information sector slide, not a job loss unique to data centers or AI.
• 2:26
  SPEAKER_00: The 8,000 job monthly drop has already been revised.
• 2:30
  SPEAKER_00: The September Employment Situation Report, released on October 2, 2026, updated the seasonally adjusted figures.
• 2:38
  SPEAKER_00: It showed 455,400 jobs in July and 453,800 in August, a revised monthly decline of 1,600, not 7,700.
• 2:51
  SPEAKER_00: For the full table, please open this article on can'tfindjob.com.
• 2:56
  SPEAKER_00: This does not mean the first report was a mistake.
• 2:59
  SPEAKER_00: BLS payroll estimates are preliminary.
• 3:02
  SPEAKER_00: The agency updates them as more employer reports arrive, and seasonally adjusted estimates can change as factors are recalculated.
• 3:10
  SPEAKER_00: As of October 14, 2026, the October 2nd estimate was the latest available, but further revision was still possible.
• 3:19
  SPEAKER_00: So use the figures carefully.
• 3:21
  SPEAKER_00: Negative 7,700 describes what BLS first reported on September 4th.
• 3:27
  SPEAKER_00: The latest published month-to-month estimate was negative 1600.
• 3:32
  SPEAKER_00: The first report's August 2025 comparison is also tied to that report's data vintage.
• 3:37
  SPEAKER_00: Don't mix it with revised figures and present the result as one consistent release.
• 3:43
  SPEAKER_00: What this payroll category actually counts, the industry name sounds like data centers, but the category is wider than physical data center sites.
• 3:52
  SPEAKER_00: The Census Bureau's industry definition includes businesses that provide computing infrastructure and data processing, as well as web hosting.
• 4:00
  SPEAKER_00: Cloud storage, application hosting, and related services.
• 4:04
  SPEAKER_00: It does not describe one building type or one occupation.
• 4:08
  SPEAKER_00: The payroll figure is an estimate of jobs at establishments classified in that industry.
• 4:13
  SPEAKER_00: It counts employees on payroll who received pay during the pay period that includes the 12th of the month.
• 4:20
  SPEAKER_00: Full-time and part-time employees count.
• 4:22
  SPEAKER_00: A person with jobs at two establishments can be counted on both payrolls.
• 4:27
  SPEAKER_00: Self-employed people and proprietors are not included.
• 4:30
  SPEAKER_00: That creates important limits.
• 4:32
  SPEAKER_00: It is not a count of workers inside data center buildings.
• 4:36
  SPEAKER_00: BLS classifies an establishment by its main business activity, not simply by where a person works.
• 4:42
  SPEAKER_00: It is not an occupation count.
• 4:45
  SPEAKER_00: The table does not tell us whether jobs were in operations, software, security, maintenance, or management.
• 4:52
  SPEAKER_00: It is not a layoff count.
• 4:53
  SPEAKER_00: A smaller payroll total can reflect fewer hires, more departures, restructuring, or other changes.
• 5:00
  SPEAKER_00: This table does not identify the cause.
• 5:03
  SPEAKER_00: It can miss work done by outside vendors.
• 5:06
  SPEAKER_00: BLS guidance says contract workers are counted under their employer of record, rather than automatically under the industry where the work takes place.
• 5:14
  SPEAKER_00: For job seekers, that means data center employment is not one clean bucket.
• 5:19
  SPEAKER_00: Some work may be counted under computing and hosting, other work at the same facility may sit under construction, utilities, engineering, or a contractor's industry.
• 5:30
  SPEAKER_00: What openings and hires say and what they do not, BLS's Job Openings and Labor Turnover Survey, released on September 29th, offers another clue about the broader information sector.
• 5:42
  SPEAKER_00: In August, it estimated 123,000 job openings, up from 78,000 in July, but below 141,000 a year earlier.
• 5:52
  SPEAKER_00: It estimated 44,000 hires, down from 62,000 in July and 69,000 in August 2025.
• 6:01
  SPEAKER_00: That is a mixed signal, more openings than the month before, but fewer hires.
• 6:06
  SPEAKER_00: It also points to softer hiring than a year earlier, but GOLTS reports these figures for the whole information sector, not just computing infrastructure and hosting.
• 6:16
  SPEAKER_00: Its openings figure is a snapshot of qualifying positions open at month's end.
• 6:21
  SPEAKER_00: Hires are changes to payrolls over the full month.
• 6:24
  SPEAKER_00: Neither measure tells us how many people a data center employer hired.
• 6:28
  SPEAKER_00: Does this prove AI is cutting data center jobs?
• 6:31
  SPEAKER_00: No, the payroll figures show a change in employment.
• 6:35
  SPEAKER_00: They do not name the cause.
• 6:36
  SPEAKER_00: The first August report offers one useful test.
• 6:40
  SPEAKER_00: Publishing and broadcasting also lost jobs.
• 6:43
  SPEAKER_00: That pattern fits broader weakness across parts of information better than a simple story in which AI alone caused the hosting industry drop.
• 6:52
  SPEAKER_00: It does not rule out AI's role.
• 6:55
  SPEAKER_00: It does mean the payroll table cannot establish it.
• 6:59
  SPEAKER_00: Data center investment is a separate question from payroll employment.
• 7:03
  SPEAKER_00: A company can spend more on servers, power, cooling, and buildings without adding the same number of permanent operating jobs.
• 7:12
  SPEAKER_00: More computing capacity does not require one-for-one growth in headcount.
• 7:17
  SPEAKER_00: That is a reasoned possibility, not something this BLS table proves.
• 7:22
  SPEAKER_00: Construction data need the same care.
• 7:25
  SPEAKER_00: In the first August report, construction employment rose by 22,000, including an 8,000 increase among non-residential specialty trade contractors.
• 7:36
  SPEAKER_00: Those totals cover many kinds of work.
• 7:39
  SPEAKER_00: They do not isolate data center projects, so they cannot confirm how much hiring came from a data center buildout.
• 7:47
  SPEAKER_00: The careful conclusion is simple.
• 7:49
  SPEAKER_00: AI and data center investment may help explain the changing mix of jobs, but these payroll and jails figures cannot show that link on their own.
• 7:59
  SPEAKER_00: They measure industry employment and hiring, not investment, AI use, or the number of workers assigned to particular facilities.
• 8:08
  SPEAKER_00: What data center and hosting job seekers should do.
• 8:11
  SPEAKER_00: For workers, the lesson is not to abandon this field.
• 8:15
  SPEAKER_00: It is to look beyond a single industry label or job title.
• 8:19
  SPEAKER_00: Search across the whole infrastructure chain.
• 8:22
  SPEAKER_00: Search for roles with cloud and hosting providers, but also consider employers in construction, utilities, engineering, equipment, and contracting.
• 8:33
  SPEAKER_00: Similar work can appear in different payroll industries because it is counted under the employer's main activity.
• 8:40
  SPEAKER_00: Build skills that travel, pair digital skills with the ability to keep real systems working.
• 8:46
  SPEAKER_00: Depending on your path, that might mean learning cloud operations, networking, cybersecurity, power systems, cooling, automation, or fault diagnosis.
• 8:57
  SPEAKER_00: The goal is not to guess one perfect future job title, it is to become useful across more than one kind of employer.
• 9:04
  SPEAKER_00: Prepare to work with automation, AI tools can speed up routine tasks.
• 9:09
  SPEAKER_00: Workers who can use those tools, check their output, spot failures, and make sound decisions when systems break will be better prepared for changing work.
• 9:19
  SPEAKER_00: The safer bet is not to compete with software at routine speed.
• 9:23
  SPEAKER_00: It is to learn how to guide, verify, and repair the systems around it.
• 9:29
  SPEAKER_00: Conclusion: read the category, not just the headline.
• 9:32
  SPEAKER_00: The September 4 report showed data hosting employment at 453,300, down 7,700 in one month, and 26,100 from a year earlier.
• 9:45
  SPEAKER_00: It also showed losses in publishing and broadcasting.
• 9:48
  SPEAKER_00: Then, on October 2nd, BLS revised the monthly decline to 1,600.
• 9:54
  SPEAKER_00: That is the most useful reading of the numbers.
• 9:57
  SPEAKER_00: The first estimate signaled a weak year-over-year trend, but its dramatic monthly drop did not hold up in the next release.
• 10:04
  SPEAKER_00: And the industry category is broader than physical data centers, so it cannot tell us exactly which jobs vanished or why.
• 10:12
  SPEAKER_00: AI infrastructure may keep growing, while traditional payroll jobs change.
• 10:17
  SPEAKER_00: For job seekers, the practical move is to build skills that apply across the system.
• 10:22
  SPEAKER_00: Not only the building, but also the power, networks, software, security, and services that make computing work.
• 10:29
  SPEAKER_00: All links to sources are available in the text version of this article.
• 10:33
  SPEAKER_00: You can find the full article at can'tfindjob.comslash blog.
• 10:38
  SPEAKER_00: Thanks for listening.
• 10:39
  SPEAKER_00: If today's episode hit close to home, stop scrolling job boards that weren't built for this new reality.
• 10:46
  SPEAKER_00: Check out Claw Earn on AI Agent Store.
• 10:48
  SPEAKER_00: AI, the first jobs marketplace designed for both humans and AI agents, so you can start earning, no matter which side of the AI revolution you're on.