The shortest distance between an AI budget and a payslip
Most stories about AI and jobs live in forecasts. This one lives on a workbench in Winnsboro, Louisiana, where students in work overalls are learning to splice fibre optic cable in a four-week course at Louisiana Delta Community College, and leaving with offers to work on the largest data centre Meta has ever built.
Business Insider reported on 18 October that the college’s Data Center Technician program — priced at close to US$3,700, aimed squarely at Meta’s Hyperion project in rural Richland Parish — has become a genuine hiring pipeline. Its inaugural class graduated in March: all 11 students received job offers from IES Communications, a contractor working on the Meta build. The program now produces about 24 graduates a month, with a backlog of applicants trying to get in.
The economics for students are stark. Trey Wadley, 30, a former automotive worker, and John Crowe Jr., who previously averaged US$14 an hour at FedEx, both told Business Insider they hope to land roles paying US$25 to US$30 an hour. Program director Dan McCarty said entry-level positions can reach US$24 an hour, with graduates telling him they hit US$27 to US$28 within six months. For a region with no state minimum wage law and an agriculture-heavy economy, that is the difference between leaving and staying. “In the past, a lot of people, including myself, had to leave here to find the good jobs,” McCarty said. “Now they’re 30 minutes down the road.”
The numbers behind the boom
Hyperion is under construction in Richland Parish at nearly 10 million square feet — more than 3,600 acres of what currently resembles, in the words of one description, a large airport with no planes. Meta’s own figures put over 7,500 trade workers on-site at peak construction, with about 1,000 permanent operational jobs once the facility is fully online by 2036.
The company has underwritten the training pipeline around the build. Meta pledged US$5 million in scholarships covering data centre-related courses at Delta for anyone graduating from a Richland Parish high school, and last year launched a broader “America’s Workforce Academy” — a reported US$115 million program offering free four-to-five-week construction training in four states, with housing, transport and a daily stipend, aiming for job placement at completion. Meta’s own policy staff have been blunt about why: there is, as one director put it, “a huge gap in the labour market when it comes to construction workers, power generation, grid modernization.”
That gap is the quiet engine of the whole story. The AI buildout now employs electricians, pipefitters, carpenters, fibre technicians and even squirrel-resistant cable splicers — an instructor’s joke that made the classroom laugh, and a real occupational hazard in rural Louisiana. These are physical, local, hard-to-automate jobs sitting directly underneath an AI economy usually described in white-collar terms.
Why a blue-collar path is the AI jobs story that pays
The Louisiana numbers rhyme with wage data published earlier this year: a Business Insider report in August found the median pay for data-centre facility managers was US$134,000 — 64 percent above other facility managers — while project managers in the sector earned a median US$115,000, about 12 percent above peers elsewhere. A site that follows the money in trades employment has tracked the same pattern: the AI supply chain is bidding up hands, not just heads.
There is also a defence of the pathway’s durability worth stating plainly. Skills instructor John Bradley told Business Insider the training is deliberately transferable: “It doesn’t have to be a Meta data center… You still got banks, you still got IT departments.” That matters for anyone wary of betting a career on one construction project — fibre, power and facilities skills outlive any single campus.
The caution column is real, though. A Gallup poll from earlier this year found 7 in 10 Americans would rather not live near a data centre, citing environmental strain and higher power bills — and Richland Parish residents told the outlet they have mixed feelings. Meanwhile the region’s long-term employment prize, roughly 1,000 permanent roles by 2036, is modest next to its construction-phase peak. Buildouts end.
What New Zealand can and can’t take from it
New Zealand has no Hyperion and, on current settings, is unlikely to get one — the IMF’s recent assessment of New Zealand’s AI position noted the country’s gaps in capital markets depth and diffusion rather than gigawatt campuses. But the transferable idea is the training architecture, not the building: employer-funded, weeks-long, credential-backed micro-courses pointed at real projects, with hiring pipelines attached before the first class starts.
There are signs the same logic applies here in miniature. When New Zealand employers name AI skills in a record share of job ads, what they are buying is closer to “can operate and maintain the new machinery” than frontier research. And the strongest hiring demand in NZ’s own postings data sits not in software but in healthcare, consumer services and the public sector — workplaces that need technicians, not theorists.
A genuinely NZ version would look less like Meta’s academy and more like what Delta’s director described: employers telling a polytech exactly which skills are missing, equipment money moving the same month, and a promised job at the end of four weeks. The US$3,700 tuition against a US$13-an-hour pay rise pays for itself in under a year — a repayment math any NZ trades programme could replicate, if the projects exist to hire at the end of it.
🔍 THE BOTTOM LINE
- Meta’s Hyperion data centre in rural Richland Parish, Louisiana — nearly 10 million square feet — is generating over 7,500 peak construction trade jobs and about 1,000 permanent operational roles by 2036, per Meta’s figures.
- Louisiana Delta Community College’s four-week Data Center Technician course (about US$3,700) placed all 11 graduates of its first class with a project contractor; it now graduates about 24 students a month against a backlog.
- Reported entry pay of up to US$24 an hour, reaching US$27–28 within six months, versus US$14 an hour in prior warehouse work, anchors the “why not get in now” appeal in a region with no state minimum wage.
- Meta has pledged US$5 million in local scholarships and runs a US$115 million Workforce Academy across four states; 70 percent of Americans tell Gallup they don’t want a data centre nearby, so community consent remains part of the bargain.
- The transferable lesson for NZ: short, employer-funded, credential-backed training pointed at named projects converts AI budgets into wages faster than any reskilling masterplan.
❓ FAQ
Are data centre jobs really a path into AI work? The direct AI content of the roles is low — fibre splicing and facilities maintenance are not model engineering. But they sit on the critical path of AI infrastructure, pay a documented premium over comparable facilities work, and the technician skills transfer to banks, IT departments and telecoms, per the instructors who teach them.
What does the training cost and pay? About US$3,700 for four weeks of coursework at Louisiana Delta Community College, with reported entry wages up to US$24 an hour and US$27–28 within six months. Against a previous US$14 an hour job, that is roughly a US$27,000 a-year bump for a one-time course fee.
Could a pathway like this work in New Zealand? The structure could — employer-specified skills, weeks rather than years, a job at the end. The difference is scale: NZ has no gigawatt campus under construction, so the equivalent would attach the same architecture to local infrastructure and grid projects, where employers already report difficulty finding skilled staff.
What are the risks in betting a career on AI buildouts? Construction phases end; the permanent figure at Hyperion is about 1,000 roles by 2036 against a 7,500-worker peak. Environmental concern in host communities is real, per polling. The mitigant, instructors say, is transferability — the same trades serve power, telecoms and facilities work anywhere.
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
