Vertiv to Create Another 800 Jobs in Anderson County
Greg Wilson/Anderson Observer
In Anderson County, South Carolina, a million square feet of industrial ambition is being assembled in service of the invisible engines of the internet age. Vertiv, the Ohio-based company that has made its name tending to the power and cooling needs of data centers and other digital strongholds, is quietly enlarging its North American footprint with a cluster of new manufacturing facilities on the county’s industrial fringes.
The numbers, in the language of economic development, are reassuringly blunt: about 560,000 square feet of additional floorspace and a projection of as many as 800 full-time and contracted skilled jobs, an infusion of electricians, engineers, testers, and project managers into a labor market still learning to speak fluently about artificial intelligence and high‑performance computing. The project, combined with existing operations in the area, nudges Vertiv’s regional presence past the one‑million‑square‑foot mark, a scale that suggests less an outpost than a campus devoted to the infrastructure behind the cloud.
For Anand Sanghi, the company’s president for the Americas, the expansion is not simply a bet on Anderson County but on an emergent geography of AI, in which small Southern communities find themselves yoked to the appetites of distant server farms. The locations in the county, he has said, are “strategic,” a word that in this context means close enough to customers to matter and far enough from coastal real estate to remain feasible, and the new capacity is meant to anticipate the demands of data centers whose computing loads are measured less in spreadsheets than in neural networks.
Inside these buildings, Vertiv plans to produce the kinds of prefabricated, integrated systems that have become the industry’s answer to the old custom‑built approach—modules that arrive at a site already engineered, assembled, and tested, like giant, metal matryoshka dolls of electricity and coolant. The company’s integrated power modules, it says, can compress the schedule for deploying power systems by as much as half, while the SmartRun overhead infrastructure—an ungainly name for a unified run of busway, liquid‑cooling pipes, network cabling, and physical containment—is advertised as capable of making a white‑space fit‑out up to eighty‑five per cent faster than traditional methods.
The promise of speed, here, is less about impatience than about competition, about how quickly a data‑center operator can move from blueprint to billable compute cycles in a market where AI workloads grow at a pace that makes last year’s facilities feel faintly antique. Vertiv’s pitch, rendered in the careful prose of corporate strategy, is that power, cooling, IT hardware, and service ought to behave less like separate trades negotiating over a job site and more like facets of a single, end‑to‑end system, designed for the next generation or two of chips whose heat must eventually be ushered back into the world as waste—or, increasingly, as something to be reused.
On the ground in Anderson County, the expansion has a more tactile register. Tommy Haughton, the senior director of operations for the new facilities, talks about hands‑on training, support teams, and the effort to make new hires “feel confident and ready to succeed,” the sort of language that nods toward the reality that for many recruits the path to the AI revolution will begin not with an algorithm but with a torque wrench and a safety orientation.
If there is a quiet irony in all this, it lies in the contrast between the futuristic aura of artificial intelligence and the resolutely physical character of the infrastructure that supports it. The systems that allow an LLM to draft a paragraph or a recommendation engine to refresh a feed depend, in the end, on busways, pipes, and switchgear assembled in places like Anderson County, in buildings whose square footage can be counted, whose jobs can be tallied, and whose presence in the landscape—low, broad, and humming—is harder to romanticize than the abstractions they enable.
