At a glance
- At each site, generation, cooling and compute are to be engineered together rather than retrofitted into an existing facility, the companies said, under a framework that runs from site screening and interconnection strategy to project development, community engagement, and responses to grants and solicitations.
- The companies say the problem is power availability, not compute capacity: interconnection queues and transmission delays have added several years to conventional grid-connected projects, and Flux Core co-founder and chief executive Reginald York puts the premise in one line—"The industry does not have a compute problem so much as a power and timeline problem."
- None of that value is priced yet: the memorandum is non-binding and names no site, and a preferment is a designation, short of a power purchase agreement.
Data Center Dynamics reports that Tersis Technologies and Flux Core Data Systems and Energy have signed a non-binding memorandum of understanding to put decentralized data centers on waste-to-energy sites, with each company named the other's preferred partner—Flux Core for compute infrastructure at Tersis project sites, Tersis for on-site energy at Flux Core deployments.
At each site, generation, cooling and compute are to be engineered together rather than retrofitted into an existing facility, the companies said, under a framework that runs from site screening and interconnection strategy to project development, community engagement, and responses to grants and solicitations. A shared pipeline of candidate sites will be reviewed on a regular basis, and selected opportunities are meant to advance into project-level agreements across municipal, industrial, institutional, defense, rural and edge-computing applications.
The technology on each side supports the other: Tersis says its SynGenic V3 platform converts waste feedstocks into power and recovered products on site, delivering baseload-capable output of 10-30MW per site, while Flux Core's data center units arrive pre-configured, scale in increments as demand appears, accept off-grid and renewable power, and can be relocated when required.
The companies say the problem is power availability, not compute capacity: interconnection queues and transmission delays have added several years to conventional grid-connected projects, and Flux Core co-founder and chief executive Reginald York puts the premise in one line—"The industry does not have a compute problem so much as a power and timeline problem." The arrangement is the grid-slot thesis at distributed scale: a baseload-capable generator attaches an anchor load, and a compute provider sites its boxes where generation already exists. Tersis chief executive Antonio Uccello describes the aim as converting suitable waste feedstocks into useful on-site energy for AI compute closer to where it is needed, and the logic follows from the siting—a data center co-located with operating generation is no longer waiting on a transmission line, leaving the site's power availability as the scarce asset the two sides are trying to pair.
None of that value is priced yet: the memorandum is non-binding and names no site, and a preferment is a designation, short of a power purchase agreement. What exists is an option on a shared candidate pipeline. The first project-level agreement to emerge from the pipeline is where the trade gets priced, because the power price struck for a 10-30MW baseload plant with an on-site tenant will establish what the grid-slot trade is worth when a generator brings its own grid.
Reported by the Energy Transition desk from Data Center Dynamics. Published . Corrections