Vulcan AIVulcan AI

ABOUT

Twenty years on the iron the AI buildout now depends on

The brief in motion

Jon Moen: Twenty Years on the Iron AI Depends On

Jon Moen, Founder and Principal of Vulcan AI

Founder and Principal

Jon Moen

Twenty years and fifty-plus data centers before founding Vulcan AI.

The constraint moved, and most of the industry has not caught up

Vulcan AI is built on a specific, testable observation: the constraint on AI capacity is no longer silicon, capital, or talent. It is powered, cooled floor space, and the standard playbook for creating that floor space is too slow for the market it is trying to serve. Vulcan AI exists to compress that timeline for owners, operators, and investors who need AI capacity measured in hundreds of megawatts, not in press releases.

Dark forge scene with purple-tinted steel anvil and controlled orange fire
Engineering judgment forged on the hardware.

Twenty years of the exact iron this industry depends on

Jon Moen founded Vulcan AI after twenty years building high-density computing infrastructure and deploying more than fifty data centers. As Technical Director of EKWB USA, he designed, built, and benchmarked liquid-cooled GPU systems at the highest end of the market, shipping $500K servers to MIT, Cornell, CrowdStrike, the US Navy, NATO, and Fortune 500 operators. Those are not reference logos borrowed for a slide. They are institutions that do not accept a cooling failure as an inconvenience, and the systems Jon built for them had to hold up accordingly.

Why he founded Vulcan AI

Two patterns kept repeating across fifty deployments. The first: projects that measured their density ceiling early moved fast and spent capital on the right problems, while projects that discovered it late paid for that discovery in change orders. The second: the fastest path to real AI capacity increasingly ran not through new construction, but through industrial sites, retired power plants, closed mills, dead malls, that had gone dark for reasons unrelated to their power infrastructure. The interconnection agreements, substations, and permits that took the original owner years to secure were often still live, waiting for someone who understood both the thermal engineering and the retrofit economics to put them back to work. Vulcan AI is built to run both plays: measure the constraint before it becomes a change order, and find the power that already exists before waiting years for new power that does not.

The retrofit thesis, proven in public

The pattern is no longer theoretical. A Somerset, NY coal plant is now TeraWulf's 500 MW Lake Mariner AI campus. A paper mill in Lufkin, Texas, closed since 2007, is scaling toward 1.1 GW by 2028. A dead mall in San Antonio became Rackspace's first widely known mall-to-data-center conversion. Each of these traces the same shape: industrial or retail infrastructure that lost its original purpose while keeping the power, water, and structure a new purpose could use.

What Vulcan AI does

Vulcan AI works on the owner's side across four engagements: the AI Data-Center Readiness Assessment, a two-week audit of a facility against AI-density requirements; the Liquid-Cooling Design Review, an independent check of a cooling design before capital is committed; Retrofit Site Acquisition Advisory, finding and qualifying industrial sites with live power rights; and Deployment Oversight, owner-representative engineering through the build to first load. Most engagements begin with the readiness assessment, and each service page lists a starting price with final scope set per engagement.

Next step

Work with the person who has done this fifty times

See how the practice is staffed, or send your project details directly.