Kaufman & Company, the permanent capital holding company founded by real estate investor Daniel Kaufman, has announced that its Project Zero micro data center prototype is now deployed on site in Detroit. The announcement also details ZeroPoint, the company’s platform for developing the physical infrastructure behind artificial intelligence. The move matters because it marks a tangible step from design to deployment in the race to build compute capacity, and it suggests a model that could be replicated in existing buildings and older industrial cities rather than only on greenfield land.
Project Zero is a modular, vendor-agnostic micro data center system built by Kaufman Development. Each unit is manufactured in a factory, shipped flat-packed, and assembled on site, giving operators compact, low-latency edge compute without a purpose-built facility. A containment design keeps cold supply air separate from hot exhaust, which is intended to reduce cooling loads and energy costs. The company’s design target is a 20% to 30% lower total cost of ownership than standard containerized systems; that figure will be measured as the Detroit prototype matures.
“We are starting with something we can stand next to and test,” Kaufman said. “A working unit in Detroit tells us more than any rendering. Prove the model, learn what breaks, then scale it with discipline.” That pragmatic approach is significant because it prioritizes real-world validation over speculative design, potentially accelerating the path to scalable, cost-effective AI infrastructure.
Alongside Project Zero, ZeroPoint includes two hyperscale projects in predevelopment: the Rio Grande Data Campus in El Paso, Texas, and the Wasatch Data Campus in Utah. LoneStar Kaufman Development Partners and ServerDomes Global, which is independently owned and operated, are joint venture partners on both campuses, with Kaufman Ventures as an investor. Kaufman said the platform is built on the same view that shapes the company’s housing work: infrastructure should leave something useful behind in the places where it is built.
“Compute is becoming as basic to a local economy as power and water,” Kaufman said. “If we are going to build infrastructure for machines, we have to be just as serious about the people and the neighborhoods around it. Small, modular sites let us put capacity where it is needed, including in existing buildings and older industrial cities, instead of only on greenfield land.” This community-oriented approach could differentiate ZeroPoint in a market often criticized for prioritizing speed and scale over local impact.
ZeroPoint is one of eight brands Kaufman & Company introduced this month across its Residential, Places and Infrastructure divisions. More information is available at thekaufmanco.com/platform-zeropoint. The company, whose platform represents more than $2 billion in project value and more than 10,000 housing units across more than 25 years of building, lending and investing, operates through 10 companies including Kaufman Development and Kaufman Ventures. Its entry into AI infrastructure, anchored by a working prototype and a clear cost-reduction target, could signal a new phase in the buildout of edge compute capacity—one that leverages existing industrial assets and aims to deliver lasting local benefits. As AI demand surges, the ability to deploy modular, low-latency data centers quickly and affordably may become a competitive advantage, and Kaufman’s experiment in Detroit will be watched as an early test of that thesis.
