AI compute closer to where it's needed.
FabricIn: the fabric of infrastructure for AI.
FabricIn connects underused power, space, and fiber at hundreds of small sites into one orchestrated network. This allows AI companies to run inference near their users for lower cost and faster TTFT, without waiting for new data center capacity.
Centralized cloud wasn't built for this.
AI inference and agentic workloads are outgrowing centralized hyperscale cloud. The industry is building new capacity in the wrong shape to catch up.
Demand has changed
Enterprises need answers in milliseconds, need to keep data inside a specific city, state, or country, and don't want to own and depreciate scarce AI hardware themselves. Token consumption on AI platforms grew roughly 10x year over year through the end of 2025, and agentic AI tasks generate far more network traffic than a person browsing directly.
Supply is stuck
New hyperscale data centers commonly take 24 to 36 months to reach power, gated by utility interconnection queues and long equipment lead times, and a growing number of major metro areas are limiting new construction even for privately powered sites.
Capacity is already out there
Hundreds of small sites, from cell tower ground space to commercial building plant rooms, already have power and fiber in place, sitting underused.
Nobody is aggregating it
Connecting that stranded capacity into something AI companies can actually run workloads on is a role nobody neutral has taken on yet. That's the gap FabricIn is built to fill. This category doesn't have an established leader yet, and the operator who defines what "AI-ready" infrastructure actually means sets the standard everyone else gets measured against.
Source: Cisco, "AI Impact on Wide Area Networks" (May 2026)
One network, two kinds of sites, orchestrated in real time.
FabricIn combines two site types into a single fabric, then routes each workload to wherever it runs best. FabricIn sits above the sites as a neutral layer, not an infrastructure landlord.
Commercial real estate
Larger sites inside Class A commercial buildings, where multi-megawatt power is already available. Fast to activate. Most of the buildout is networking and interior fit-out, not new construction.
Tower & infrastructure real estate
Compact, modular sites on existing tower and infrastructure company real estate. Standardized and quick to deploy, reaching markets a large hub can't cover alone.
A fiber mesh connects sites within each metro at low latency. FabricIn's orchestration layer then decides, in real time, which site should handle each workload, based on latency, where the data is allowed to live, available power, and current capacity.
The kind of company that trusts us with their infrastructure.
We're in active partnership discussions across the categories below, with one live pilot already underway. Most of these relationships are still confidential, so we're describing them by type rather than by name. Reach out if you'd like specifics.
Tower & infrastructure companies
Among the largest owners of real estate and power in the U.S., providing the Edge Spoke footprint.
National fiber carriers
Dark-fiber and network partners who mesh our sites together at metro scale.
Commercial real estate
Class A building owners in dense urban markets, hosting our Anchor Hub sites.
AI-native infrastructure providers
Fast-growing compute companies piloting workloads on the fabric today.
Three kinds of conversations, one network.
Figuring out if FabricIn is for you
FabricIn is built for teams running AI inference or agentic workloads who need to be physically close to their users, or need data to stay within a specific city, state, or country. It's built for teams that want this without spending years building infrastructure themselves.
- You operate your own GPU fleet, or work with a systems integrator or managed-compute partner who provisions and operates hardware on your behalf
- Latency to your end users matters. Every hop counts
- You have data residency, sovereignty, or compliance requirements tied to a specific location
- You're evaluating several markets at once and can't wait 24 to 36 months for new data center capacity
New demand for capacity you already have
If you operate fiber, tower, or commercial real estate assets, FabricIn brings AI inference demand to space, power, and fiber capacity that would otherwise sit underused. You don't have to build a compute business of your own to capture it. We handle site normalization, orchestration, and tenant demand; you keep doing what you do best.
AI infrastructure that stays close to home
We're piloting distributed AI infrastructure with municipal and utility partners. Together we're exploring how local, low-latency compute can support economic development, public-sector AI adoption, and smarter use of existing grid and real estate assets, without a new, multi-year data center build in your community.
The people building the fabric.
FabricIn is a small team of infrastructure operators, not a large company pretending to be one.
Daniel Herb
Previously Chief Product Officer at IONX and at Vapor IO, where he built one of the first distributed edge networks in the U.S.
LinkedIn ↗
Aaron Hinkle
Previously VP of Solutions and Network Architecture at Vapor IO.
LinkedIn ↗
Mark Meuleman
Previously CRO at IONX, growing neutral-host revenue from pilot stage to more than $10M with Fortune 100 brands.
LinkedIn ↗
Lee Dawson
Previously led service-provider partnerships for IONX's neutral-host model, with an innovation background at major national carriers.
LinkedIn ↗We're a small team building something big.
FabricIn is early. We're in stealth mode, Seattle-based, and growing. We're a capital-efficient, milestone-driven build, not a "move fast and break things" consumer startup: measured claims, real engineering, ground-floor ownership. If you want to help build the neutral host layer for distributed AI, we'd like to hear from you, even if nothing below is an exact match.
Let's talk.
Whether you're an AI company evaluating infrastructure, a partner with power, space, or fiber to offer, a city or utility exploring a pilot, or considering joining the team, reach out directly.