Bloom Raises $3.6M to Build an AI Marketplace for U.S. Manufacturing

Detroit-based Bloom has raised $3.6 million in seed funding to build a marketplace for U.S. manufacturing capacity.

The round was led by SNAK Venture Partners, a firm focused on marketplaces, with participation from Flyover Capital and Mana Ventures, alongside Detroit Venture Partners, Invest Detroit Ventures and Michigan Outdoor Innovation Fund TechCrunch. The raise was reported on October 7, 2026.

Bloom was co-founded in 2023 by Justin Kosmides, who serves as chief executive. The company is based in Detroit.

Its current product is a marketplace that connects buyers and sellers of manufacturing capacity. Customers use supply-chain AI agents, software helpers that search and shortlist suppliers, to find suppliers, parts, and manufacturing and engineering services. Bloom does not handle logistics or production itself. The agents handle supplier discovery and matching.

To date the company reports over 2,000 matches for more than 140 companies. Memberships have grown fivefold, with low churn.

That is a shift from where Bloom started. It originally handled logistics, manufacturing and supply chain work for e-bike and e-scooter companies, working as a service provider inside the mobility supply chain. It has since moved to a marketplace model where it does not take on that operational work directly.

Kosmides describes the ambition as an AI-driven version of Alibaba's marketplace for contract manufacturers in China, applied to American manufacturing. In practice that means a two-sided discovery service for domestic contract manufacturing, components and engineering services.

The expansion beyond micromobility came from inbound demand. Robotics startups and drone makers seeking domestic supply chains approached the company beyond the original e-bike and e-scooter base. Bloom widened its scope to help drone and robotics companies find U.S.-based manufacturers and shippers. By May 2026, the company said it had generated as much revenue in five months as in all of 2025 TechCrunch, alongside the fivefold rise in memberships.

The broader context here is that supplier discovery for low-volume, high-mix hardware, or small runs of many different parts, remains fragmented. Procurement teams rely on referrals, trade shows and manual requests for quotes. Contract manufacturers face idle capacity and unqualified inquiries. A marketplace that organizes capability data and routes demand to the right shops addresses that bottleneck.

In my view, the technical question is how much of that matching can be left to agents without eroding trust. Supplier selection turns on tolerances, certifications, lead times and yield history, details that are often poorly structured and settled offline. Automated matching holds only if supplier profiles stay current and the agent can check constraints before introducing the parties. Match volume matters less than how many matches turn into purchase orders and repeat use.

In this author's view, Bloom's other bet is to stay broad on categories but narrow on geography. It assumes domestic sourcing is a strong enough filter to create liquidity across mobility, robotics and drones, which avoids the slow start of a single-industry exchange. That choice puts pressure on search and classification, since a CNC shop, a battery pack assembler and a freight forwarder do not describe capacity in the same terms. The seed funding gives the company room to test that idea, and if it can keep churn low while growing the supplier side, it could become routine infrastructure for hardware teams that want to source domestically without building a supplier network from scratch.