Portland-based semiconductor systems company Mueon has raised $65 million in Series A funding to develop a three-dimensional computing architecture designed for AI data centers.
Maverick Silicon led the round, with participation from Intel Capital, Geodesic Alliances Fund, Oregon Venture Fund, M-Ventures, Wellington Capital, IQT and Defined. The financing follows a $15.5 million seed round in 2025, bringing total capital raised to more than $80 million.
The company was founded by former Intel engineers Wilfred Gomes, its CEO, and Jack Hwang, its CTO. Their experience spans advanced semiconductor packaging, silicon manufacturing and system architecture.
Mueon's core technology, Cubelets, is a modular architecture that integrates compute, memory, interconnects, power delivery and thermal management into stackable three-dimensional systems.
Traditional server designs rely on components distributed across boards and racks. As AI workloads expand, moving data between processors and memory consumes substantial energy and introduces performance limitations. Mueon aims to shorten those distances and increase the computing capacity that can fit within a given physical footprint.
The company is targeting up to tenfold improvements in power and cost efficiency within the same rack footprint. That figure is an architectural target, not a demonstrated production result.
The Series A will fund engineering, software integration, system validation and the transition toward production readiness. Mueon is also working with industry partners to validate its approach, with companies including d-Matrix and Ampere Computing publicly discussing its potential.
The market is attractive because AI infrastructure is increasingly constrained by available power, cooling and data movement, not simply processor performance.
Mueon's challenge is that advanced packaging and three-dimensional integration create difficult manufacturing and thermal problems. Higher density can improve efficiency, but it also concentrates heat and complicates servicing.
The company must prove that its architecture delivers meaningful system-level improvements under real AI workloads and can be manufactured at a competitive cost. Until then, its projected efficiency gains remain an engineering objective rather than established economics.