Portland-based Mueon, which is developing semiconductor-based systems for AI and hyperscale computing, has announced their launch from stealth with $15.5M in seed funding and the debut of Cubelets, stackable, modular components designed to redefine compute efficiency at scale. Mueon’s idea is to build data centre scale Infrastructure for next-gen AI Workloads.
Cubelets are more than a chip. They are a platform, offering a path to more sustainable and cost-effective data centre operations. They reduce the need for massive cooling infrastructure and enable linear scalability as demand grows.
“I’ve known Wilfred Gomes, Mueon’s co-founder and CEO, for over five years, said Srini Ananth, Managing Director at Intel Capital, the lead investor in the funding round. “As an Intel Fellow, Wilfred helped shape the future of microprocessor design through breakthroughs in 3D integration, Foveros packaging, and combining logic and memory for better performance and efficiency. His vision for Mueon reflects that same system-level thinking.
“When he approached us with the idea, it didn’t take long to decide to invest,” Ananth said. “Wilfred also teamed up with Jack Hwang, a former Intel Principal Engineer with over 30 years of experience in semiconductor process technology, including advanced materials, etching techniques, and scaling innovations for advanced logic nodes and next-generation chips. They have recruited a strong team of industry experts in Memory, Thermals and Power delivery to enable this vision.”
“Thrilled to announced Mueon Corporation seed funding by Intel Capital, Geodesic Capital, and Oregon Venture Fund,” Gomes said. “Grateful to my co-founder Jack Hwang and the entire team at Mueon Corporation. “We are honored to collaborate with, d-Matrix, Subramanian Iyer/UCLAand our Memory, Foundry, EDA, Equipment Supplier partners to push the envelope for next Gen Data Centres/AI.”
Hyperscale operators are forecast to invest more than $200 billion annually in new capacity by 2027, much of it constrained by power-hungry rack-based systems. Traditional architectures create hotspots, drive up costs, and restrict scalability, making the need for a new design clear.
Mueon addresses these challenges with Cubelets, compact building blocks that provide a platform to tightly integrate compute, memory, power delivery, and thermal management. The architecture enables up to 10x improvements in density, energy efficiency, and faster deployment, while remaining fully compatible with today’s AI and cloud software stacks. By enabling clusters that scale linearly with demand, Cubelets reduce upfront capital expenditures, simplify deployment, and minimize the need for massive cooling and power infrastructure.
“Mueon was founded to rethink the system building blocks of the data centre from the ground up,” said Gomes, co-founder and CEO of Mueon Corporation. “Cubelets give our partners a new unit of compute by bringing together compute, memory, and power delivery, thermals, delivering efficiency and cost savings that today’s data centers urgently need.”
Intel Capital led the round with participation from Geodesic Alliance Fund and Oregon Venture Fund. Intel Capital Managing Director Srini Ananth will join Mueon’s board of directors.
Hyperscale operators are forecast to invest more than $200 billion annually in new capacity by 2027, much of it constrained by power-hungry rack-based systems. Traditional architectures create hotspots, drive up costs, and restrict scalability, making the need for a new design clear.
Mueon addresses these challenges with Cubelets, compact building blocks that provide a platform to tightly integrate compute, memory, power delivery, and thermal management. The architecture enables up to 10x improvements in density, energy efficiency, and faster deployment, while remaining fully compatible with today’s AI and cloud software stacks. By enabling clusters that scale linearly with demand, Cubelets reduce upfront capital expenditures, simplify deployment, and minimize the need for massive cooling and power infrastructure.
“The Mueon team is reimagining data centre design at the fundamental level,” said Srini Ananth, Intel Capital Managing Director. “Their modular approach offers customers a practical path to scale AI systems with better economics and sustainability, while Mueon’s deep technical expertise and vision for data center efficiency could redefine how AI infrastructure is built globally.”
Mueon’s vision is to democratize high-performance computing by making infrastructure more flexible and sustainable. Its modular approach allows cloud providers, enterprises, and research institutions to deploy dense compute clusters in locations and environments that were previously cost- or power-prohibitive.
“Mueon’s vision for a “data centre in a box” is exactly that: a bold reimagining of how AI infrastructure can be built, scaled, and sustained,” Ananth stated. “We’re proud to support Mueon on this journey and excited to see how Cubelets will redefine the future of AI computing.”
To enable this vision, Mueon is closely working with industry, government, and academia partners to deliver all-in-one integration, with compute, memory, power, and thermals in a single modular unit. They also pledge up to 10x gains in density, energy savings, and deployment speed, and lower cooling and power demands, cutting both costs and environmental footprint
“Mueon’s Cubelets architecture has unlocked new possibilities for d-Matrix,” said Sudeep Bhoja, founder and CTO at d-Matrix. “Combined with d-Matrix’s digital in-memory compute technology, it’s helping us build a new class of smaller, denser infrastructure tailored for AI workloads that require ultra-low latency. This modular approach has allowed us to experiment faster, optimize performance, and deploy AI models at scales that were previously impractical.”
“For decades, large-scale data centers have been constrained by power and cooling limits,” said Subramanian Iyer, Distinguished Professor at UCLA and Director of UCLA CHIPS, former Director of the NAPMP, and Ex-IBM Fellow. “Mueon’s approach leverages the scale down and scale out capability of advanced packaging with innovative system architectures that can make the infamous ‘memory wall’ a relic of the past. This will result in a versatile HPC platform with a high degree of modularity and extensibility, and one where most of the energy is spent in compute rather than communication and power delivery inefficiencies.”
The funding will accelerate prototype development, system scaling, and top engineering hires. Mueon is initially focused on AI inference and training workloads that are constrained by today’s rack-based systems, with strong demand for deploying large models at scale at lower cost and power. The company is collaborating with leading fabrication, manufacturing, and foundry partners to validate Cubelets and bring the architecture to market.