A Portland semiconductor startup led by a former Intel Fellow has closed a $65 million Series A round to commercialize what it calls Cubelets, stackable computing modules designed to squeeze far more processing power into the same data center footprint. Mueon announced Tuesday that Maverick Silicon led the round, with participation from Intel Capital, In-Q-Tel, and six other investors.
The financing brings Mueon's total raised to more than $80 million since the company emerged from stealth in September 2025. That's a brisk pace for a hardware startup tackling one of the industry's thorniest problems: how to cram ever more compute into facilities already straining under electrical and cooling constraints.
Mueon's pitch centers on modularity. Rather than building systems around a single accelerator chip, the company integrates compute, memory, interconnect, power delivery and thermal management into three-dimensional blocks that fit inside standard data center racks. The approach, Mueon said, can deliver up to tenfold gains in power and cost efficiency compared to conventional designs.
From Intel's labs to a bet on composability
Wilfred Gomes spent three decades at Intel before co-founding Mueon with Jack Hwang, another former Intel technologist. During his tenure, Gomes rose to Intel Fellow and led development of Foveros, the chipmaker's 3D packaging technology. His fingerprints are on microprocessor designs including Meteor Lake and Lakefield, according to Intel publications and conference presentations.
Now Gomes is betting that data centers need more than incremental improvements in chip design. "We envision composable data centers, where components can be mixed and matched at the data center level," he told The Information, which reported Tuesday that Mueon has not disclosed revenue or valuation.
The startup allows customers to supply their own AI chips for assembly into Cubelets modules, The Information said. That flexibility may prove crucial as hyperscalers and enterprise buyers experiment with different silicon architectures, from GPUs to custom ASICs.
Lowering voltage, raising density
Mueon's architecture relies on lower-voltage power delivery and two-sided cooling to pack components closer together. The company treats "packaging, interconnect, power, and thermals as first-class design variables," according to its announcement, rather than optimizing around any single component.
The target: megawatt-class density, a benchmark tied to surging electricity consumption. The International Energy Agency projected in a 2025 report that global data center power demand will roughly double to approximately 950 terawatt-hours by 2030, climbing to about 3% of worldwide electricity use from 1.5% in 2024, according to an Axios analysis published in August 2026.
Whether Mueon's approach can deliver on those efficiency claims at scale remains to be seen. Hardware startups often struggle to translate prototype performance into mass production, particularly when integrating technologies across multiple domains.
Intelligence money and industry backing
Intel Capital, which led Mueon's $15.5 million seed round in September, continued its support in the Series A. Srini Ananth, a managing director at Intel Capital, joined Mueon's board during the seed round and remains on it. Manish Muthal, senior managing director at Maverick Silicon, joined alongside him in the new round.
In-Q-Tel, the venture arm of U.S. intelligence agencies, appeared on the cap table for the first time, according to CB Insights. Other new investors include M-Ventures and Wellington Capital, alongside Oregon Venture Fund, Geodesic Alliances Fund and Defined.
The startup also secured public endorsements from potential customers and partners. Sudeep Bhoja, co-founder and CTO of d-Matrix, a digital in-memory compute startup that raised $275 million at a $2 billion valuation in November, said Mueon's architecture "creates new possibilities for partners such as d-Matrix."
Rohit Vidwans, co-CEO of Ampere Computing, noted in the same release that "large-scale data centers have been constrained by power and cooling limits for years." Those constraints have only intensified as AI workloads proliferate.
Hiring up for production push
Mueon plans to use the capital to expand engineering and operations teams, advance from platform development into system validation, and prepare for production. LinkedIn shows the company currently employs between 11 and 50 people in the Portland metro area, a modest headcount that will need to grow substantially if Mueon reaches manufacturing scale.
"This financing gives us the scale to bring the next generation of compute and memory system-scale integration to market," Gomes said in the announcement. Muthal of Maverick Silicon pointed to the team's combined expertise: "They combine deep knowledge in the technologies that define modern data centers with the experience to deploy those technologies at system scale."
The funding arrives as venture investors show renewed appetite for infrastructure hardware, particularly companies promising to ease data center bottlenecks. Whether Mueon can execute on a vision that requires coordinating silicon packaging, power management and thermal engineering in a single modular platform will determine if this round proves prescient or premature.
