Some companies chase horsepower. Mythic went hunting for wattage. On Dec 17, 2025, the Austin-born analog rebels closed a $125M Series D led by DCVC, with New Enterprise Associates, Atreides Management, Honda Motor Co., and Lockheed Martin stepping in with conviction. That pushes total capital raised north of $290M, not to flex valuation math, but to scale production of AI silicon that thinks more like a brain and less like a space heater. GPUs roar and gulp power. Mythic whispers, sips electrons, and still finishes the sentence.
The roots explain the posture. Founded in 2012 as Isocline out of the Univ of Michigan, Mythic came up through U.S. Air Force SBIR work solving compute vision for high altitude drones and GPS signal acquisition. Mike Henry and Dave Fick were not chasing hype cycles. They were chasing physics. Laura Fick’s PhD research became the core analog compute engine, storing neural network weights directly on flash transistors so computation happens inside memory. Less movement, less waste, more signal. This is compute-in-memory that treats energy like a scarce resource, not an afterthought.
The silicon backs it up. The M1076 Analog Matrix Processor delivers 25 TOPS at roughly 3 to 4W, holding up to 80M parameters on chip with deterministic execution that defense and automotive teams can actually trust. Full HD YOLOv5 runs in 33ms. A 16-chip PCIe configuration reaches 400 TOPS at 75W. That is not lab theater. That is production hardware already deployed with customers across the U.S., Europe, and APAC, spanning robotics, industrial vision, smart infrastructure, and mission critical defense workloads.
Leadership matters when the ambition is this technical. Taner Ozcelik stepped in as CEO in June 2024 after building NVIDIA’s automotive business and running sensing organizations where power budgets decide who ships. Dave Fick returned to the CTO role after navigating Mythic through its toughest financial stretch and refocusing the roadmap. This is a team that understands scars, yield curves, and what it takes to move from clever chip to repeatable manufacturing. DCVC’s Matt Ocko has been involved since 2016 because deep tech like this compounds quietly, then shows up everywhere.
Honda Motor Co. and Lockheed Martin are not tourists. They are strategic investors with real deployment paths into trillion-dollar automotive and defense markets that cannot afford inefficient inference. Mythic’s analog compute-in-memory platform is built for edge and server AI where performance per watt is the deciding factor. With the M2000 series on deck, the company is scaling an advantage rooted in math, materials, and restraint. Mythic does not need to shout. The numbers already carry.
Startups Startup Funding Venture Capital Series D AI Semiconductor GPU Hardware Deep Tech Infrastructure Technology Innovation Tech Ecosystem Startup Ecosystem Hiring Tech Hiring DCTalks
