Mythic
Mythic is an Austin-based semiconductor company founded in 2012 by Mike Henry and Dave Fick as a spinout from the University of Michigan, originally called Isocline.
Profile
Mythic designs energy-efficient AI inference chips using analog compute-in-memory architecture, competing with Nvidia on power efficiency rather than raw performance.
Mythic is an Austin-based semiconductor company founded in 2012 by Mike Henry and Dave Fick as a spinout from the University of Michigan, originally called Isocline. The startup specializes in analog compute-in-memory (CIM) processors designed to deliver AI inference with dramatically lower power consumption than conventional GPUs. The company faced near-collapse in late 2022 when it ran out of cash, but pivoted under co-founder Fick's leadership as CEO in March 2023 and secured a $13 million survival round.
Mythic subsequently raised $125 million in an oversubscribed Series D led by DCVC in December 2025, backed by strategic investors Honda, Lockheed Martin, SoftBank, and others—bringing total funding to approximately $290 million. The company reported $6.4 million in revenue as of mid-2025. Under CEO Taner Ozcelik, a Nvidia veteran who joined in June 2024, Mythic repositioned from building all components in-house to using mature semiconductor process nodes and partner relationships.
In May 2026, Mythic acquired Videantis, a European digital processor IP company with over 25 million chips deployed, to build a hybrid analog-digital platform spanning edge devices, robotics, automotive, and data centers. Mythic's Analog Processing Units (APUs) claim 100-fold efficiency over standard GPUs, targeting defense, automotive (with Honda licensing for next-gen vehicles), robotics, and large language model inference.
Who buys this
- Defense and military applications (validated by U.S. Department of Defense)
- Automotive OEMs and Tier-1 suppliers (co-development with Honda for software-defined vehicles)
- Robotics and autonomous systems (drones, AMRs, industrial robots)
- Edge IoT and surveillance (computer vision, object detection)
- Data center inference (trillion-parameter LLMs)
Publicly disclosed clients
- Honda Motor Co. (joint development of automotive-grade SOC)
- Lockheed Martin (strategic investor, defense applications)
- U.S. Department of Defense
Strengths and what to watch
Strengths
- Acquired Videantis in May 2026, gaining proven digital processor IP deployed in 25+ million chips and deep relationships with European automakers, creating a rare analog-digital hybrid platform
- Strong investor syndicate including Honda and Lockheed Martin signals real automotive and defense traction beyond venture-only backing
- Emerged from near-bankruptcy (late 2022) with restructured operations and now generating revenue; $125M Series D in Dec 2025 provides runway without imminent funding pressure
Watch for
- Chip production and customer acquisition still ramping: $6.4M revenue in 2025 is small relative to $290M raised; execution risk on automotive SoCs planned for late 2020s/early 2030s remains high
- CEO transitions and leadership continuity: founder Fick moved to CTO in June 2024 when Nvidia veteran Ozcelik took over; successful integration of new leadership and Videantis acquisition teams is unproven
- Competitive moat unproven in production: analog CIM architecture is novel but competing against entrenched GPU players (Nvidia) and rising analog-AI startups (Groq, SambaNova); claimed 100x efficiency gains need real-world validation at scale
Recent moves
Key Information
- Industry
- Edge AI
- Founded
- 2012
- Headquarters
- Austin, USA
Frequently Asked Questions
What is Mythic?
Mythic is an Austin-based semiconductor startup founded in 2012 that designs analog compute-in-memory chips for AI inference. The company specializes in processors consuming dramatically less power than conventional GPUs. It targets applications in defense, automotive, robotics, and data centers, claiming 100-fold efficiency improvements over standard AI accelerators.
What is analog compute-in-memory technology?
Analog compute-in-memory (CIM) is an architecture that performs AI inference directly within memory, eliminating data movement between processor and storage. This reduces power consumption dramatically compared to conventional GPUs, which fetch data separately. Mythic's CIM processors claim up to 100 times better power efficiency for AI inference workloads.
How does Mythic compare to Nvidia?
Mythic competes on power efficiency rather than raw performance. Its analog compute-in-memory architecture claims 100-fold efficiency gains over standard GPUs. While Nvidia dominates gaming and cloud training, Mythic targets edge inference, automotive, and defense applications where power consumption is critical and processing speed is secondary.
What is Mythic's partnership with Honda?
Honda and Mythic announced joint development of an energy-efficient analog AI chip for next-generation vehicles. The partnership targets software-defined vehicles with deployment expected in the late 2020s or early 2030s. Honda is also a strategic investor in Mythic's Series D funding round, backing the technology with capital and automotive expertise.
Why did Mythic acquire Videantis?
In May 2026, Mythic acquired Videantis, a European digital processor IP company with over 25 million chips deployed. The acquisition creates a hybrid analog-digital platform for edge devices, robotics, autonomous vehicles, and data centers, combining Mythic's analog efficiency with Videantis's proven digital technology and automaker relationships.
Why did Mythic nearly collapse in 2022?
Mythic ran out of cash in late 2022 despite raising over $165 million. The company lacked significant revenue as chips hadn't achieved market traction. In March 2023, Mythic secured a $13 million survival round, restructured operations, and pivoted toward defense and automotive markets where customer validation and demand were stronger.
How Mythic compares
Direct head-to-head against 2 competitors. Picked by 7wData.
Mythic
- Positioning
- Mythic designs energy-efficient AI inference chips using analog compute-in-memory architecture, competing with Nvidia on power efficiency rather than raw performance.
- Customer segments
- Defense and military applications (validated by U.S. Department of Defense)
- Strengths
- Acquired Videantis in May 2026, gaining proven digital processor IP deployed in 25+ million chips and deep relationships with European automakers, creating a rare analog-digital hybrid platform
- Watch for
- Chip production and customer acquisition still ramping: $6.4M revenue in 2025 is small relative to $290M raised; execution risk on automotive SoCs planned for late 2020s/early 2030s remains high
- Recent moves
- Mythic Acquires Videantis to Build the World's Most Energy-Efficient AI Compute Platform
Hailo
- Positioning
- Israeli edge AI inference chipmaker competing on power efficiency for cameras, automotive ADAS, and industrial robots.
- Customer segments
- Automotive OEMs, Tier-1 camera suppliers, industrial robotics makers, and defense integrators; hardware and firmware engineering buyers.
- Strengths
- Hailo-10H processor runs generative AI inference on camera and embedded hardware using structure-driven dataflow, minimizing on-chip data movement.
- Watch for
- Valuation dropped from $1.2B to under $500M by early 2026; January 2026 layoffs and SPAC liquidity urgency signal financial distress.
- Recent moves
- SPAC merger memorandum signed March 2026 to address urgent liquidity; Hailo-10H edge chip launched July 2025.
SambaNova Systems
- Positioning
- Enterprise and government inference hardware vendor claiming 3x lower cost than Nvidia B200 via purpose-built RDU processors.
- Customer segments
- Enterprise IT, cloud providers, sovereign AI programs, and US government agencies; AI infrastructure and data center ops buyers.
- Strengths
- SN50 RDU fits 10-trillion-parameter model inference in a single rack, removing multi-GPU networking overhead required by comparable GPU clusters.
- Watch for
- SN50 ships H2 2026; enterprises planning 2026 procurement must rely on vendor-stated benchmarks before hardware is in their hands.
- Recent moves
- SN50 RDU launched February 2026; Intel multi-year strategic partnership for agentic inference deployment announced April 2026.
Sources
- mythic.ai — Company overview, product descriptions (APUs, Starlight sensors), market segments (automotive, robotics, defense, data centers)
- www.businesswire.com — Videantis acquisition details, combined roadmap, Videantis technology and customer relationships (25M+ chips deployed, European automaker penetration)
- mythic.ai — Honda partnership, joint development of automotive SoC for software-defined vehicles, deployment timeline (late 2020s/early 2030s)
- www.dcvc.com — $125M Series D funding round, investor list (DCVC lead, NEA, Atreides, SoftBank, Honda, Lockheed Martin), technology claims and market applications
- techcrunch.com — November 2022 cash crisis, Dave Fick becoming CEO, $13M survival round in March 2023, strategic pivot to defense/government markets
- www.theregister.com — 2022 near-bankruptcy circumstances, company running out of runway before revenue, previous $165M+ raised, IP of Engineering Ty Garibay statement
- www.businesswire.com — Taner Ozcelik as CEO (appointed June 2024), Ozcelik background at Nvidia (automotive division SVP & GM), ON Semiconductor, Luminar, Dave Fick transition to CTO
- getlatka.com — $6.4M revenue in 2025, 58-person team headcount
- www.edge-ai-vision.com — Videantis acquisition strategic rationale, hybrid analog-digital architecture, Videantis proven software stack and automotive relationships, deployment roadmap (drones, robots, autonomous vehicles, data centers)