Mythic Acquires Videantis, One of Europe’s Leading Digital Processor IP Companies, to Build the World’s Most Energy-Efficient AI Compute Platform

Mythic, a pioneer in analog compute-in-memory and architect of the industry’s most energy-efficient AI acceleration technology, today announced it has acquired Videantis GmbH, one of Europe’s leading digital processor IP companies.

This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20260519255958/en/

The transaction unites Mythic’s breakthrough analog compute platform with Videantis’ highly differentiated, unified digital processor architecture and production-proven software stack — accelerating Mythic’s delivery of a new class of hybrid AI compute platform with a 100x energy efficiency advantage over conventional GPU-based systems.

This deal builds on Mythic’s recently announced agreement with Honda to co-develop next-generation AI chips for future vehicles — a validation of Mythic’s game-changing architecture in one of the world’s most demanding production environments. Together, the Honda collaboration and the Videantis acquisition position Mythic to scale its hybrid AI computing on a single platform – from single camera drones, to factory robots, to autonomous platforms, all the way to data center deployments.

This progress follows on the heels of Mythic’s oversubscribed $125M funding round led by prominent deep-tech investors including DCVC, NEA, Future Ventures, S3 Ventures, Atreides Management, and SoftBank. eCAPITAL, a prominent deep tech VC in Europe has also become a shareholder in Mythic as part of the acquisition.

“AI’s present energy consumption is ruinously wasteful, untenable, and destabilizing,” said Dr. Taner Ozcelik, CEO of Mythic. “With the addition of the amazing team at Videantis, Mythic will provide the world a top-performing AI chip that uses one percent of the energy used by today’s state-of-the art GPUs. We are manufacturing this chip in the US and Europe – providing our customers safety against geopolitical instability. We seek to ensure that the future of AI is sustainable, stable, and abundant.”

With this transaction, Mythic significantly strengthens its position across data center, automotive, robotics, and defense markets, and establishes a powerful transatlantic AI semiconductorplatform rooted in both Silicon Valley and Germany. Europe’s AI future is assured and America’s AI future is turbo-charged.

A Rare European Technology Asset — Proven at Scale

Videantis is among Europe’s most distinguished semiconductor innovators. The company was recently selected in the top 1% of companies by the European Innovation Council (EIC), earning one of the continent’s most competitive innovation grants.

Its processor IP has powered more than 25 million chips shipped worldwide, with production programs continuing to scale. The company maintains deep penetration across all top 3 European automotive manufacturers and has enabled compact, safety-critical systems such as Automatic Emergency Braking (AEB) deployed inside ultra-compact 20mm × 20mm camera modules.

Critically, Videantis has achieved zero field defects across its deployed production base — an extraordinary accomplishment in automotive-grade silicon and a testament to the robustness of its architecture and engineering discipline.

Videantis’ unified processors are shipped in increasing, major volumes today, inside chips from a top three global semiconductor company in market capitalization, whose silicon powers large portions of the vehicle backbone — reflecting the rare transition of European processor IP from innovation to industry-standard deployment.

Videantis’ v-MP6000UDX unified processor platform runs deep learning, classical computer vision, signal and image processing, video encode and decode, and AI inference workloads — all on one architecture. Rather than stitching together specialized blocks such as NPUs, GPUs, Streaming Machines, Tensor Cores, DSPs, ISPs, and codecs, Videantis employs a scalable array of identical processing cores, delivering high utilization at low power for real-time workloads.

“For more than a decade, we have continuously refined a unified architecture designed for scalability, determinism, and efficiency,” said Dr. Hans-Joachim Stolberg, Founder and CEO of Videantis. “We’ve always believed energy efficiency comes from controlling data movement — explicit memory management beats coherence traffic. Mythic’s compute-in-memory approach applies the same philosophy to the extreme limit. Together we can deliver higher system-level efficiency than either of us alone.”

Videantis was founded by five industry veterans and is led by Hans-Joachim Stolberg. All five founders will join Mythic and continue to drive the evolution of the unified digital architecture and build the new hybrid computer. Videantis will continue to operate as a wholly owned subsidiary of Mythic.

Unified Architecture: A Fundamentally Different Approach from GPUs

Unlike GPU-centric systems built from multiple heterogeneous processing blocks, Videantis employs a unified hardware and software architecture designed for determinism, scalability, and efficiency.

Unified Hardware

While modern GPUs integrate multiple heterogeneous processing engines and subsystems, Videantis’ architecture is unified from the ground up. Its unique combination of:

• VLIW (Very Long Instruction Word)
• SIMD (Single Instruction Multiple Data)

creates a scalable, deterministic compute engine capable of handling vision, signal processing, SLAM, Large Language Models (LLMs), non-max suppression, attention calculations, video processing, AI inference, and other high-performance workloads within a single coherent framework.

A defining characteristic of the architecture is its extraordinary runtime adaptability. The same cluster of unified cores can be dynamically partitioned and adapted to support vastly different workload mixes depending on the application workload. This flexibility is a key reason for Videantis’ incredible success in automotive programs, where requirements evolve across model generations and feature sets. The architecture effectively delivers FPGA-like reconfigurability with ASIC-level efficiency — a rare combination that enables long product lifecycles and diverse application workloads without sacrificing performance per watt.

Instead of relying on increasingly complex memory coherence mechanisms and high-bandwidth interconnect fabrics, Videantis minimizes data movement through intelligent, explicit memory management. This architectural purity avoids the inefficiency of “hodgepodge” processor integration common in GPU-centric designs and enables exceptional performance per watt and scalability.

On its own, Videantis’ digital processor architecture delivers highly compelling advantages over traditional GPU and TPU approaches in power-constrained and safety-critical systems.

Unified Software

The unified hardware is matched by a decade-hardened unified software stack, perfected throughout more than a decade of numerous production automotive programs.

Much like NVIDIA’s Unified Driver Architecture (UDA) and CUDA ecosystem were foundational to GPU adoption, Videantis has built a cohesive compute and software architecture that simplifies deployment and scales across applications — purpose-built for deterministic, safety-certified, and power-efficient AI systems.

This software maturity dramatically reduces integration friction and accelerates system-level deployment — a critical advantage in automotive, robotics, defense, and data center environments.

From Analog + Digital to a New Class of Hybrid AI Computer

Mythic has consistently positioned itself around a bold mission: achieving up to 100x improvement in energy efficiency versus traditional AI acceleration platforms.

The addition of Videantis accelerates this mission.

Both companies share a common architectural philosophy: energy efficiency comes from keeping data local and minimizing data movement. Videantis achieves this through explicit memory control and intelligent scheduling across unified cores. Mythic pushes this principle further by collapsing compute and memory into a single physical plane through analog compute-in-memory.

Mythic’s analog in-memory compute architecture delivers unmatched efficiency for dense AI inference. Videantis’ unified digital architecture provides the flexible, scalable, reconfigurable, high-performance digital backbone required to orchestrate and expand that compute.

Together, they form a true hybrid compute platform:

• Analog for ultra-efficient AI matrix operations
• Unified digital for control, programmability, and high-performance workloads

“We’ve always envisioned Mythic as a hybrid system, combining the extreme efficiency and performance of analog computing with the programmability and flexibility of digital computing,” said Dr. Dave Fick, Co-Founder and CTO of Mythic. “Videantis adds a key piece of the puzzle, providing flexible computing — handling not only digital operations for attention calculations, but also non-max suppression, video codecs, SLAM, and many other high-performance compute algorithms.”

This hybrid model contrasts sharply with GPU-centric systems, which attempt to solve all workloads with increasingly power-hungry digital architectures that rely on extensive memory bandwidth utilization.

By combining best-in-class analog and best-in-class digital architectures, Mythic effectively doubles its architectural advantage — accelerating time to market for the most energy-efficient AI computing platform in the world.

Accelerating Data Center Disruption

AI data centers are facing an unprecedented energy crisis. Power — not raw throughput — is becoming the primary constraint.

This acquisition significantly accelerates Mythic’s entry into the data center market by expediting development of the digital half of its hybrid architecture and enabling delivery of more complete, deployable AI systems with end-to-end efficiency and performance.

The result is a compute platform designed not merely to compete with GPUs and TPUs, but to redefine the performance-per-watt envelope for AI industry.

Mythic’s hybrid architecture aims to enable data center operators to deliver more AI capability per watt, per rack, and per dollar — addressing one of the most urgent structural challenges in AI infrastructure.

A European AI Champion

Europe now has its own AI semiconductor champion — one with deep roots in German engineering excellence and the ambition to compete globally.

With a strong base in Germany and the US, Mythic now represents a transatlantic AI computing platform uniquely positioned to serve Europe, the United States, and global markets.

Dr. Taner Ozcelik, CEO of Mythic, and ex-NVIDIA executive, said:

“Videantis is the best digital architecture we evaluated worldwide. Its unified hardware and software model is elegant, scalable, extraordinarily flexible, and production-proven with zero defects in the field. When combined with Mythic’s analog compute platform, we effectively double our architectural advantage and accelerate the development of the hybrid computer the world needs for the AI era.

This acquisition brings together Silicon Valley innovation and German engineering excellence to build the most energy efficient compute platform on the planet.”

About Mythic

Mythic is a pioneer in analog AI computing, ushering in a new era of accelerated computing with its Analog Processing Units (APUs) – a fundamentally different compute-in-memory approach that collapses computation and memory into a single plane. By overcoming the energy bottlenecks of traditional digital architectures, Mythic’s chips deliver unparalleled performance-per-watt (often 100x better than GPUs) for AI inference across edge devices, vehicles, and data centers. Mythic’s platform supports a wide range of AI models, and its Starlight analog co-processors can be embedded in sensors to dramatically enhance data quality. The company’s investors include DCVC, NEA, Future Ventures, Atreides Management, SoftBank, S3 Ventures, eCAPITAL and strategic partners such as Honda and Lockheed Martin. Learn more at mythic.ai.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260519255958/en/