Mythic

Smart security, drones and intelligent manufacturing robotics require one key factor: instantaneous computing for #AI inference. Since digital systems store billions (or trillions) of parameters separately, they are unable to compute in a timely and energy-efficient manner.   Analog solutions like Mythic’s can avoid memory bottlenecks, resulting in accurate and instantaneous high-resolution inference. Mythic achieves this by storing AI parameters directly in the processor. For more information, visit: https://bit.ly/3LH6EZg

The rapid adoption of AI is causing a significant amount of electricity usage. It’s reported that AI technologies consume approximately 7% of the world’s electricity. To meet AI demands in a sustainable manner, designers need alternative hardware solutions.   Check out this EE Times | Electronic Engineering Times article, which looks at how analog solutions like Mythic’s pave a new path for sustainable AI: https://bit.ly/3S1JsbK  

Our new CEO Taner Ozcelik was recognized by the Austin Business Journal as a “new name to know” in Austin's technology and startup ecosystem. Pulling from his storied career, Taner will lead Mythic in unleashing AI’s potential with new solutions that bring order-of-magnitude improvements in power, cost efficiency, and performance scalability.   To learn more about Taner’s extensive background and what’s next for Mythic, visit: https://bit.ly/3VR4MBO

We’re thrilled to announce veteran semiconductor executive Taner Ozcelik as our new CEO. Benefiting from his extensive experience at NVIDIA and On Semi, Mythic will fuel new markets for AI inference through our ultra-efficient analog compute-in-memory technology. Looking toward the future, our next-generation product will deliver significant improvements in power and cost efficiency, achieving unprecedented levels of performance scalability. To learn more about Ozcelik’s storied career and his plans for leading Mythic, visit: https://bit.ly/4cEBDAF

AI and IoT expansion may be newer, but the computing accompanying these two industry segments has quite a history and distinctly different use cases. And despite us being surrounded by the impact of these emerging technologies in our daily lives, we rarely stop to discover how this technology works. Dave Fick outlines the key differentiators between the computing methods that make this functional – analog and digital – and how analog is re-emerging as a crucial need for real-time AI processing. For the full story, head over to Electronic Design: https://bit.ly/4bamtlK

Mythic’s Dave Fick recently joined Daniel Litwin of MarketScale to discuss the crucial role edge computing plays in advancing Smart Cities.   The future of developed smart cities requires thousands of cameras, and connecting those cameras to one server is costly, requiring an enormous amount of bandwidth and halting growth. Instead, decision-makers can opt for local computing at the edge, which helps significantly reduce costs and latencies while improving scalability. This is especially true in security and surveillance applications that would benefit from AI inferencing. Check out the episode to hear more about smart city development: https://bit.ly/3KoE1iM

The AI hardware industry is moving fast, making it increasingly difficult for developers to future-proof AI-powered products. This presents an urgent need for scalable solutions. Mythic’s AMP provides powerful compute in small edge devices and can scale in volume to deliver server-class compute in data centers. Whether it’s one AMP or multiple, Mythic offers ultra HD resolution for inferencing with support for multiple neural network applications. To learn more, visit: https://bit.ly/3pecYzN

This week’s unveiling of OpenAI’s Chat GPT-4o and Google DeepMind’s Project Astra are incredibly exciting for the AI industry. However, smaller applications with limited bandwidth require power solutions designed for efficient memory and data flow. In-memory computing achieved using analog processors presents a unique opportunity, delivering real-time performance, high data throughput, and low latency. Ultimately, this accelerates the adoption of LLMs across a breadth of consumer and industrial devices.   We discuss this more in our blog about how Mythic tackles the challenge of LLM adoption in its day-to-day operations: https://bit.ly/47RRk5w

Unite.AI shared an insightful breakdown of the intersection between edge computing and AI, pointing to applications like autonomous drones as a key use case. Hardware that enables deep neural network processing is crucial for secure and accurate performance. https://bit.ly/4b9LsGj   At Mythic, our analog computing AI processors enable deep neural network processing directly within the device, making them ideal for airborne and ground-based drone models. Learn more at https://bit.ly/3XdaUVh.

While large language models (LLMs) are exciting and show great promise, they present a unique set of challenges, particularly around memory bandwidth. If not addressed, the power of LLMs within devices remain untapped. With Mythic’s unique in-memory computing, data is pinpointed to exactly when it’s needed, without a memory cache. The results are enhanced AI inferencing at up to 25 TOPS.   Read here for more on how Mythic plans to harness LLMS: https://bit.ly/47RRk5w