SiMa.ai Expands Edge AI Market with New MLSoC Modalix Chip
SiMa.ai, an edge computer chip and software startup, has recently raised $70 million from notable investors including Dell Technologies Capital. The company is enhancing its presence in the edge AI sector with the launch of the MLSoC Modalix chip, which is smaller and more energy-efficient than its predecessor.
Revolutionary Chip Design
Engineered at 6 nanometers, the Modalix chip is significantly smaller than the previous 16-nanometer MLSoC model from the San Jose-based startup. This advancement empowers the Modalix platform to support advanced AI models—Convolutional Neural Networks (CNNs), Transformers, and Generative AI—while maintaining industry-leading energy efficiency and scalable performance.
CEO Krishna Rangasayee expressed enthusiasm during a video interview: “We're building on our ONE Platform for Edge AI, enhancing both our silicon capabilities and accompanying software.”
Diverse Applications
The Modalix family is tailored for a variety of applications, including industrial automation, healthcare, smart vision systems, and aerospace. Rangasayee noted, “Robotics, embodied AI, and sensory information are the future. Modalix is ideal for generative AI-centric architecture that fosters innovations like human-robot interactions.”
His vision extends further, predicting a future where “every appliance, every device, can communicate, express emotions, and visualize information.”
Advancing Edge AI Technology
Generative AI is rapidly influencing sectors but has historically been limited to desktop and mobile environments. SiMa.ai and competitors like Nvidia are pushing this technology into dedicated devices used in fields such as robotics and drones.
“We consistently outperform immediate competitors 10x in real-world applications, and with Modalix, our advantage extends even further,” Rangasayee claimed.
Multimodal Processing Capabilities
The MLSoC Modalix platform supports multimodal AI processing, integrating text, image, and audio inputs. It can run various models, including Meta’s Llama 2-7B parameter model, opening new possibilities for edge reasoning. Rangasayee stated, “The combination of audio, video, and text inputs represents a significant shift we’re addressing.”
This new family offers configurations ranging from 25 to 200 TOPS, specifically designed to manage demanding AI workloads while minimizing power consumption. “Modalix reflects the significant advancements of the past two years, now allowing you to run everything from CNNs to cutting-edge models on a single chip,” he added.
Addressing Edge Computing Challenges
SiMa.ai’s technology is crafted to tackle key hurdles in edge applications, particularly performance-per-watt. Rangasayee emphasized the importance of “frames per second per watt” and “inferences per second per watt,” asserting that Modalix pushes their lead further.
The goal is to free customers from concerns about power and cooling limitations. “With Modalix, low power and high performance are guaranteed, reshaping possibilities at the edge,” he concluded.
Industry Recognition
The capabilities of SiMa.ai's MLSoC Modalix are gaining traction among industry leaders. Arye Barnehama, CEO of Elementary, praised its energy efficiency and high performance, aligning with the needs of their vision inspection systems. Vaibhav Ghadiok, CTO of Hayden AI, also highlighted its relevance for multimodal AI in power-constrained edge devices.
User-Friendly Deployment
SiMa.ai further supports its platform through the Palette Edgematic software stack, a no-code, drag-and-drop solution aimed at simplifying AI deployment for non-specialist developers. The platform also features integrated Image Signal Processor (ISP) capabilities, PCIe Gen 5 support, and eight Arm Cortex-A65 CPUs, making it adaptable to a wide range of AI workloads.
Market Positioning and Future Outlook
The launch of the MLSoC Modalix family signifies SiMa.ai's ambition to compete with giants like Nvidia. While Nvidia dominates cloud-based AI, SiMa.ai specializes in real-time, on-device processing. Rangasayee noted that their chips surpass Nvidia’s in both performance and power efficiency for edge AI applications.
With predictions of the global edge computing market doubling in the coming years, driven by AI advancements and the demand for real-time decision-making, SiMa.ai's MLSoC Modalix is well-positioned to meet this growing need.
The introduction of the MLSoC Modalix family reinforces SiMa.ai's leadership in edge AI, combining high performance, energy efficiency, and simplified deployment—a compelling option for industries eager to leverage AI capabilities at the edge. Backed by investors like Dell Technologies Capital and a growing roster of industry partners, SiMa.ai is poised to lead the next wave of edge AI innovation.
Ranking
