Ambiq Expands AI Development Support with neuralSPOT SDK V1.2.0 Beta for Apollo510 and Apollo510B
Ambiq
September 23, 2025
neuralSPOT for the Apollo510 (4)
AUSTIN, Texas, Sept. 23, 2025 (GLOBE NEWSWIRE) -- Ambiq Micro, Inc. (NYSE: AMBQ), a technology leader in ultra-low-power semiconductor solutions for edge AI, today announced the availability of the latest release of the neuralSPOT SDK and Toolkit for the Apollo510 and Apollo510B, Ambiq’s flagship systems-on-chips (SoC) Family with the highest performance and the most power efficiency among its generations for advanced edge AI applications.
Feature highlights:
- Apollo510 Support and Enhancements – Unlocks the full potential of Ambiq’s award-winning SoCs with advanced AI performance characterization tools
- HeliaRT Beta Integration – Ambiq’s Apollo-specific AI runtime, delivering up to 3X faster inference and improved energy efficiency compared to LiteRT implementations
- Experimental HeliaAOT Integration – Ahead-of-time AI compiler for performance-driven deployment
- Rich Example Library – More than a dozen ready-to-deploy models for Apollo510, Apollo510B, Apollo4 Plus, Apollo4 Lite, and Apollo3 Plus
neuralSPOT – Accelerating Edge AI Development with Ready-to-Use AI Models
The neuralSPOT SDK and Toolkit bridges the gap between AI model creation, deployment, and optimization—empowering developers to move from concept to prototype in minutes, not days. neuralSPOT V1.2.0 Beta comes with example implementations of popular AI models including:
- Human activity recognition
- ECG monitoring
- Keyword spotting
- Speech enhancement
- Speaker identification
Whether you’re building next-generation AI models or showcasing innovative demos, neuralSPOT delivers the tools you need to succeed.
“With its AI-focused vector engine, the Apollo510 is a joy to develop for,” said Carlos Morales, VP of AI at Ambiq. “We continue to discover new ways to push the boundaries of performance and efficiency, and neuralSPOT is key to enabling those optimizations.”
Apollo510 SoC Family – Powering the Future of Edge AI
The Apollo510—winner of the Best Hardware Award at Embedded World 2024—sets the standard for performance and power efficiency in edge AI. It also drives heartKIT™ AI SDK, a collection of models for heart-vital monitoring and analytics, which earned the Best AI Solutions Award at Embedded World 2025 and is now available within the neuralSPOT SDK library.
Learn more about the latest updates in neuralSPOT V1.2.0 Beta and get started on developing edge AI applications with neuralSPOT today.
About Ambiq
Headquartered in Austin, Texas, Ambiq’s mission is to enable intelligence (artificial intelligence (AI) and beyond) everywhere by delivering the lowest power semiconductor solutions. Ambiq enables its customers to deliver AI compute at the edge where power consumption challenges are the most severe. Ambiq’s technology innovations, built on the patented and proprietary subthreshold power optimized technology (SPOT®), fundamentally deliver a multi-fold improvement in power consumption over traditional semiconductor designs. Ambiq has powered over 280 million devices to date. For more information, visit www.ambiq.com.
Copilot:
Ambiq has not publicly disclosed the exact semiconductor node sizes for the Apollo510 and Apollo510B SoCs. However, based on their use of Ambiq’s proprietary Subthreshold Power Optimized Technology (SPOT) and the Arm Cortex-M55 core running up to 250 MHz, it's likely that these chips are fabricated on a mature low-power node, possibly in the 40nm to 22nm range, optimized for ultra-low leakage and energy efficiency.
Ambiq’s previous Apollo families (Apollo3, Apollo4) were known to use TSMC’s 40nm ULP process, and unless there's been a major shift in foundry or process strategy, Apollo5-series chips like the 510/510B may continue on similar or slightly refined nodes. That said, the presence of advanced features like Helium vector extensions, Bluetooth 5.4, and 2.5D GPU acceleration suggests aggressive optimization rather than a leap to bleeding-edge nodes.
If you're benchmarking against other edge AI SoCs or evaluating for display backplane integration, I can help map out comparative process nodes and leakage profiles. Want to dig into that? |
