Synaptics Incorporated (Nasdaq: SYNA) introduced its Astra® SR200 Series on October 20, 2026, a new family of processors built around an AI engine that borrows its core intelligence from Google Research’s open-source Coral technology. The launch, first reported by VentureBeat, marks one of the clearest signals yet that Google wants its machine-learning silicon designs to spread across the industry rather than stay locked inside Pixel phones and Nest cameras.
The Astra SR200 Series lands at a moment when battery-powered AI wearables, smart cameras, and industrial sensors are multiplying faster than the chips built to run them efficiently. Synaptics is betting that a shared, open NPU core, rather than a dozen competing proprietary designs, is what finally makes on-device AI affordable at scale. Whether that bet pays off will depend on how many device makers actually adopt the Torq™ AI engine once review units ship.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Google and Synaptics Actually Announced
The Astra SR200 Series consists of three processors: the Astra SR210, the Astra SR230, and the Astra SR250. Each chip is built around what Synaptics calls the Torq AI engine, and each one embeds the Coral NPU from Google Research directly into the silicon. That detail is the real story here: Coral has existed for years as a Google-branded edge AI product line, but this is a case of Google’s NPU architecture itself, not just a Google-badged board, being licensed into a third party’s commercial chip family.
Each processor in the Astra SR200 Series targets a different tier of on-device intelligence. The SR210 delivers 200 GOPS of AI-processing performance and is aimed at the simplest sensing tasks. The SR230 steps up to 300 GOPS and adds support for vision and display workloads, making it a candidate for smart glasses or camera viewfinders that need to render UI alongside inference. The SR250 tops the lineup at 400 GOPS and is positioned for multimodal applications, the kind of workload that mixes audio, vision, and sensor fusion in a single pipeline.
All three chips pair Arm Cortex-M85 and Arm Cortex-M52 cores with the Coral NPU, a combination that keeps general-purpose control tasks on the Arm cores while routing matrix-heavy AI inference to the dedicated accelerator. That architecture is a familiar pattern in microcontroller-class AI silicon, but the Coral NPU component is what differentiates the Astra SR200 line from competing microcontroller families.
Astra SR200 Series Spec Comparison
The table below lays out how the three new processors compare on paper, based on the specifications Synaptics disclosed at launch.
| Model | AI Performance | Core Combination | Added Capability | Primary Target |
|---|---|---|---|---|
| Astra SR210 | 200 GOPS | Arm Cortex-M85 + Cortex-M52 + Coral NPU | Baseline sensing | Battery-powered sensing devices |
| Astra SR230 | 300 GOPS | Arm Cortex-M85 + Cortex-M52 + Coral NPU | Vision and display support | Connected cameras, streamers |
| Astra SR250 | 400 GOPS | Arm Cortex-M85 + Cortex-M52 + Coral NPU | Multimodal processing | AI wearables, industrial vision/voice |
Synaptics has not published pricing for any of the three parts, and the company’s announcement does not include a specific energy-efficiency figure for the new series. Readers should treat any percentage energy-savings claim circulating online as unconfirmed until Synaptics or Google publishes independent benchmark data.
The Coral NPU’s Road From Dev Board to Production Silicon
To understand why the Astra SR200 Series matters, it helps to trace how Google’s Coral NPU got here. Back on March 10, 2026, Google Research and Synaptics jointly announced a limited-edition Coral Dev Board built around the Astra SL2610 product line, as detailed in Synaptics’ official announcement. That board integrated a 1 TOPS Synaptics Torq NPU, giving developers their first hands-on access to Google’s open NPU design running on Synaptics hardware.
Seven months later, that developer-focused collaboration has turned into a shipping product line. Google describes the Coral NPU as open-source intellectual property, and the company’s own developer documentation, available through Google’s Coral announcements page, identifies the Synaptics Torq NPU subsystem as the first production implementation of Google’s open-source Coral NPU ML core. That distinction matters: a dev board is a prototype vehicle, but a production implementation inside a commercially available processor series is a market test of whether the open design can compete with closed, proprietary NPU architectures from larger chipmakers.
Google also used its broader October 2026 research retrospective to frame the Coral NPU as part of a larger push toward accessible, efficient AI hardware. In a post on the Research@ 2026 blog, the company laid out ten initiatives meant to expand who gets to build with AI, and open silicon IP for edge devices fits squarely into that narrative.
The Coralboard Platform Arrives Through DigiKey
Alongside the Astra SR200 announcement, Synaptics confirmed that its Coralboard™ platform became available through electronics distributor DigiKey as of October 20, 2026. The announcement accompanying that distribution deal does not list a price for the Coralboard, so any dollar figure attached to it in secondhand coverage should be treated with caution until Synaptics or DigiKey publishes an official list price.
Making the board available through a mainstream parts distributor, rather than keeping it behind a direct sales relationship, is itself a signal. DigiKey is where hobbyists, startups, and procurement teams at larger manufacturers go to source components in volumes ranging from one unit to thousands. That distribution choice suggests Synaptics wants the Coral NPU design to reach small embedded-systems teams, not just large OEM partners with existing Synaptics relationships.
Coral NPU Timeline: Dev Board to Commercial Chip Family
| Date | Milestone | Hardware | Status |
|---|---|---|---|
| March 10, 2026 | Coral Dev Board (next-gen) announced | Astra SL2610, 1 TOPS Torq NPU | Limited-edition developer hardware |
| 2026 (ongoing) | Coral NPU documented as open-source IP | Coral NPU ML core | Open specification, third-party implementation enabled |
| October 20, 2026 | Astra SR200 Series launch (SR210, SR230, SR250) | Torq AI engine + Coral NPU, Arm Cortex-M85/M52 | Commercial processor series |
| October 20, 2026 | Coralboard platform listed at DigiKey | Astra SL2610-based | Available through distributor, price unconfirmed |
That sequence, from a niche developer board in March to a three-chip commercial family in October, covers roughly seven months. For context, chip design cycles at this scale typically run 18 to 24 months from architectural spec to shipping silicon, which means the Astra SR200 Series was almost certainly already deep in development before the Coral Dev Board’s public unveiling. The dev board functioned more as a public preview of an NPU core that Synaptics’ engineering teams had been integrating for considerably longer.
Who Buys an Astra SR200: Target Applications Explained
Synaptics named five categories of target applications for the new series: AI wearables, battery-powered sensing devices, connected cameras, streamers, and industrial vision and voice systems. Each category stresses a different constraint, and the three-tier chip lineup appears designed to match those constraints rather than offer one undifferentiated part.
- AI wearables need the lowest possible power draw since they run on small batteries worn against skin for hours at a time; this is the segment Synaptics has separately flagged as facing new legal scrutiny, as covered in our report on AI wearables hitting wiretap law limits in over a dozen US states.
- Battery-powered sensing devices cover everything from environmental monitors to asset trackers, where the SR210’s 200 GOPS tier is the likely fit.
- Connected cameras and streamers require the vision and display pipeline that only the SR230 and SR250 support, since basic sensing silicon cannot decode and render video.
- Industrial vision and voice systems sit at the high end, where multimodal processing across audio and image inputs simultaneously points toward the SR250.
That segmentation strategy mirrors how Nvidia and Intel structure consumer chip tiers, scaling a single underlying architecture across a performance ladder rather than designing from scratch for each market. The difference here is that the ladder is built around a licensed, open-source accelerator core instead of a wholly proprietary one.
Why an Open-Source NPU Changes the Edge AI Chip Business
Most AI accelerators embedded in microcontrollers today are closed, vendor-specific designs. Qualcomm, Ambiq, and MediaTek each maintain proprietary NPU architectures that only they can modify, and licensing those designs typically means committing to a single vendor’s roadmap and tooling. Google’s decision to publish the Coral NPU core as open-source IP flips that model: any chipmaker with the engineering resources to integrate it can build a Coral-based processor, and Synaptics has now proven that’s possible at commercial scale with a three-part product family.
For Google, the upside is ecosystem gravity. If the Coral NPU becomes the common accelerator baseline across several chipmakers’ wearable and industrial silicon, Google’s own software stack, including whatever lightweight on-device models it optimizes for that core, gets a much larger addressable hardware base without Google having to fabricate a single chip itself. That is the same playbook Arm has run successfully for decades with its CPU core licensing business, just applied to an AI accelerator instead of a general-purpose processor.
For Synaptics, being first to ship a commercial Coral NPU implementation gives it a credibility edge over rival embedded-chip vendors who have not yet announced similar integrations. Being first also carries risk: if Google opens the same NPU license to additional chipmakers soon, as the open-source model implies it eventually will, Synaptics’ head start could shrink quickly once a second or third vendor ships competing Coral-based silicon.
Competitive Landscape: How the Astra SR200 Stacks Up
The broader AI chip market has spent much of 2026 focused on data-center silicon, from Nvidia’s push into workstation-class parts covered in our analysis of the RTX Spark undercutting pricier AI workstations to memory suppliers racing to keep up with demand, a trend visible in Micron’s reported 331% jump in HBM4 sales ahead of its October 21 webinar. The Astra SR200 Series sits at the opposite end of that spectrum entirely: it is edge silicon meant to run inside a wristband or a doorbell camera, not a data-center rack.
That distinction matters because the edge AI chip market has its own competitive dynamics, separate from the GPU shortage narrative dominating headlines elsewhere in semiconductors. Supply constraints on advanced memory, the kind reshaping deals like Samsung’s reported tie-up with TSMC’s 3nm process for Nvidia’s NVHBM memory chips, barely touch microcontroller-class silicon like the Astra SR200, which uses far simpler process nodes and much smaller die sizes. Synaptics is competing on integration and software ecosystem rather than on leading-edge fabrication access.
On the AI model side of the equation, the push toward open, efficient designs that the Coral NPU represents echoes a pattern already playing out in large language models. Nvidia’s own release of smaller open-weight models, detailed in our coverage of Nvidia’s three open Nemotron models spanning 4GB to 120B parameters, shows the same tension between proprietary and open approaches now reaching into silicon design itself, not just the models that run on top of it.
Historical Context: Coral’s Long Road at Google
Coral began life years ago as a Google hardware brand for edge TPU accelerators aimed at hobbyists and small-scale commercial deployments, sold through Google’s own channel as development boards and USB accelerators. The lineage from that original Coral TPU product to today’s Coral NPU core represents a shift in Google’s strategy: rather than continuing to sell its own branded edge AI boards indefinitely, Google has moved toward licensing the underlying accelerator architecture to established chipmakers who already have manufacturing relationships, device-maker customers, and distribution through partners like DigiKey.
That evolution tracks a broader pattern across the semiconductor industry in 2026, where chip earnings and AI infrastructure spending have become the dominant storyline, a dynamic our newsroom has tracked in pieces like Intel’s claim of unique 300mm InP silicon photonics capability at scale. Large chipmakers are increasingly choosing to open parts of their technology stack rather than guard every layer, calculating that ecosystem adoption creates more long-term value than exclusivity.
Market Impact: What This Means for Device Makers
For companies building AI wearables, smart cameras, or industrial sensing products, the Astra SR200 Series offers a new option that did not exist a week ago: a tiered chip family with a shared, Google-backed AI core and a documented upgrade path from the simplest SR210 configuration up to the multimodal SR250. That tiering reduces the engineering risk of locking into a single chip’s capabilities too early in a product’s design cycle, since firmware and software built for the Torq AI engine should, in principle, port across the three parts with less rework than switching between entirely different vendor architectures.
The lack of published pricing is the biggest open question for product teams evaluating the series today. Component cost drives bill-of-materials decisions more than almost any other factor in consumer wearables, where margins are thin and volumes need to be large to justify a new design. Until Synaptics publishes per-unit pricing, likely tied to order volume as is standard in embedded silicon sales, device makers cannot yet model whether the Astra SR200 undercuts or exceeds the cost of existing microcontroller-plus-NPU combinations from Qualcomm or MediaTek.
There is also a software-ecosystem question. An open-source NPU core is only as useful as the tooling and model library built around it. Google’s own documentation efforts, tracked through its Coral announcements channel, will need to keep pace with Synaptics’ hardware releases, or developers evaluating the Astra SR200 Series could find themselves with capable silicon and an immature software stack, a mismatch that has slowed adoption of other promising edge AI chips in the past.
What Reports Say About the Road Ahead
Coverage leading up to the launch had been tracking the announcement for days. A piece from R&D World noted that Google had been teasing the Synaptics edge AI chip ahead of its October 20 launch date, suggesting both companies coordinated a staged rollout of information rather than a single surprise announcement. That kind of lead-up is typical for B2B silicon launches, where the customer base, device OEMs and industrial integrators, needs lead time to evaluate specs before committing design resources.
Reports also describe the Astra SR200 line as part of a broader low-power edge AI push, though specific efficiency percentages attached to that framing in some outlets remain unconfirmed by Synaptics’ own materials. Readers evaluating the chips for a real design should wait for datasheet-level power figures rather than relying on preview coverage.
Predictions: Where This Goes Next
Based on the pattern Google and Synaptics have followed so far, from dev board in March to shipping chip family in October, a few outcomes look likely over the next 12 to 18 months.
- A second licensee emerges. If the Coral NPU is genuinely open IP, expect at least one more established microcontroller vendor to announce its own Coral-based silicon within a year, testing whether Synaptics’ first-mover advantage holds.
- Pricing arrives alongside design wins, not before. Synaptics will likely disclose Astra SR200 pricing only once it can pair the announcement with named device-maker customers, a common tactic to frame cost in the context of real adoption.
- Software tooling becomes the bottleneck. The success of the Torq AI engine will hinge less on raw GOPS numbers and more on how quickly Google and Synaptics ship model-conversion tools, SDKs, and reference firmware that make the Coral NPU easy to target.
- Wearable makers face the regulatory question in parallel. Any AI wearable built on this silicon will still need to navigate the wiretap-law scrutiny already facing the category, meaning hardware capability alone will not resolve the legal friction some manufacturers are encountering.
- Coral’s TPU-era branding fades further. Expect Google to continue referring to the technology primarily as the “Coral NPU” rather than reviving Coral’s original edge TPU identity, completing a years-long rebrand of the product line.
The Bigger Picture for Edge AI Silicon
The Astra SR200 Series is a small announcement in absolute dollar terms compared to the hundreds of billions of dollars flowing into data-center AI infrastructure elsewhere in the industry this year. But it addresses a different, equally important question: how does on-device AI actually reach the wearables, cameras, and industrial sensors that most consumers and businesses will interact with daily, as opposed to the cloud-based models running in hyperscale data centers.
Google’s choice to open-source the Coral NPU core rather than keep it proprietary suggests the company has concluded that owning the software and model layer matters more than owning the silicon itself. If that bet proves correct, the Astra SR200 Series will be remembered as the first proof point in a much longer list of third-party chips built on Google’s open NPU design. If adoption stalls, it may instead be remembered as an interesting experiment that never scaled past its first commercial partner.
Frequently Asked Questions
What is the Synaptics Astra SR200 Series?
It is a new family of three AI processors, the SR210, SR230, and SR250, announced by Synaptics Incorporated on October 20, 2026. Each chip combines Arm Cortex-M85 and Cortex-M52 cores with Google Research’s Coral NPU inside Synaptics’ Torq AI engine.
What is the Coral NPU?
The Coral NPU is an AI accelerator core that Google Research describes as open-source intellectual property. Synaptics’ Torq NPU subsystem is identified by Google’s developer materials as the first production implementation of that open-source design.
How much AI performance do the new chips offer?
The Astra SR210 delivers 200 GOPS, the SR230 delivers 300 GOPS and adds vision and display support, and the SR250 delivers 400 GOPS with support for multimodal applications.
What devices will use the Astra SR200 Series?
Synaptics has named AI wearables, battery-powered sensing devices, connected cameras, streamers, and industrial vision and voice systems as target applications.
How much does the Astra SR200 Series cost?
Synaptics has not published pricing for the SR210, SR230, or SR250 as of the October 20, 2026 announcement.
What is the Coralboard?
The Coralboard is a Synaptics development platform that became available through distributor DigiKey as of October 20, 2026. No price has been disclosed for it.
Is this the same as the Coral Dev Board announced earlier in 2026?
No. The Coral Dev Board, announced on March 10, 2026 and built around the Astra SL2610 line with a 1 TOPS Torq NPU, was a limited-edition developer board. The Astra SR200 Series is a separate, newer commercial processor family announced in October.
Are the energy-savings claims about these chips confirmed?
No. Specific efficiency figures, including any claim of a fixed percentage reduction in energy use, have not been confirmed in Synaptics’ own announcement materials and should be treated as unverified until an official datasheet is published.
