<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>hw.dev Signal</title>
    <link>https://hw.dev</link>
    <description>Short, high-signal posts on meaningful developments in hardware development. What happened, why it matters, what it enables, operator read.</description>
    <language>en-us</language>
    <lastBuildDate>Thu, 30 Apr 2026 12:04:39 GMT</lastBuildDate>
    <atom:link href="https://hw.dev/signal/feed.xml" rel="self" type="application/rss+xml" />
    <generator>hw.dev static export</generator>
  <item>
    <title>Agileo&apos;s Agil&apos;EDA Targets SEMI Freeze 3 Upgrade Ahead of Mid-2026 Deadline</title>
    <link>https://hw.dev/signal/agileo-semi-eda-freeze3</link>
    <guid isPermaLink="true">https://hw.dev/signal/agileo-semi-eda-freeze3</guid>
    <pubDate>Thu, 30 Apr 2026 12:00:00 GMT</pubDate>
    <description>Agileo Automation launches Agil&apos;EDA, a software implementation of the SEMI Equipment Data Acquisition standard that decouples high-frequency data collection from equipment control and is already architected for the Freeze 3 gRPC upgrade due mid-2026.</description>
    <category>signal</category>
    <category>eda</category>
    <category>manufacturing</category>
    <category>tools</category>
  </item>
  <item>
    <title>AI Demand Forces ASML and Aixtron to Raise Outlooks as Equipment Lead Times Stretch</title>
    <link>https://hw.dev/signal/asml-aixtron-fab-capacity-expansion-2026</link>
    <guid isPermaLink="true">https://hw.dev/signal/asml-aixtron-fab-capacity-expansion-2026</guid>
    <pubDate>Thu, 30 Apr 2026 12:00:00 GMT</pubDate>
    <description>ASML and Aixtron both raised 2026 revenue outlooks this April as AI-driven orders for EUV lithography, SiC, and GaN deposition equipment push wafer fab tool lead times past 12 months.</description>
    <category>signal</category>
    <category>semiconductor</category>
    <category>manufacturing</category>
    <category>ai-hardware</category>
  </item>
  <item>
    <title>Tenstorrent Galaxy Blackhole: Integrated Inference at 23 PFLOPS</title>
    <link>https://hw.dev/signal/tenstorrent-galaxy-blackhole-servers</link>
    <guid isPermaLink="true">https://hw.dev/signal/tenstorrent-galaxy-blackhole-servers</guid>
    <pubDate>Thu, 30 Apr 2026 12:00:00 GMT</pubDate>
    <description>Tenstorrent&apos;s Galaxy Blackhole 6U server packs 32 Blackhole chips and 23 PFLOPS into a disaggregation-free inference platform, hitting 350 tokens/s on DeepSeek-671B in Blitz mode.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>tools</category>
  </item>
  <item>
    <title>UK Bets on a National AI Hardware Plan, Gallium Supply Chain Exposed</title>
    <link>https://hw.dev/signal/uk-ai-hardware-plan-kendall-2026</link>
    <guid isPermaLink="true">https://hw.dev/signal/uk-ai-hardware-plan-kendall-2026</guid>
    <pubDate>Thu, 30 Apr 2026 12:00:00 GMT</pubDate>
    <description>UK Technology Secretary Liz Kendall announced a national AI hardware plan targeting 5% of the projected $1T AI chip market, while the speech itself exposed Britain&apos;s critical gallium supply-chain vulnerability.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
    <category>supply-chain</category>
  </item>
  <item>
    <title>Cadence Lifts Its Revenue Forecast as AI Chip Design Demand Holds</title>
    <link>https://hw.dev/signal/cadence-revenue-forecast-ai-eda-2026</link>
    <guid isPermaLink="true">https://hw.dev/signal/cadence-revenue-forecast-ai-eda-2026</guid>
    <pubDate>Wed, 29 Apr 2026 12:00:00 GMT</pubDate>
    <description>Cadence raised its 2026 revenue forecast to $6.13-6.23B (up from $5.9-6B) as AI SoC complexity drives EDA demand, beating Q1 estimates on both revenue and adjusted EPS.</description>
    <category>signal</category>
    <category>eda</category>
    <category>ai-hardware</category>
    <category>tools</category>
  </item>
  <item>
    <title>Google Splits Its 8th-Gen TPU Into Two: One for Training, One for Inference</title>
    <link>https://hw.dev/signal/google-tpu-8t-8i-agentic-era</link>
    <guid isPermaLink="true">https://hw.dev/signal/google-tpu-8t-8i-agentic-era</guid>
    <pubDate>Wed, 29 Apr 2026 12:00:00 GMT</pubDate>
    <description>Google&apos;s 8th-gen TPU splits into two purpose-built chips: TPU 8t for training at 121 ExaFlops per superpod, and TPU 8i for low-latency inference with 3x the on-chip SRAM of its predecessor.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>Why Manual PCB Signoff Doesn&apos;t Scale -- and What Automated DRC Actually Buys You</title>
    <link>https://hw.dev/signal/pcb-signoff-hyperlynx-drc-automation</link>
    <guid isPermaLink="true">https://hw.dev/signal/pcb-signoff-hyperlynx-drc-automation</guid>
    <pubDate>Wed, 29 Apr 2026 12:00:00 GMT</pubDate>
    <description>Siemens HyperLynx DRC automates PCB electrical signoff across SI, PI, and EMI/EMC domains, shifting rule-based verification earlier in the design cycle and replacing the layer-by-layer manual inspection that misses high-speed net violations.</description>
    <category>signal</category>
    <category>pcb</category>
    <category>eda</category>
    <category>verification</category>
  </item>
  <item>
    <title>RISC-V Is Quietly Becoming the Connective Tissue of AI Hardware</title>
    <link>https://hw.dev/signal/risc-v-ai-hardware-open-foundation</link>
    <guid isPermaLink="true">https://hw.dev/signal/risc-v-ai-hardware-open-foundation</guid>
    <pubDate>Wed, 29 Apr 2026 12:00:00 GMT</pubDate>
    <description>Jon Peddie Research maps how RISC-V is scaling from always-on keyword detection up through chiplet-based inference SoCs, with three distinct NPU integration architectures now in commercial production.</description>
    <category>signal</category>
    <category>risc-v</category>
    <category>ai-hardware</category>
    <category>embedded</category>
  </item>
  <item>
    <title>Chip Sovereignty Has Moved Past Chips -- AMD CTO on the System-Level Shift</title>
    <link>https://hw.dev/signal/chip-sovereignty-systems-not-chips-amd-papermaster</link>
    <guid isPermaLink="true">https://hw.dev/signal/chip-sovereignty-systems-not-chips-amd-papermaster</guid>
    <pubDate>Tue, 28 Apr 2026 12:00:00 GMT</pubDate>
    <description>AMD CTO Mark Papermaster and Chip War author Chris Miller argue that semiconductor strategic value has shifted from fab ownership to system architecture -- AI accelerators, advanced packaging, memory bandwidth, and the software stack that ties them together.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>chiplets</category>
    <category>semiconductor</category>
    <category>trends</category>
  </item>
  <item>
    <title>Cisco&apos;s Universal Quantum Switch Tackles Interoperability at Room Temperature</title>
    <link>https://hw.dev/signal/cisco-universal-quantum-switch</link>
    <guid isPermaLink="true">https://hw.dev/signal/cisco-universal-quantum-switch</guid>
    <pubDate>Tue, 28 Apr 2026 12:00:00 GMT</pubDate>
    <description>Cisco&apos;s working quantum switch prototype routes quantum information between different vendor systems at room temperature over standard telecom fiber, with only 4% fidelity loss.</description>
    <category>signal</category>
    <category>trends</category>
    <category>testing</category>
  </item>
  <item>
    <title>Copper Sintering Challenges Silver for Wide Band Gap Die Attach</title>
    <link>https://hw.dev/signal/copper-sintering-wbg-die-attach</link>
    <guid isPermaLink="true">https://hw.dev/signal/copper-sintering-wbg-die-attach</guid>
    <pubDate>Tue, 28 Apr 2026 12:00:00 GMT</pubDate>
    <description>Copper pressure-assisted sintering is closing in on silver as the preferred die attach method for GaN and SiC power devices, offering comparable thermal conductivity at lower cost and better long-term reliability.</description>
    <category>signal</category>
    <category>manufacturing</category>
    <category>trends</category>
  </item>
  <item>
    <title>DRAM Crunch Forces AI System Architects to Rethink Memory</title>
    <link>https://hw.dev/signal/dram-crunch-ai-system-design</link>
    <guid isPermaLink="true">https://hw.dev/signal/dram-crunch-ai-system-design</guid>
    <pubDate>Tue, 28 Apr 2026 12:00:00 GMT</pubDate>
    <description>DRAM prices up 3-4x and high-capacity modules on extended lead times are pushing hardware teams to design AI systems around smaller models and on-chip inference.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>embedded</category>
    <category>supply-chain</category>
  </item>
  <item>
    <title>EU DARE Project Scrambles After Codasip Divests Its RISC-V Business</title>
    <link>https://hw.dev/signal/eu-dare-codasip-riscv-divestiture</link>
    <guid isPermaLink="true">https://hw.dev/signal/eu-dare-codasip-riscv-divestiture</guid>
    <pubDate>Tue, 28 Apr 2026 12:00:00 GMT</pubDate>
    <description>Codasip has divested its low-end RISC-V processor business -- and Codasip Studio EDA tool rights -- to an undisclosed U.S. public semiconductor company, leaving the EU&apos;s 240M euro DARE supercomputing project searching for a replacement partner.</description>
    <category>signal</category>
    <category>risc-v</category>
    <category>eda</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>What Hormuz Exposed About Our Semiconductor Supply Chain</title>
    <link>https://hw.dev/signal/hormuz-semiconductor-helium-supply-chain</link>
    <guid isPermaLink="true">https://hw.dev/signal/hormuz-semiconductor-helium-supply-chain</guid>
    <pubDate>Tue, 28 Apr 2026 12:00:00 GMT</pubDate>
    <description>The Strait of Hormuz closure has cut global helium supply and collapsed Gulf air cargo, exposing the semiconductor industry&apos;s unhedged dependencies on both -- with cascading effects now visible in memory pricing, tool delivery timelines, and cleanroom operations.</description>
    <category>signal</category>
    <category>supply-chain</category>
    <category>semiconductor</category>
    <category>manufacturing</category>
  </item>
  <item>
    <title>YMTC Bets on Domestic Equipment for New NAND Fab -- and Xtacking 4.0 Is the Wildcard</title>
    <link>https://hw.dev/signal/ymtc-nand-xtacking4-new-fab-2026</link>
    <guid isPermaLink="true">https://hw.dev/signal/ymtc-nand-xtacking4-new-fab-2026</guid>
    <pubDate>Tue, 28 Apr 2026 12:00:00 GMT</pubDate>
    <description>YMTC is building a new NAND fab in Wuhan where over 50% of equipment comes from Chinese domestic suppliers -- a significant shift from pre-sanction fabs -- while its Xtacking 4.0 architecture at 270 layers continues to close the gap with SK Hynix and Samsung.</description>
    <category>signal</category>
    <category>semiconductor</category>
    <category>supply-chain</category>
    <category>manufacturing</category>
  </item>
  <item>
    <title>Broadcom and Meta Lock In Multiyear Deal for Industry&apos;s First 2nm AI Compute Accelerator</title>
    <link>https://hw.dev/signal/broadcom-meta-mtia-2nm-accelerator</link>
    <guid isPermaLink="true">https://hw.dev/signal/broadcom-meta-mtia-2nm-accelerator</guid>
    <pubDate>Mon, 27 Apr 2026 12:00:00 GMT</pubDate>
    <description>Meta and Broadcom signed a multiyear, multigenerational agreement through 2029 to co-develop Meta&apos;s MTIA AI chips, with the first result targeting the industry&apos;s first 2nm AI compute accelerator.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>Google Splits AI Inference Chip Design Across Broadcom, MediaTek, and Marvell</title>
    <link>https://hw.dev/signal/google-inference-asic-broadcom-marvell-mediatek</link>
    <guid isPermaLink="true">https://hw.dev/signal/google-inference-asic-broadcom-marvell-mediatek</guid>
    <pubDate>Mon, 27 Apr 2026 12:00:00 GMT</pubDate>
    <description>Google is running parallel custom ASIC programs at Broadcom, MediaTek, and Marvell -- each optimizing for a different cost and performance point in the inference supply chain, with two million memory processing units potentially in the pipeline.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
    <category>supply-chain</category>
  </item>
  <item>
    <title>Qualcomm Eyes Samsung SF2 2nm for Snapdragon 8 Elite 2 as TSMC Dual-Source Strategy Takes Shape</title>
    <link>https://hw.dev/signal/qualcomm-samsung-2nm-snapdragon-foundry</link>
    <guid isPermaLink="true">https://hw.dev/signal/qualcomm-samsung-2nm-snapdragon-foundry</guid>
    <pubDate>Mon, 27 Apr 2026 12:00:00 GMT</pubDate>
    <description>Qualcomm CEO Cristiano Amon visited Samsung and SK Hynix in Korea to explore manufacturing the Snapdragon 8 Elite 2 on Samsung&apos;s SF2 2nm process -- a return to Samsung foundry after a five-year absence driven by dual-sourcing strategy.</description>
    <category>signal</category>
    <category>semiconductor</category>
    <category>supply-chain</category>
  </item>
  <item>
    <title>Samsung Semiconductor Workers Threaten 18-Day Strike as AI Chip Boom Widens Pay Gap</title>
    <link>https://hw.dev/signal/samsung-workers-strike-threat-chip-supply</link>
    <guid isPermaLink="true">https://hw.dev/signal/samsung-workers-strike-threat-chip-supply</guid>
    <pubDate>Mon, 27 Apr 2026 12:00:00 GMT</pubDate>
    <description>40,000 Samsung workers rallied at the Pyeongtaek chip complex demanding a 7% wage hike and elimination of the bonus cap, threatening an 18-day strike starting May 21 if talks fail.</description>
    <category>signal</category>
    <category>supply-chain</category>
    <category>manufacturing</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>EDA Vendors Race to Certify on TSMC Angstrom-Era Nodes</title>
    <link>https://hw.dev/signal/eda-tsmc-angstrom-roadmap-2026</link>
    <guid isPermaLink="true">https://hw.dev/signal/eda-tsmc-angstrom-roadmap-2026</guid>
    <pubDate>Sun, 26 Apr 2026 12:00:00 GMT</pubDate>
    <description>At TSMC&apos;s 2026 North America Technology Symposium, Synopsys, Cadence, and Siemens each announced IP and flow certification for A14/A16/N2 nodes as TSMC extended the roadmap to A13 targeting 2029 production.</description>
    <category>signal</category>
    <category>eda</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>Google Splits TPU Gen 8 Into Dedicated Training and Inference Chips</title>
    <link>https://hw.dev/signal/google-tpu8-training-inference-split</link>
    <guid isPermaLink="true">https://hw.dev/signal/google-tpu8-training-inference-split</guid>
    <pubDate>Sun, 26 Apr 2026 12:00:00 GMT</pubDate>
    <description>Google&apos;s eighth-generation TPU ships as two distinct chips -- 8t for training at 121 ExaFLOPs and 8i for inference -- acknowledging that the two workloads have incompatible hardware requirements.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>Meta Bets on CPUs: Multibillion Deal for AWS Graviton5 Targets Agentic AI</title>
    <link>https://hw.dev/signal/meta-aws-graviton5-agentic-ai</link>
    <guid isPermaLink="true">https://hw.dev/signal/meta-aws-graviton5-agentic-ai</guid>
    <pubDate>Sun, 26 Apr 2026 12:00:00 GMT</pubDate>
    <description>Meta signed a multibillion-dollar multi-year deal for hundreds of thousands of AWS Graviton5 CPUs -- 3nm, 192 cores, 5x larger cache than Graviton4 -- to handle agentic AI workloads that GPUs serve inefficiently.</description>
    <category>signal</category>
    <category>ai-hardware</category>
  </item>
  <item>
    <title>RISC-V Crosses Into Production: Automotive and IoT Ship at Scale</title>
    <link>https://hw.dev/signal/risc-v-embedded-world-2026-automotive</link>
    <guid isPermaLink="true">https://hw.dev/signal/risc-v-embedded-world-2026-automotive</guid>
    <pubDate>Sun, 26 Apr 2026 12:00:00 GMT</pubDate>
    <description>Embedded World 2026 confirmed RISC-V is past the exploratory phase: Andes shipped over one million units in a consumer smartwatch, Nuclei has billions of SoCs deployed, and Infineon committed AURIX to RISC-V for next-generation automotive MCUs.</description>
    <category>signal</category>
    <category>risc-v</category>
    <category>embedded</category>
  </item>
  <item>
    <title>Google TPU 8 Splits Into Training and Inference Chips, Claims 3x Speed</title>
    <link>https://hw.dev/signal/google-tpu8-million-chip-cluster</link>
    <guid isPermaLink="true">https://hw.dev/signal/google-tpu8-million-chip-cluster</guid>
    <pubDate>Sat, 25 Apr 2026 12:00:00 GMT</pubDate>
    <description>Google&apos;s eighth-generation TPUs split into dedicated training (8t) and inference (8i) chips, claiming 3x faster training and 80% better performance per dollar -- and the ability to cluster 1 million TPUs together.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>Meta Bets on Arm CPUs Over GPUs for AI Agent Inference</title>
    <link>https://hw.dev/signal/meta-amazon-graviton-agent-inference</link>
    <guid isPermaLink="true">https://hw.dev/signal/meta-amazon-graviton-agent-inference</guid>
    <pubDate>Sat, 25 Apr 2026 12:00:00 GMT</pubDate>
    <description>Meta secured millions of AWS Graviton Arm CPUs for AI agent workloads -- a structural signal that inference for agentic tasks is separating from GPU territory on cost and latency grounds.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
    <category>trends</category>
  </item>
  <item>
    <title>Synopsys and Cadence Ship Production EDA Flows for TSMC&apos;s Angstrom-Era Nodes</title>
    <link>https://hw.dev/signal/synopsys-cadence-tsmc-angstrom-eda</link>
    <guid isPermaLink="true">https://hw.dev/signal/synopsys-cadence-tsmc-angstrom-eda</guid>
    <pubDate>Sat, 25 Apr 2026 12:00:00 GMT</pubDate>
    <description>At the TSMC Technology Symposium, Synopsys and Cadence both certified production flows for N2, A16, and A14 nodes -- Cadence claiming 7% better placed area on 2nm and Synopsys 5.5x 3DIC productivity -- while Siemens confirmed its Canopus AI acquisition targets manufacturing metrology.</description>
    <category>signal</category>
    <category>eda</category>
    <category>semiconductor</category>
    <category>chiplets</category>
  </item>
  <item>
    <title>TSMC&apos;s CoWoS Roadmap Scales to 14x Reticles and 24 HBM Stacks by 2029</title>
    <link>https://hw.dev/signal/tsmc-cowos-14x-reticle-system-on-wafer</link>
    <guid isPermaLink="true">https://hw.dev/signal/tsmc-cowos-14x-reticle-system-on-wafer</guid>
    <pubDate>Sat, 25 Apr 2026 12:00:00 GMT</pubDate>
    <description>TSMC&apos;s 2026 Technology Symposium detailed CoWoS interposers scaling to 14x reticle size and 24 HBM stacks by 2029, with System-on-Wafer targeting 40-plus reticles and 64 HBM stacks -- redefining what &apos;a chip&apos; means for AI infrastructure.</description>
    <category>signal</category>
    <category>semiconductor</category>
    <category>chiplets</category>
    <category>manufacturing</category>
  </item>
  <item>
    <title>Hafnium Oxide Memristors Achieve Stable Neuromorphic Switching at One Million Times Lower Current</title>
    <link>https://hw.dev/signal/cambridge-hafnium-memristor</link>
    <guid isPermaLink="true">https://hw.dev/signal/cambridge-hafnium-memristor</guid>
    <pubDate>Fri, 24 Apr 2026 12:00:00 GMT</pubDate>
    <description>Cambridge researchers built a hafnium oxide memristor that switches at currents a million times lower than existing oxide devices and demonstrates stable analog conductance across tens of thousands of cycles.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
    <category>embedded</category>
  </item>
  <item>
    <title>AI Agent Designs a Complete RISC-V CPU Core in 12 Hours</title>
    <link>https://hw.dev/signal/design-conductor-risc-v-ai-cpu</link>
    <guid isPermaLink="true">https://hw.dev/signal/design-conductor-risc-v-ai-cpu</guid>
    <pubDate>Fri, 24 Apr 2026 12:00:00 GMT</pubDate>
    <description>Verkor.io&apos;s Design Conductor agentic AI system produced a verified 1.48 GHz RISC-V core from a 219-word spec -- the first end-to-end autonomous CPU design from spec to GDSII layout.</description>
    <category>signal</category>
    <category>risc-v</category>
    <category>eda</category>
    <category>ai-hardware</category>
    <category>tools</category>
  </item>
  <item>
    <title>Siemens and TSMC Expand AI Automation Across the Full Design Stack</title>
    <link>https://hw.dev/signal/siemens-tsmc-ai-eda-collaboration</link>
    <guid isPermaLink="true">https://hw.dev/signal/siemens-tsmc-ai-eda-collaboration</guid>
    <pubDate>Fri, 24 Apr 2026 12:00:00 GMT</pubDate>
    <description>Siemens and TSMC certified Fuse EDA AI Agent across N2P, A16, and A14 nodes while extending 3DFabric and silicon photonics support -- a systematic push to make agentic EDA the default, not a beta feature.</description>
    <category>signal</category>
    <category>eda</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
  </item>
  <item>
    <title>Challenger+ RP2350 Pairs Dual-Core RISC-V MCU with NB-IoT and GNSS</title>
    <link>https://hw.dev/signal/challenger-rp2350-nb-iot-gnss</link>
    <guid isPermaLink="true">https://hw.dev/signal/challenger-rp2350-nb-iot-gnss</guid>
    <pubDate>Thu, 23 Apr 2026 12:00:00 GMT</pubDate>
    <description>iLABs&apos; Challenger+ RP2350 board integrates NB-IoT (LTE Cat NB2), GPS/Galileo, and Wi-Fi positioning on a Feather-compatible form factor -- making a compelling case for outdoor IoT trackers without a cellular module hat.</description>
    <category>signal</category>
    <category>embedded</category>
    <category>risc-v</category>
    <category>aiot</category>
  </item>
  <item>
    <title>Espressif ESP-Claw Brings Local AI Agents to ESP32</title>
    <link>https://hw.dev/signal/espressif-esp-claw-edge-ai-esp32</link>
    <guid isPermaLink="true">https://hw.dev/signal/espressif-esp-claw-edge-ai-esp32</guid>
    <pubDate>Thu, 23 Apr 2026 12:00:00 GMT</pubDate>
    <description>Espressif released ESP-Claw, a Chat-as-Coding edge agent framework that runs LLM-driven decision loops locally on ESP32-S3 hardware -- no cloud required unless you want it.</description>
    <category>signal</category>
    <category>embedded</category>
    <category>ai-hardware</category>
    <category>aiot</category>
  </item>
  <item>
    <title>GigaDevice GD32F5HC: Cortex-M33 at 200 MHz with TrustZone and Hardware Crypto</title>
    <link>https://hw.dev/signal/gigadevice-gd32f5hc-cortex-m33-200mhz</link>
    <guid isPermaLink="true">https://hw.dev/signal/gigadevice-gd32f5hc-cortex-m33-200mhz</guid>
    <pubDate>Thu, 23 Apr 2026 12:00:00 GMT</pubDate>
    <description>GigaDevice launched the GD32F5HC, a Cortex-M33 MCU running at 200 MHz with 2MB Flash, TrustZone, and a hardware crypto block -- competing directly in the secure edge compute tier dominated by STM32 and NXP.</description>
    <category>signal</category>
    <category>embedded</category>
    <category>semiconductor</category>
    <category>testing</category>
  </item>
  <item>
    <title>Microchip PIC MCUs Gain CPLD-Like Configurable Logic Blocks</title>
    <link>https://hw.dev/signal/microchip-pic-configurable-logic-blocks</link>
    <guid isPermaLink="true">https://hw.dev/signal/microchip-pic-configurable-logic-blocks</guid>
    <pubDate>Thu, 23 Apr 2026 12:00:00 GMT</pubDate>
    <description>Microchip launched two 8-bit PIC families -- PIC16F132 and PIC18-Q35 -- with on-chip Configurable Logic Blocks that let you offload digital glue logic from firmware to hardware without an external CPLD or FPGA.</description>
    <category>signal</category>
    <category>embedded</category>
    <category>fpga</category>
    <category>tools</category>
  </item>
  <item>
    <title>NVIDIA NemoClaw: When the World&apos;s Biggest Chip Company Bets on OpenClaw</title>
    <link>https://hw.dev/signal/nvidia-nemoclaw-openclaw-hardware-ai-agents</link>
    <guid isPermaLink="true">https://hw.dev/signal/nvidia-nemoclaw-openclaw-hardware-ai-agents</guid>
    <pubDate>Thu, 23 Apr 2026 12:00:00 GMT</pubDate>
    <description>NVIDIA&apos;s NemoClaw wraps OpenClaw with a sandboxed, policy-enforced security stack -- and what that means for hardware engineers running always-on AI agents is bigger than the press release lets on.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>tools</category>
    <category>embedded</category>
    <category>software</category>
  </item>
  <item>
    <title>The 2026 Guide to Autonomous PCB Design: Quilter vs. DeepPCB vs. Flux.ai</title>
    <link>https://hw.dev/signal/autonomous-pcb-design-2026-guide</link>
    <guid isPermaLink="true">https://hw.dev/signal/autonomous-pcb-design-2026-guide</guid>
    <pubDate>Wed, 22 Apr 2026 12:00:00 GMT</pubDate>
    <description>Quilter published a head-to-head comparison of the three leading AI PCB autorouters -- itself, DeepPCB, and Flux.ai -- and the technical differentiation is sharper than the marketing suggests.</description>
    <category>signal</category>
    <category>pcb</category>
    <category>eda</category>
    <category>tools</category>
    <category>ai-hardware</category>
  </item>
  <item>
    <title>Cadence and Nvidia Deepen Partnership at CadenceLIVE Silicon Valley 2026</title>
    <link>https://hw.dev/signal/cadence-nvidia-cadencelive-2026</link>
    <guid isPermaLink="true">https://hw.dev/signal/cadence-nvidia-cadencelive-2026</guid>
    <pubDate>Tue, 21 Apr 2026 12:00:00 GMT</pubDate>
    <description>At CadenceLIVE Silicon Valley, Cadence and Nvidia announced an expanded partnership combining Cadence&apos;s EDA and SDA toolchains with Nvidia CUDA-X, AI physics models, and Omniverse -- targeting up to 100x speedups in simulation and verification workflows.</description>
    <category>signal</category>
    <category>eda</category>
    <category>ai-hardware</category>
    <category>verification</category>
    <category>tools</category>
  </item>
  <item>
    <title>Positron AI&apos;s Asimov Chip Goes Live in Oracle Cloud for MOE Inference</title>
    <link>https://hw.dev/signal/positron-ai-oracle-inference-asimov</link>
    <guid isPermaLink="true">https://hw.dev/signal/positron-ai-oracle-inference-asimov</guid>
    <pubDate>Tue, 21 Apr 2026 12:00:00 GMT</pubDate>
    <description>Positron AI has deployed tens of millions of dollars worth of Asimov-based inference systems into Oracle Cloud, targeting mixture-of-experts model inference -- one of the first non-Nvidia, non-AMD chips to reach production scale in a major cloud provider&apos;s AI infrastructure.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>semiconductor</category>
    <category>chiplets</category>
  </item>
  <item>
    <title>Siemens EDA Makes the Case for Agentic Automation Across the Full Design Lifecycle</title>
    <link>https://hw.dev/signal/siemens-eda-agentic-ai-pcb-semiconductor</link>
    <guid isPermaLink="true">https://hw.dev/signal/siemens-eda-agentic-ai-pcb-semiconductor</guid>
    <pubDate>Tue, 21 Apr 2026 12:00:00 GMT</pubDate>
    <description>Siemens EDA&apos;s product team argues that AI copilots are no longer sufficient -- what the industry needs is an agentic orchestration layer that spans the full design lifecycle from concept to manufacturing sign-off, capable of multi-step reasoning across fragmented EDA toolchains.</description>
    <category>signal</category>
    <category>eda</category>
    <category>pcb</category>
    <category>verification</category>
    <category>tools</category>
    <category>software</category>
  </item>
  <item>
    <title>Syntacore SCR RISC-V IP Ships Full Zephyr 4.3 Support</title>
    <link>https://hw.dev/signal/syntacore-zephyr-4-3-risc-v</link>
    <guid isPermaLink="true">https://hw.dev/signal/syntacore-zephyr-4-3-risc-v</guid>
    <pubDate>Tue, 21 Apr 2026 12:00:00 GMT</pubDate>
    <description>Syntacore&apos;s SCR RISC-V IP portfolio now ships with full, pre-configured Zephyr 4.3 RTOS support including multi-core and out-of-the-box deployment -- a small update with real friction-reduction value for embedded teams starting new RISC-V designs.</description>
    <category>signal</category>
    <category>risc-v</category>
    <category>embedded</category>
    <category>tools</category>
  </item>
  <item>
    <title>Cadence and NVIDIA Bring 30x Simulation Speedup to Physical AI</title>
    <link>https://hw.dev/signal/cadence-nvidia-physical-ai-digital-twin</link>
    <guid isPermaLink="true">https://hw.dev/signal/cadence-nvidia-physical-ai-digital-twin</guid>
    <pubDate>Mon, 20 Apr 2026 12:00:00 GMT</pubDate>
    <description>Cadence and NVIDIA expanded their partnership at CadenceLIVE to integrate the Reality Digital Twin Platform with Omniverse and Isaac, delivering 30x multi-physics simulation speedup on Blackwell GPUs -- targeting the Sim-to-Real gap in robotics and physical AI.</description>
    <category>signal</category>
    <category>eda</category>
    <category>ai-hardware</category>
    <category>tools</category>
    <category>software</category>
  </item>
  <item>
    <title>Siemens and Nvidia Push Chip Verification to Trillion-Cycle Scale</title>
    <link>https://hw.dev/signal/siemens-nvidia-trillion-cycle-verification</link>
    <guid isPermaLink="true">https://hw.dev/signal/siemens-nvidia-trillion-cycle-verification</guid>
    <pubDate>Mon, 20 Apr 2026 12:00:00 GMT</pubDate>
    <description>Siemens&apos; Veloce proFPGA CS paired with Nvidia GPU architecture promises trillion design cycles in days -- a direct response to verification becoming the dominant bottleneck in AI chip development.</description>
    <category>signal</category>
    <category>verification</category>
    <category>eda</category>
    <category>ai-hardware</category>
    <category>fpga</category>
    <category>tools</category>
  </item>
  <item>
    <title>Teradyne Acquires TestInsight to Close the Design-to-Test Gap</title>
    <link>https://hw.dev/signal/teradyne-testinsight-ate-acquisition</link>
    <guid isPermaLink="true">https://hw.dev/signal/teradyne-testinsight-ate-acquisition</guid>
    <pubDate>Mon, 20 Apr 2026 12:00:00 GMT</pubDate>
    <description>Teradyne&apos;s acquisition of TestInsight brings pre-silicon validation and automated test pattern generation in-house -- a direct play to reduce the handoff friction between design teams and ATE platforms.</description>
    <category>signal</category>
    <category>testing</category>
    <category>tools</category>
    <category>semiconductor</category>
    <category>software</category>
    <category>verification</category>
  </item>
  <item>
    <title>TSMC Commits $56B Capex as AI Chip Demand Outpaces Fabs</title>
    <link>https://hw.dev/signal/tsmc-56b-capex-3nm-ai-demand</link>
    <guid isPermaLink="true">https://hw.dev/signal/tsmc-56b-capex-3nm-ai-demand</guid>
    <pubDate>Mon, 20 Apr 2026 12:00:00 GMT</pubDate>
    <description>TSMC raised its 2026 capex to nearly $56B -- and CEO C.C. Wei acknowledged they&apos;ll still fall short of demand in 2027, with three new 3nm fabs racing to come online across Japan, Taiwan, and the US.</description>
    <category>signal</category>
    <category>semiconductor</category>
    <category>manufacturing</category>
    <category>ai-hardware</category>
    <category>supply-chain</category>
  </item>
  <item>
    <title>CadenceLive 2026: Agentic AI Stacks Land in EDA</title>
    <link>https://hw.dev/signal/cadencelive-2026-ai-agent-stacks</link>
    <guid isPermaLink="true">https://hw.dev/signal/cadencelive-2026-ai-agent-stacks</guid>
    <pubDate>Sun, 19 Apr 2026 12:00:00 GMT</pubDate>
    <description>Cadence unveiled a hierarchical agentic AI architecture at CadenceLive, pairing an orchestrator layer with domain-specific super-agents across RTL design and verification — and opened the business model question nobody wants to answer.</description>
    <category>signal</category>
    <category>eda</category>
    <category>ai-hardware</category>
    <category>tools</category>
    <category>verification</category>
  </item>
  <item>
    <title>EU&apos;s €240M RISC-V Program Scrambles After Codasip&apos;s Strategic Pivot</title>
    <link>https://hw.dev/signal/eu-dare-codasip-risc-v-pivot</link>
    <guid isPermaLink="true">https://hw.dev/signal/eu-dare-codasip-risc-v-pivot</guid>
    <pubDate>Sun, 19 Apr 2026 12:00:00 GMT</pubDate>
    <description>The €240M EU DARE RISC-V initiative is actively replacing Codasip after the Munich-based processor IP company divested key business units to an undisclosed US semiconductor firm, leaving European digital sovereignty plans in flux.</description>
    <category>signal</category>
    <category>risc-v</category>
    <category>embedded</category>
    <category>supply-chain</category>
    <category>trends</category>
  </item>
  <item>
    <title>Intel Signs On for Terafab: A 100M Sq-Ft Chip Campus with Tesla and SpaceX</title>
    <link>https://hw.dev/signal/intel-terafab-tesla-spacex</link>
    <guid isPermaLink="true">https://hw.dev/signal/intel-terafab-tesla-spacex</guid>
    <pubDate>Sun, 19 Apr 2026 12:00:00 GMT</pubDate>
    <description>Intel will partner with Tesla, SpaceX, and xAI to build Terafab — a proposed 100 million square-foot collocated semiconductor campus in Austin targeting one terawatt of annual compute output.</description>
    <category>signal</category>
    <category>manufacturing</category>
    <category>supply-chain</category>
    <category>ai-hardware</category>
    <category>trends</category>
  </item>
  <item>
    <title>Tesla AI5 Tapes Out on 3nm: &apos;Radical Simplicity&apos; Meets Sovereign Silicon</title>
    <link>https://hw.dev/signal/tesla-ai5-tape-out-3nm</link>
    <guid isPermaLink="true">https://hw.dev/signal/tesla-ai5-tape-out-3nm</guid>
    <pubDate>Sun, 19 Apr 2026 12:00:00 GMT</pubDate>
    <description>Tesla taped out its AI5 chip on 3nm — dual-sourced from TSMC Arizona and Samsung Taylor — claiming 8-10x the compute of HW4, 192GB LPDDR5X, and a philosophy of stripping away everything that isn&apos;t neural network inference.</description>
    <category>signal</category>
    <category>ai-hardware</category>
    <category>chiplets</category>
    <category>manufacturing</category>
    <category>embedded</category>
  </item>
  <item>
    <title>Siemens Launches Fuse EDA AI Agent for PCB and Semiconductor Workflows</title>
    <link>https://hw.dev/signal/siemens-fuse-eda-ai-agent</link>
    <guid isPermaLink="true">https://hw.dev/signal/siemens-fuse-eda-ai-agent</guid>
    <pubDate>Sun, 15 Mar 2026 12:00:00 GMT</pubDate>
    <description>Siemens announced Fuse EDA AI Agent, a domain-scoped autonomous agent that plans and orchestrates multi-tool workflows spanning design, verification, and manufacturing sign-off for semiconductors, 3D ICs, and PCBs. This is the biggest incumbent response to AI-native EDA challengers yet.</description>
    <category>signal</category>
    <category>eda</category>
    <category>ai-agent</category>
    <category>pcb</category>
    <category>semiconductor</category>
    <category>automation</category>
  </item>
  <item>
    <title>RISC-V at Embedded World 2026: Production-Ready, Automotive-Grade, AI-Native</title>
    <link>https://hw.dev/signal/risc-v-embedded-world-2026</link>
    <guid isPermaLink="true">https://hw.dev/signal/risc-v-embedded-world-2026</guid>
    <pubDate>Sun, 01 Mar 2026 12:00:00 GMT</pubDate>
    <description>Embedded World 2026 marked a decisive shift in RISC-V conversations — from architectural promise to production readiness. Automotive-grade cores, AI-native extensions, and wearable deployments (including the Amazfit T-Rex 3 Pro) showed the ISA has crossed the chasm from experimentation to volume production.</description>
    <category>signal</category>
    <category>risc-v</category>
    <category>embedded-world</category>
    <category>automotive</category>
    <category>ai</category>
    <category>production</category>
  </item>
  </channel>
</rss>