application-specific integrated circuit
PulseAugur coverage of application-specific integrated circuit — every cluster mentioning application-specific integrated circuit across labs, papers, and developer communities, ranked by signal.
4 day(s) with sentiment data
AI Compute Demands Driving PCB Industry Upgrades for ASIC/GPU Clusters
The increasing demand for AI computing power, particularly from GPU/ASIC clusters, is forcing significant upgrades in the Printed Circuit Board (PCB) industry. This includes a shift towards higher layer counts and high-density interconnect boards to support the advanced requirements of AI hardware.
Chinese AI Labs Prioritizing Custom ASICs Over NVIDIA
Multiple Chinese AI labs (Zhipu, DeepSeek) are actively developing custom AI chips to mitigate reliance on NVIDIA. This is driven by export control concerns and a desire to avoid vendor lock-in, indicating a strategic shift towards vertical integration within China's AI sector.
BitLogic Framework to Accelerate ASIC Adoption for Boolean Logic Neural Networks
The newly developed BitLogic framework unifies training for neural networks using Boolean logic operations, enabling deployment across various hardware including ASICs. This standardization could significantly lower the barrier to entry for developing and deploying Boolean logic-based ASICs for AI tasks.
ASIC manufacturers to offer specialized AI training chips leveraging Boolean logic
The development of the BitLogic framework, which enables gradient-based neural network training using Boolean logic operations on FPGAs and ASICs, suggests a future where ASICs are optimized for this type of computation. This could lead to specialized AI training chips that offer significant efficiency gains over traditional multiply-accumulate architectures.
ASIC feasibility studies for neural spike sorting indicate a growing niche market
The DBNN research highlights successful ASIC feasibility studies for low-power, on-node neural spike sorting in brain-computer interfaces. This indicates a specific, high-value niche where ASICs are being seriously considered due to their power and area advantages, suggesting potential for specialized ASIC development in the biomedical field.
-
MediaTek plans $5B AI datacenter push with custom ASIC chips
MediaTek is planning a significant investment of $5 billion to enter the AI datacenter market. The Taiwanese semiconductor company aims to capture up to 20% of the estimated $80 billion market with its upcoming custom A…
-
Shunluo Electronics begins mass production of AI inductors for servers
Shunluo Electronics has begun mass production and shipment of AI inductors, which are crucial components for power supply in AI servers for various chips like GPUs and CPUs. The company has gained recognition and is sup…
-
Southbound Capital Sees Volatility Amidst AI Tech Developments
Southbound capital saw a net buying of 3 billion Hong Kong dollars, followed by a net selling of over 5 billion Hong Kong dollars, indicating significant volatility in market sentiment. In related tech news, Shunluo Ele…
-
Chinese telcos embrace TPUs for AI compute, focusing on token economics
Chinese telecom operators are increasingly investing in AI computing power, moving beyond traditional GPUs to explore other chip architectures like TPUs and ASICs. This shift is driven by a focus on cost-efficiency, par…
-
Chinese AI Labs Pursue Custom Chips Amid NVIDIA Dependence Concerns · 1 source tracked
Chinese AI labs like Zhipu and DeepSeek are developing custom AI chips to reduce reliance on NVIDIA, driven by concerns over export controls and vendor lock-in. This move signals a trend toward vertical integration with…
-
China's first offshore wind converter station ships; AI drives PCB upgrades
A significant milestone has been reached for China's first integrated offshore wind power direct current transmission project, with the core hub, the offshore converter station, beginning its transport. This project is …
-
BitLogic framework unifies training for FPGA-native neural networks
Researchers have developed BitLogic, a unified framework designed to standardize the training and evaluation of gradient-based neural networks that utilize Boolean logic operations instead of traditional multiply-accumu…
-
DBNN enables accurate, low-power neural spike sorting for brain-computer interfaces
Researchers have developed a Deep Binarized Neural Network (DBNN) for on-node spike sorting in implantable brain-computer interfaces. This DBNN system, featuring two binarized hidden layers, enables multiplier-free infe…
-
CITIC Securities: FPSO demand rises amid deep-sea oil focus, AI drives PCB upgrades
A report from CITIC Securities indicates a surge in demand for Floating Production Storage and Offloading (FPSO) units, driven by the increasing focus on deep-sea oil and gas exploration due to insufficient conventional…
-
AI computing power drives demand for advanced PCBs, boosting equipment and materials sector
A recent report from CITIC Securities highlights a significant surge in demand for printed circuit boards (PCBs) driven by the expansion of AI computing power. This demand is pushing for upgrades in PCB technology, requ…
-
OpenRegistry launches live ABR data access tool for investigators
OpenRegistry has launched a new tool that provides direct access to Australia's live Australian Business Register (ABR) data via MCP. This tool aims to assist investigators and journalists by offering real-time informat…
-
Apple and Broadcom extend custom chip deal through 2031 · 4 sources tracked
Broadcom and Apple have extended their technology collaboration through 2031, with Broadcom agreeing to develop and supply a range of custom ASIC chips for Apple's future products. This multiyear agreement builds upon t…
-
Samsung Foundry's 4nm capacity sold out amid AI chip demand surge
Samsung Foundry is experiencing a surge in demand for its advanced semiconductor manufacturing processes, particularly for AI-related chips. The company's 4nm process is nearly sold out, with capacity for next year also…
-
OpenAI and Broadcom unveil Jalapeño AI inference chip · 10 sources tracked
OpenAI and Broadcom have unveiled Jalapeño, their first jointly-developed AI inference processor. This custom chip is designed to run large language models more efficiently, promising better performance per watt than cu…
-
NeuronFabric architecture enables on-chip transformer training
Researchers have introduced NeuronFabric, a software reference architecture designed for on-chip transformer training using local Adam updates. A C# prototype demonstrates the feasibility of this approach, handling forw…
-
New ITP-STDP engine slashes SNN training energy use
Researchers have developed a new learning engine called ITP-STDP for training spiking neural networks (SNNs) that significantly reduces hardware resource utilization and energy consumption. This novel approach optimizes…
-
Australian Business Register data aids investigations but lacks ownership details
The Australian Business Register (ABR), managed by the Australian Taxation Office, provides an 11-digit identifier (ABN) for various entities, including companies, charities, and sole traders. This register is crucial f…
-
China redesigns AI chip industry amid US export curbs
China's AI chip industry is undergoing a significant redesign in response to US export controls, pushing domestic companies to explore alternatives to Nvidia's dominant GPUs. Major players like Huawei and Cambricon are …
-
Deep learning model predicts Bitcoin mining hardware ROI
Researchers have developed MineROI-Net, a deep learning framework using a Transformer architecture to predict the profitability of Bitcoin mining hardware purchases. The model classifies acquisition timing into profitab…
-
AI industry pivots to token economics, focusing on inference computing centers
The AI industry is shifting its focus from model parameters to computational efficiency, with "token economics" emerging as a new value unit. This transition is driving demand for "token factories" – intelligent computi…