The AI Chips Powering Tomorrow's Industries
Driving transformative impact across every sector
Introduction
The rise of artificial intelligence (AI) is transforming every industry, from healthcare and finance to transportation and manufacturing. At the heart of this AI revolution lie specialized chips designed to handle the unique demands of machine learning and deep learning workloads. As AI models become increasingly sophisticated and data-intensive, traditional CPUs and even GPUs are struggling to keep pace. This has sparked a wave of innovation in AI-specific semiconductors, attracting billions in investment from tech giants, chip makers, and a dynamic ecosystem of startups.
AI chips represent one of the hottest areas of investment and innovation in the semiconductor industry today. The global AI chip market is projected to grow from $8 billion in 2020 to over $70 billion by 2026, potentially becoming the largest semiconductor category by the end of the decade. Tech titans like Google, Amazon, Apple, and Microsoft are investing heavily to develop their own custom AI silicon. Legacy semiconductor players like Nvidia, Intel, and AMD are pivoting to become AI computing powerhouses. And over a hundred VC-backed startups like Cerebras Systems, Graphcore, SambaNova, and Mythic are pushing the boundaries of AI chip architectures.
In this deep dive, we'll explore why AI chips have become so crucial, who the key players are, and how these powerful processors are driving transformative impact across every sector. The future of computing is getting smarter by the day, and AI chips are at the forefront of this intelligent revolution.
From Basic Transistors to Advanced Neural Networks
Tracing the Progressive Evolution of Semiconductor Technologies
1940s-50s:
The invention of the transistor in 1947.
Development of early integrated circuits(ICs) by Texas Instruments and Fairchild Semiconductor
1960s
Introduction of Metal-Oxide-Semiconductor Field-Effect Transistors (MOSFETs).
Formulation of Moore's Law by Gordon Moore in 1965
1970s
Adoption of Complementary Metal-Oxide-Semiconductor (CMOS) technology.
Introduction of microprocessors, such as the Intel 4004 (1971) and Intel 8080 (1974).
1980s
Advancements in process technology enable higher transistor densities.
Initial research on Extreme ultraviolet lithography (EUV).
Introduction of System-on-Chip (SoC) integration, combining multiple functionalities on a single chip.
Emergence of Application-Specific Integrated Circuits (ASICs) for specialized applications.
1990s
Introduction of FinFET transistors for improved performance and power efficiency.
Expansion of the semiconductor industry into consumer electronics, telecommunications, and networking.
2000s
Development of advanced process nodes, such as 45nm, 32nm, and 22nm.
Introduction of 3D integration technologies for stacking multiple layers of chips.
The first EUV prototype is created.
Proliferation of smartphones and IoT devices driven by highly integrated SoCs.
2010s
Scaling of semiconductor technology to sub-10nm nodes.
Initial trials with EUV machines.
Emergence of specialized process technologies like Fully Depleted Silicon-on-Insulator (FD-SOI).
Advancements in heterogeneous computing architectures and GPU acceleration.
2020s
Focus on advanced process nodes and EUV lithography.
Exploration of emerging technologies like quantum computing, neuromorphic computing, and silicon photonics.
Integration of AI and machine learning accelerators into semiconductor chips for edge and cloud applications.
Key Trends in Advanced Chip Development
AI & Machine Learning
Integration of specialized hardware accelerators for AI and machine learning workloads to speed up the processing of complex algorithms.
Edge Computing
Semiconductor chips designed for edge computing applications, allowing processing to occur closer to the data source, resulting in lower latency and more efficiency.
Advanced Nodes
Continued development of smaller process nodes, including sub-10nm1 and EUV lithography2, to enable higher transistor density and improved performance.
Quantum Computing
Growing investment in R&D of quantum computing technologies aiming to revolutionize computing capabilities for specific applications.
5G & IoT
Development of semiconductor chips tailored for 5G connectivity and IoT devices emphasizing low power consumption, high bandwidth, and security.
Security & Trust
Chip-level security features to protect against cybersecurity threats, including hardware-level encryption, secure boot, and tamper-resistant technologies.
Automotive
Advancements in semiconductor chips for automotive applications, including autonomous driving, ADAS, and in-vehicle infotainment.
Sustainability
Emphasis on energy-efficient designs, eco-friendly manufacturing processes, and materials to reduce the environmental impact of semiconductor production.
System Integration
Integration of diverse components such as CPUs, GPUs, accelerators, and memory on a single chip or package enhancing efficiency and performance.
Supply Chain
Heightened focus on supply chain resilience and diversification in response to disruptions, including geopolitical tensions and natural disasters.
Race for Dominance as competition Intensifies
These collaborations, investments, and acquisitions demonstrate a concerted push to bolster AI capabilities in the semiconductor industry, following the launch of OpenAI's ChatGPT in November 2022.
NVIDIA
Collaboration with NTT, Red Hat, and Fujitsu: As part of the IOWN initiative, this collaboration aims to enhance and extend the potential for real-time artificial intelligence (AI) data analysis at the edge. The solution leverages technologies developed by the IOWN Global Forum and is built upon the infrastructure of Red Hat OpenShift.
Support for AI chip startup Enfabrica: Enfabrica has successfully secured $125 million in Series B funding, with support from collaborators in the IOWN initiative.
Investment in AI research startup Runway AI: Runway AI received investment for its Series C fundraising round, furthering its efforts in AI research and development.
Joint AI venture with Dell Technologies: A partnership with Dell Technologies has led to the launch of a joint AI venture, aiming to advance AI capabilities in various applications.
Acquisition of OmniML: OmniML, a startup specializing in edge AI, has been acquired, further strengthening the capabilities in edge AI technology and applications.
INTEL
Joined forces with Microsoft to bring AI capabilities to personal computers.
Collaboration with VMware to accelerate the adoption of AI across data centers, public clouds, and edge.
Partnership with IT service provider Wipro to accelerate advanced AI chip design and development
Intel’s 18A technology was selected by Microsoft for producing AI chips.
SAMSUNG
Partnership with Naver Corp. to develop next-generation AI chips.
Collaboration with South Korea’s KT Group to develop full-stack AI capabilities.
Invested in AI chip startup Tenstorrent along with providing it foundry services.
ASML
Announces a new chip-making plant in Japan.
Shipped first high-NA EUV machine to Intel for its 1.8nm chips.
SK HYNIX
Unveiled a prototype of AiMX, a generative AI accelerator card that combines GDDR6 chips to enhance AI performance.
Microsoft and Google have been notably proactive in enhancing their AI chip capacities through investments or acquisitions of startups. Apple's latest acquisition of DarwinAI signifies its aim to fortify its own chip capabilities.
Semiconductor venture landscape
Over the past five years, AI has emerged as the leading theme in the overall venture capital deal value for the semiconductor sector, with almost 27% of the deal value from 2019 to 2023 going towards AI.
While AI-focused startups are attracting the most funding, investors are also diversifying their investments by funding a broad range of AI chip startups. This trend is indicative of the increasing importance of AI in the semiconductor sector and the need for companies to develop AI-optimized chips.
Semiconductor venture deals trend ($ Mn): 2019 - 2023
Venture capital funds are investing in diverse AI chip startups, including fabless manufacturers and on-device AI solution providers.
A Glimpse into AI chips R&D ecosystem
The AI chip R&D landscape is marked by dynamic competition among market players; however, it is geographically concentrated with the US and China being the key hubs of innovation.
Top 10 patent filers (2019-YTD)
The AI chip patent landscape signifies a dynamic and competitive landscape in the AI chip market:
Illustrative enterprise patent filings in 2023:
Apple: Systems and methods for task switching in neural network processor
Rebellions Inc.: Neural processing device and method for job scheduling thereof
MediaTek: Always-on artificial intelligence (AI) security
Nvidia: Fairly utilizing multiple contexts sharing cryptographic hardware
Samsung: Hbm-based memory lookup engine for deep learning accelerator
Microsoft: Reduced power consumption analog or hybrid mac neural network
Sector impact of advanced chips
Advanced chips are set to revolutionize multiple sectors, boosting performance and energy efficiency in technology, enabling autonomous driving and personalized healthcare, optimizing telecom networks, and enhancing energy management.
Technology Sector: Trends & Innovations
Increased Performance: AI chips boost tech applications, such as data centers, cloud, and edge computing, by accelerating AI computations.
Energy Efficiency: Task-specific chip optimization can improve data center and computing device energy efficiency.
Tenstorrent boosts AI performance using inference cards
Tenstorrent's Grayskull inference cards, e75 and e150, aim to enhance edge-based machine learning applications with efficient processing and high performance.
The Grayskull inference cards are positioned as a solution for enhancing the performance of edge-based machine learning applications.
The use of Tensix cores with the open-source RISC-V instruction set architecture demonstrates a commitment to open and accessible technology.
The integration of hardware accelerators for networking and data compression/decompression tasks highlights the focus on efficient processing.
The availability of two models, e75 and e150, allows for scalability in meeting different inference operation demands.
The introduction of TT-Buda and TT-Metalium software stacks indicates a focus on providing versatile tools for developers.
The launch of the Grayskull inference cards positions Tenstorrent at the forefront of machine learning innovation, particularly in the realm of edge computing and inference acceleration.
The availability of optimized hardware like the Grayskull cards is seen as a key factor in enabling more sophisticated and efficient AI applications.
Cerebras Systems brings generative AI-focused WSE-3 chip
The US-based chip startup Cerebras Systems has introduced the Wafer Scale Engine 3 (WSE-3), the third generation of its AI chip. The WSE-3 is designed for training large AI models and currently powers Cerebras' CS-3 AI supercomputer.
The WSE-3's enhanced capabilities are expected to accelerate advancements in AI technology, particularly in the training of large-scale generative AI models like GPT-4.
The WSE-3 chip represents a significant advancement in AI chip technology.
The chip's doubled performance is akin to a true Moore's Law step in the industry.
The WSE-3 chip is tailored for training AI models, enabling efficient refinement of neural weights or parameters.
The chip's enhanced capabilities are facilitated by a reduction in transistor size and an increase in transistor count.
Cerebras has maintained a balance between compute and memory in the WSE-3 chip.
The WSE-3 chip is expected to have significant implications for AI research and development.
The availability of advanced AI chips like the WSE-3 will play a crucial role in driving innovation and unlocking new possibilities in artificial intelligence.
Celestial AI addresses bandwidth challenges with optical interconnect platform
Celestial AI introduces Photonic Fabric, an optical interconnect platform designed for AI computing and memory infrastructure, offering improvements in bandwidth, memory capacity, latency, and power consumption.
Photonic Fabric is designed to address challenges in bandwidth, memory capacity, latency, and power consumption in AI computing and memory infrastructure.
The technology offers a 25 times increase in bandwidth and memory capacity and reductions in latency and power consumption by up to 10 times.
The platform comprises three components: optical interconnects, Memory Fabric, and Compute Fabric, which collectively target the resolution of performance bottlenecks in AI workloads.
The scalability of Photonic Fabric makes it suitable for various AI applications across industries.
Celestial AI has previewed its solution at the OFC Conference, indicating its commitment to driving adoption and innovation in AI infrastructure.
The technology has the potential to revolutionize AI computing and memory infrastructure.
As AI workloads continue to grow in complexity and scale, solutions like Photonic Fabric will be highly sought after.
Diraq Develops Silicon-Based Qubits
Diraq, an Australian quantum computing startup, has developed quantum processors using traditional silicon chip technologies. The startup's technology harnesses existing manufacturing processes to produce quantum chips, potentially enabling scalable production of qubits using existing semiconductor industry infrastructure.
Unlike other qubit technologies, Diraq's silicon-based qubits offer inherent stability and compatibility with existing semiconductor manufacturing processes. This approach integrates qubits directly onto silicon chips, creating more robust and reliable quantum computing systems.
Diraq's technology could lower the barrier to entry for organizations looking to adopt quantum computing technology, leading to widespread adoption across industries.
The startup recently secured $15 million in a Series A-2 funding round led by Quantonation, which they aim to use to advance their research in pioneering qubit development.
Diraq's silicon-based qubits offer inherent stability and compatibility with existing semiconductor manufacturing processes.
This compatibility suggests that these quantum components can be manufactured using current facilities and processes, obviating the need for substantial investment in new fabrication plants specifically for quantum chips.
Diraq's approach mirrors the broader trend of utilizing existing semiconductor technologies to advance quantum computing capabilities.
Silicon-based approaches offer a promising avenue for overcoming scalability, stability, and cost challenges in quantum computing.
Diraq's technology holds significant potential across various industries including finance, healthcare, materials science, and logistics.
Diraq's approach diverges from superconducting and ion-trapped qubits pursued by entities like IBM and IonQ.
Diraq's technology could lower the barrier to entry for organizations looking to adopt quantum computing technology, paving the way for widespread adoption across industries.
Healthcare Sector: Trends & Innovations
Medical Imaging: AI chips can accelerate image processing tasks in medical imaging equipment.
Drug Discovery: AI chips can accelerate drug discovery by analyzing vast genomic data to identify potential drug candidates.
Remote Monitoring: AI-enabled devices with advanced chips remotely monitor patients' health, detecting abnormalities for timely interventions.
Neuralink implants wireless brain chip
Neuralink has successfully implanted its brain-computer interface 'Telepathy' into a human, enabling direct communication between the brain and external devices, with potential applications in treating neurological disorders and enhancing human interaction with technology.
The wireless device consists of a chip and over 1,000 ultra-thin, flexible conductors that are inserted into the cerebral cortex by a surgical robot. These electrodes record thoughts associated with movement, allowing users to control digital devices through thought.
The US Food and Drug Administration has authorized human clinical trials for Neuralink's technology. Other organizations, such as the Ecole Polytechnique Federale in Lausanne (EPFL) and Chinese AR technology startup WIMI Hologram Cloud (WIMI), are also making notable strides in the field of BCIs.
The successful implementation of Neuralink's technology could advance the treatment of neurological disorders and enhance human interaction with technology.
The technology addresses the critical need for advanced assistive technologies for individuals with severe motor and communication limitations.
The precision and minimally invasive nature of the implant procedure combined with the high bandwidth of data processing could bring a major leap in the field of neurotechnology.
Future applications of the technology could extend to enhancing cognitive capabilities or treating mental health disorders.
Other organizations, such as EPFL and WIMI, are also making significant contributions to the field of BCIs.
Stanford University researchers have developed a BCI employing a recurrent neural network (RNN) to interpret brain signals into synthesized speech.
These advancements hold promise for applications in healthcare, neuroscience research, and human-machine interaction.
CN Bio unveils liver-on-a-chip
CN Bio, a British organ-on-a-chip (OOC) company, and Altis Biosystems, an American biotech firm, have partnered to develop a human microphysiological system (MPS) that combines their individual models.
The collaboration between CN Bio and Altis Biosystems is seen as an advancement in in vitro human organ modeling for drug development.
The integrated Gut/Liver MPS is expected to enhance the accuracy and predictability of in vitro-derived data.
The primary human microphysiological system offers a promising solution to optimize drug properties, support candidate selection, and inform clinical trial design.
The partnership intends to reduce animal usage in drug development processes.
The integrated system is anticipated to improve its utility in preclinical research and drug development processes.
The collaboration is expected to improve in vivo study design.
The partnership aims to address translatability issues between animal models and humans.
Ceremorphic accelerates drug discovery with AI chip
Ceremorphic, an AI chip startup, is revolutionizing drug discovery with its BioCompDiscoverX platform, which utilizes analog silicon technology to mimic human cells and tissues, reducing development costs and increasing success rates.
The BioCompDiscoverX platform also generates synthetic data for training drug discovery AI models, addressing the current lack of relevant data. It integrates specialized hardware accelerators for molecular dynamics, AI, and reaction processors, tailored to the requirements of drug discovery. This approach offers significant efficiency gains over traditional digital methods, potentially reducing the iteration needed in drug discovery. Ceremorphic plans to demonstrate the technology platform soon and aims for full availability by the end of 2024.
The BioCompDiscoverX platform's capability to generate synthetic data is a significant advantage in addressing the current shortfall of relevant data for training drug discovery AI models.
The integration of specialized hardware accelerators for molecular dynamics, AI, and reaction processors is tailored to the requirements of drug discovery, highlighting a substantial efficiency gain over traditional digital methods.
Ceremorphic's entry into the life sciences with the BioCompDiscoverX platform represents a significant shift in how drug discovery could be approached, promising to address the inefficiency and high costs currently plaguing the industry.
The strategic move into life sciences not only diversifies Ceremorphic's portfolio but also sets a new precedent for the application of semiconductor technology in healthcare.
Ceremorphic's focus on areas with unmet needs such as oncology and neurology demonstrates a commitment to addressing critical healthcare issues.
The accurate representation of biological processes using analog silicon technology significantly increases the success rate of potential drug candidates.
The reduction in iteration needed in drug discovery using Ceremorphic's approach could potentially reduce development costs and time.
Obatala launches obesity-on-a-chip
Obatala Sciences has launched the ObaCell Obesity-on-a-Chip Service, a microphysiological system that mimics human adipose tissue for improved research in metabolic diseases like obesity and diabetes.
The service offers a more accurate representation of human biology, improving the efficiency and effectiveness of preclinical research for therapeutic development.
Obatala's commercial launch of the ObaCell Obesity-on-a-Chip Service addresses the need for better insights and treatments for metabolic diseases like obesity and diabetes.
The company intends to capitalize on the evolving regulatory landscape and has secured a $3 million Series A finance round to support the commercialization of its research-enabling products and platform for drug discovery and development.
The ObaCell Obesity-on-a-Chip Service is a significant advancement in the organ-on-a-chip industry, providing researchers with better insights and data.
The technology overcomes size limitations, enabling the simulation of fasting and feeding in a human model system.
The commercial launch of the ObaCell Obesity-on-a-Chip Service demonstrates Obatala's development of a family of patents licensed from Harvard University in 2021.
The collaboration with Harvard University could fuel further advancements in adipocyte biology and engineering.
Obatala's solutions are aimed at accelerating the study and prevention of diseases in areas such as obesity, diabetes, cancer, and regenerative medicines for pharmaceutical companies, government labs, and researchers.
Automotive Sector: Trends & Innovations
Advanced Driver Assistance Systems (ADAS): AI chips enable real-time processing of sensor data for features like lane keeping, adaptive cruise control, and collision avoidance.
Autonomous Driving: Powerful AI chips process vast amounts of data from cameras, lidars, and radars to enable self-driving capabilities.
In-Vehicle Infotainment: AI-powered chips enhance user experience with personalized content, voice assistants, and seamless connectivity.
Qualcomm has introduced a system-on-chip for ADAS
Qualcomm Technologies has launched the Snapdragon Ride Flex SoC, an advanced system-on-chip that integrates diverse compute resources for in-vehicle computing. The SoC supports mixed-criticality workloads across digital cockpit, advanced driver assistance systems (ADAS), and autonomous driving (AD) functions. It is engineered to meet safety standards and incorporates hardware architecture that ensures isolation, freedom from interference, and quality-of-service (QoS) for specific ADAS functions.
Qualcomm anticipates that the Snapdragon Ride Flex SoC will empower automakers to distinguish their products while ensuring compatibility and interoperability.
The company aims to simplify development processes and expedite time-to-market for automakers and Tier-1 suppliers.
Qualcomm intends to expand the adoption of Snapdragon Ride Flex SoC across diverse vehicle tiers, accelerating innovation and driving growth in the automotive ecosystem.
Intel unveils AI chips for automobiles
Intel has introduced a new family of AI-enhanced software-defined vehicle (SDV) system-on-chips (SoCs). The company collaborated with China’s original equipment manufacturer (OEM) Zeekr to implement these SoCs, delivering advanced AI-driven experiences within vehicles.
Intel introduces AI-enhanced SDV SoCs for various in-vehicle AI applications like driver and passenger monitoring.
The SoCs allow for streamlining legacy electronic control unit architecture to enhance efficiency and scalability.
Collaboration with China’s OEM Zeekr to implement these SoCs for advanced AI-driven experiences within vehicles.
Initiatives extend to creating industry-standard workgroups for automotive power management and developing an open automotive chiplet platform.
Acquisition of Silicon Mobility SAS to enhance position in intelligent EV energy management systems.
Establishment of a committee with SAE International to develop a standard for Vehicle Platform Power Management (J3311).
Texas Instruments Radar Sensor Chips
TI introduces radar sensor chips for satellite radar architectures in January 2024.
AWR2544 enables radar sensors to deliver partially processed data to a central processor for ADAS decision-making through sensor fusion algorithms.
Launch-on-package (LOP) technology shrinks sensor size by up to 30%.
DRV3946-Q1 integrated contactor driver enhances system power efficiency.
DRV3901-Q1 integrated squib driver facilitates intelligent pyro fuse disconnect systems.
Aimed at enhancing automotive safety and intelligence with built-in diagnostics and functional safety support.
AMD AI Engines for Automotive
AMD introduces Versal AI Edge XA adaptive SoCs equipped with AI Engines for AI compute, vision, and signal processing.
Ryzen Embedded V2000A Series processor enables performance and multitasking for infotainment and in-vehicle experiences.
Built on 7nm process technology with 'Zen 2' cores and high-performance AMD Radeon Vega 7 graphics.
Provides high-definition graphics, enhanced security features, and support for Automotive Grade Linux and Android Automotive.
Designed to serve various automotive segments such as infotainment, advanced driver safety, and autonomous driving.
Expansion of the AMD Ryzen Embedded V2000A Series aims to provide consumers with a familiar PC-like experience in vehicles.
OMNIVISION Image Sensor for Vehicle Safety
OMNIVISION launches OX01J image sensor for automotive 360-degree surround-view systems and rear-view cameras.
Designed to offer enhanced imaging performance, compact form factor, and flexibility in system integration.
Complies with ASIL-B safety standards to ensure reliability and safety in automotive applications.
Availability for sampling and mass production by October 2024.
Compatibility with existing ISP architectures provides flexible and customizable solutions for OEMs.
Aims to adapt and enhance offerings to meet evolving automotive industry needs.
NXP Radar Chip for Autonomous Driving
NXP introduces SAF85xx one-chip family integrating high-performance radar sensing and processing.
Offers double the RF performance and boosts radar signal processing speed by up to 40% compared to previous generation.
Enables 4D sensing crucial for safety applications like automated emergency braking and blind-spot monitoring.
Provides OEMs with flexibility to meet safety requirements and increasing demand for radar sensors.
Showcased at Consumer Electronics Show in January 2023 and available for sampling for alpha customers.
Plans to extend application to other industries or develop further iterations for enhanced performance and integration.
Kneron Automotive-grade NPU Chip
Kneron introduces KL730, an automotive-grade neural processing unit (NPU) chip featuring integrated Image Signal Processor (ISP).
Designed to provide secure and energy-efficient AI capabilities across various applications.
Focus on advancing edge AI technology, particularly in natural language processing (NLP).
Secured $49 million in Series B funding for automotive expansion and team growth.
Competing with rivals like NeuReality, Hailo, and Mythic in the AI inference market.
Aims to serve applications in enterprise-edge servers, smart home devices, and advanced driving assistance systems.
Telecom Sector: Trends & Innovations
Network Optimization: AI chips enhance network traffic, prevent failures, and boost performance.
Virtual Assistants: Telecoms use AI chips for cost-effective, user-friendly virtual assistants.
Edge Computing: AI chips at the edge process data locally, minimizing latency for IoT and autonomous vehicles.
EdgeQ debuts converged 4G-5G-AI base station on a chip
The Base Station-on-a-Chip offers the world’s First 4 Carrier Aggregation for 5G.
It enables a 2x improvement in spectral utilization by performing 3 to 4 multi-carrier operations on a 4T4R small cell basis.
The chip integrates unified 4G and 5G Technology onto a single chip, addressing costly transitions from 4G to 5G.
Voice-Over-NR on a Private Network is enabled using the 4G+5G converged system-on-a-chip (SoC), enhancing data and voice services.
EdgeQ introduces a converged 5G+AI Base Station-on-a-Chip, reducing cost, power consumption, and space requirements.
The solution tackles the increasing demand for bandwidth and spectrum efficiency in a future with numerous connected devices.
Picocom unveils SoC for 5G small cell open RAN
PC805 offers an integrated solution tailored for Open RAN architecture, simplifying hardware requirements and supporting various bands.
It interfaces directly with optical distribution units (O-Dus) via open fronthaul, enhancing interoperability and reducing bill of materials.
Picocom introduces a system-on-chip (SoC) 'PC805' optimized for 5G small cell Open RAN radio units (O-RUs), facilitating diverse deployment use cases.
MediaTek introduces generative AI chipset for 5G smartphones
The Dimensity 8300 chipset harnesses GenAI for advanced AI capabilities on 5G smartphones, offering vision AI for image manipulation and personalized suggestion algorithms.
It features an octa-core CPU with four Arm Cortex-A715 cores and four Cortex-A510 cores built on Arm’s latest v9 CPU architecture.
MediaTek launches the Dimensity 8300 chipset with generative AI capabilities for enhanced user interactions on 5G smartphones.
Qualcomm launches satellite IoT chipsets for asset tracking
Qualcomm unveils the Qualcomm 212S Modem and the Qualcomm 9205S Modem, both equipped with satellite capability for industrial applications off the grid.
The chipsets empower IoT devices with satellite communication for optimal off-grid connectivity, supporting Third Generation Partnership Project (3GPP) Release 17 standards.
Qualcomm Technologies partners with Skylo to integrate chipsets with the Qualcomm Aware platform, enabling real-time asset tracking and device management in remote areas.
The chipsets offer global coverage, durability, real-time tracking, and long-term cost-effectiveness for widespread adoption.
NTT introduces prototype chip for data transmission
NTT develops a prototype chip amplifying high-frequency signals crucial for faster data transmission, potentially achieving speeds of up to 2 terabits per second (Tbps).
The chip utilizes indium phosphide material to enhance communication speeds within data centers and across undersea fiber optic cables.
Japanese telecom company NTT's prototype chip aims to improve internet speeds and data transfer rates, benefiting sectors reliant on communication infrastructure.
AMD launches adaptive radio computing chips
AMD introduces Zynq UltraScale+ RFSoC ZU63DR and Zynq UltraScale+ RFSoC ZU64DR devices, expanding its Zynq UltraScale+ RFSoC digital front-end (DFE) portfolio.
The RFSoCs enable the expansion and deployment of 4G/5G radios into markets requiring lower-cost, power, and spectrum-efficient radios.
AMD's radio computing chips aim to meet the evolving needs of the telecom sector, offering cost-effective solutions for enhanced wireless connectivity.
The RFSoC devices are anticipated to enter full production in the second quarter of 2023, supporting diverse telecommunications needs.
Power and Energy Sector: Trends & Innovations
Smart Grids: AI chips can optimize energy distribution in smart grids, balancing supply and demand in real time and reducing energy wastage.
Renewable Energy Integration: AI chips can improve the efficiency of renewable energy sources by optimizing their operation and integration into the power grid.
Integration of Hailo's Processors by Schneider Electric
Schneider Electric integrates Hailo's processors for edge computing applications, enhancing processing power and efficiency for faster and more accurate decision-making in industrial automation.
The integration of Hailo's processors is seen as an advancement in AI capabilities for edge computing applications.
Hailo's processors support a wide range of AI applications, including computer vision, natural language processing, and sensor data analysis.
The integration of Hailo-8 AI processor leads to enhancements in various manufacturing tasks such as object detection, quality control classifications, and waste reduction.
The partnership between Schneider Electric and Hailo Technologies is expected to improve operational efficiency, reduce downtime, and enhance productivity in industrial automation.
Renesas unveils LTE NB-IoT modem chipset
Japanese tech company Renesas Electronics (Renesas) has introduced the RH1NS200, a narrowband internet of things (NB-IoT) chipset tailored for the Indian smart metering market.
The chipset is designed to operate on major Indian telecommunications carriers' networks, addressing the growing demand for smart metering solutions in India. Renesas aims to enable customers to build complete NB-IoT modules, aligning with the "Make in India" initiative.
Traditional NB-IoT chipsets often consume considerable power, limiting their battery life and usability in remote or hard-to-reach areas. Emerging markets such as India offer vast opportunities for NB-IoT deployments, particularly in the smart metering segment, driven by government initiatives and infrastructure development projects.
Renesas targets specific markets such as power and water metering systems, addressing the growing demand for reliable and secure NB-IoT solutions in these sectors. Similarly, in December 2023, STMicroelectronics unveiled an evaluation kit (EVLKST8500GH-2) featuring a smart-grid chipset. This configuration enables hybrid connectivity, combining powerline and wireless communication capabilities.
Portland General Electric pilots smart grid edge AI chip
Oregon utility Portland General Electric (PGE) has piloted US-based technology company Utilidata’s smart grid edge distributed AI platform.
Distributed AI's ability to optimize grid operations in real-time represents an advancement in smart grid technology, offering utilities like PGE better resilience and flexibility in managing their grid infrastructure.
As the energy sector continues to evolve, AI-driven grid optimization solutions are expected to play a vital role in addressing grid challenges and driving sustainable energy transitions.
Utilizing Distributed AI, PGE aims to improve grid resilience, optimize resource utilization, and enhance customer satisfaction, paving the way for broader adoption of AI technologies in utility operations.
It aligns with the industry's shift towards more adaptive and responsive grid management practices.
The anticipated installation of Smart Grid Chips within PGE's Smart Grid Test Bed will offer initial real-time visibility at the grid's edge, supporting PGE's transition towards decarbonization.
Conclusion: Chips of Change
AI chips are likely to revolutionize sectors like tech, auto, healthcare, and energy, with short-term adoption in data centers and automotive advanced driver assistance systems (ADAS), and long-term impacts on autonomy and energy systems.
The transformative power of AI computation, underpinned by the rapid advancements in semiconductor technologies, is steering the next industrial leap. As industries across the board integrate AI capabilities into their operations, the role of high-performance computing becomes ever more critical. The intersection of AI and cutting-edge chip design is not just enhancing efficiencies but also driving innovation in ways previously unimaginable. This synergy promises to unlock new levels of productivity, solve complex problems, and create opportunities that will reshape the economic landscape.
As we stand on the brink of this new era, it's clear that the future belongs to those who can harness the full potential of AI-driven computing. Businesses, investors, and policymakers must collaborate to foster an environment that supports technological advancements and the adoption of AI. The path forward will require continuous innovation, investment in research and development, and a commitment to ethical considerations to ensure that the benefits of this technological leap are broadly shared.
In this dynamic landscape, staying informed and adaptable is crucial. As we witness the dawn of a new industrial age powered by AI and sophisticated chip technologies, the opportunities for growth and transformation are boundless. Embracing these advancements will not only propel industries forward but also pave the way for a future where technology enhances the human experience in unprecedented ways.
For more insights and updates on the evolving world of AI and technology, stay tuned to Curious Compass, where we explore the frontiers of innovation and its impact on our world.
Sub-10nm fabrication refers to the process of creating integrated circuits with features smaller than 10nm. This is challenging due to the need for high resolution and efficiency. Sub-10nm fabrication is often performed using a combination of additive and subtractive methods. All production "10nm" processes are based on FinFET (fin field-effect transistor) technology, which is a non-planar evolution of planar silicon CMOS technology. Sub-10nm fabrication is important for creating smaller, more powerful, and energy-efficient chips for various applications, including AI and machine learning.
EUV (Extreme Ultraviolet) lithography is a cutting-edge technology used in the semiconductor industry for manufacturing integrated circuits (ICs). It uses extreme ultraviolet (EUV) light to create intricate patterns on silicon wafers. As of 2023, ASML Holding is the only company that produces and sells EUV systems for chip manufacturing. EUV lithography enables smaller features to be built on a semiconductor, increasing device area density. It is a projection lithography approach that utilizes 13.5 nm photons to expose photoresist. EUV lithography is widely seen as the next evolution of lithography technology and is enabling the continued extension of Moore's Law.
this is interesting. I see AI infra startups like flex.ai want to build AI-chips for model training. your post goes into details on why startups/companies would want to do that.