Amolak Badesha

5 Reasons Why Legacy Nodes Remain Crucial in the 1nm Era? Legacy nodes Typically refer to older, well-established fabrication processes, often at 45 nm or 28nm and larger. 🚀 Cost-Effectiveness: Manufacturing chips at advanced nodes like 1nm is expensive due to the complexities involved. Legacy nodes are a better choice for cost-sensitive applications. Imagine building a high-performance gaming PC vs. a basic office computer. The gaming PC (1nm chip) is expensive with all the latest tech. But a basic computer (legacy node) is much cheaper to make, perfect for everyday tasks like browsing the web. 🚀 Power Efficiency: Smaller transistors in advanced nodes enable faster performance but often come at the cost of higher power consumption. Legacy nodes, with their larger transistors, can be more power-efficient, a critical factor for battery-powered devices like smartphones and wearables. Tiny transistors in new chips are like powerful engines in cars - they work great but use a lot of fuel (battery power). Bigger transistors in legacy nodes are like fuel-efficient car engines - they might not be the fastest, but they use less battery, ideal for things like smartphones and smartwatches that need to last all day. 🚀 Maturity and Reliability: Legacy nodes have been around for longer and have undergone extensive refinement, leading to highly optimized and reliable processes. This maturity ensures consistent performance and fewer bugs, crucial for applications demanding stability. Think of a recipe you've made many times. You know exactly how much of each ingredient to use and how long to cook it for. Legacy nodes are like that tried-and-tested recipe - they're reliable and we know how to make them work perfectly for everyday electronics. 🚀 Compatibility: Not all applications require the bleeding-edge performance offered by 1nm chips. Legacy nodes can integrate well with newer technologies, enabling efficient design of complex systems with a mix of performance requirements. Imagine building a house - you might need different materials for the roof, walls, and foundation. Legacy nodes can be like those different materials - they work well with newer technologies to create all sorts of electronic devices, like a phone with a powerful processor (1nm chip) but a low-power Bluetooth chip (legacy node) for connecting to headphones. 🚀 Internet of Things (IoT): The vast and growing world of IoT devices, many of them simple sensors, often doesn't need the processing power of 1nm chips. Legacy nodes are perfectly suited for these low-power, low-cost applications. Tiny sensors in things like fitness trackers or smart thermostats don't need the processing power of a fancy 1nm chip. Legacy nodes are the perfect tool for these smaller jobs, keeping costs low and battery life long. Guess, Which Country is investing huge in legacy Nodes? Tell us in the comments. A detailed post is in comments. For all semiconductor and AI content, follow TechoVedas

77

3 Comments

I have finally delved into this year's "100 Silicon Startups Worth Watching in 2024," recently published by EE Times | Electronic Engineering Times. One insight from Peter Clarke in his preface is the interesting shift in the geographical distribution of these top 100 companies. While there's a slight decline in California and North American representation (is it real?), the most striking changes are the sharp decrease in Chinese companies and the meaningful rise in European entries. Peter attributes the Chinese decline to trade tensions: "#China continues to have high levels of startup activity, but its reduced presence on the Silicon 100 reflects the trade tensions between the U.S. and China. That friction has intensified in recent years and has hit a number of Chinese startups that may have had international aspirations but have been denied access to leading-edge silicon out of foundry TSMC." #Europe, on the other hand, has seen an increase from 23 to 29 startups, with the U.K. leading the charge: "The U.K. has 12 startups on the current list, up from nine startups in version 23. The U.K., with longstanding entrepreneurial bases in Cambridge, Oxford, Bristol, and Edinburgh, has enjoyed strong representation for many years." In my view, this growth is a testament to #Europe's robust academic system, coupled with government support, investments, and incubators like Silicon Catalyst.UK. Israel continues to excel: "Israel has always punched way above its weight by population and continues to do so," with its representation growing slightly from 7 to 8 this year, including NeuReality*, POLYN Technology, NeoLogic*, Speedata.io, Pliops, Chain Reaction Ltd.*, Hailo, .Quantum Machines . (*=new additions to the 2024 list). This list provides a good overview of this dynamic and evolving industry, blending both emerging and more established technologies. It's a must-read for anyone interested in the future of #semiconductor #innovation. For the full list: https://lnkd.in/dgfrNUWX

50

6 Comments

According to a new report by the Wall Street Journal, TSMC and Samsung have been in talks with the UAE to explore building megafactories in the Middle East, on the same scale as those found in Taiwan. This could potentially reshape global supply chains. The question is whether the UAE can leverage its strategic partnerships with both the U.S. and China to become a key player in this high-stakes race. #Semiconductors #TSMC #Samsung

62

A report in the #NYTimes today indicates stimulus from #CHIPS Act already appears to be taking effect.  Done right, we could see more than 30% of the most advanced logic devices manufactured domestically by 2032, up from basically zero today. The result is a change in the total percent of semiconductor devices changing from a downward trend towards less than 10% of global #semiconductormanufacturing to more than 14%. While this may not seem like a major change, consider that all global regions are simultaneously investing heavily in production and we begin to understand how significant this reversal actually is. The National Semiconductor Economic Roadmap (#NSER, https://lnkd.in/gtSGMNjt) modeled and predicted these trends almost 18 months ago (Exhibit 3, below) based upon US leadership in manufacturing capacity investments, and how supply chain, entrepreneurship, infrastructure, and workforce development would be the cornerstones of this resurrection. We’ve seen multiple incentives planned, the directives made for a National Semiconductor Technology Center (#NSTC, #NATCAST) and National Advanced Packaging Manufacturing Program (#NAPMP), supplier funding incentives being communicated, and most recently the announcement by #NIST of the CHIPS Manufacturing USA Institute (https://lnkd.in/g_wb5dXG) dedicated to advancing innovation and productivity through digital twinning across the ecosystem to create a global digital twin for manufacturing, arguably this industry’s moonshot! Let’s align around the spirit of this stimulus and regain our leadership position in semiconductor manufacturing. https://lnkd.in/gwCzSJM5

33

5 Comments

Tackling Data Retention in SSDs (by Anil Burra from Intelligent Memory) 🤩 "SSDs have conquered the market due to their speed, durability, energy efficiency, and small form factor. However, write endurance and data retention are concerns in #embedded and #industrial applications, especially as the technology moves to ever smaller nodes." (Anil Burra) With the scaling of #NAND #flash #technology to higher densities (e.g., Triple-Level-Cell or Quad Level-Cell) and smaller process nodes, data retention becomes more challenging due to increased cell-to-cell interference and reduced margins for charge storage. Therefore, newer #technologies may have shorter specified data retention periods compared to older, more robust #SLC (Single-Level-Cell) or #MLC (Mul ti-Level-Cell) #NAND products. Two key factors affect data retention, the period over which #flashmemory can store #data without loss or corruption, in #NANDflash cells: program/erase (P/E) cycling and operating temperature. A very big thank you to Anil Burra from Intelligent Memory for this insightful article via Electronic Specifier Ltd with more background and insights via the link below 💡🙏👇 - A very good read to an important TOPIC 🤩 https://lnkd.in/eqdH7-N2 #semiconductorindustry #semiconductormanufacturing #innovation #it #ai #ssd #datastorage #automotiveindustry #embeddedsystems #ic #computer #computing #iot #memory #chip #chips #technology #tech

50

1 Comment

Stitching vias, also known as via stitching, are a crucial design practice for PCBs. By placing a series of vias around the perimeter of a ground plane or signal trace, a continuous conductive path is created, connecting the top and bottom layers of the board. #PCBdesign #viastitching

1

Jim Keller, former Intel, AMD, and Tesla engineer, aims to challenge Nvidia's dominance in AI with Tenstorrent's energy-efficient chip design. Tenstorrent's upcoming second-generation AI chip promises superior energy and cost efficiency, presenting a viable alternative to Nvidia's offerings. The innovative design of Tenstorrent's chip features independent decision-making cores, potentially reshaping AI chip technology across diverse applications. #AI #Nvidia #semiconductor #semiconductors #chip #technology #innovation #research #computing Tenstorrent NVIDIA https://lnkd.in/g6a3vAu4

31

Here’s why SemiQ made my Top 10 Power Semi Startups to Bet On list… In today's fast-paced world, the demand for power is skyrocketing and advanced semiconductors like silicon carbide (SiC) are revolutionizing efficiency. SemiQ is one of the “mature” startups on my list, but their longstanding commitment to SiC reflects how integral they’ve been to the advancement of this material. Founded as Global Power Technologies in 2007, it started as an integrated development and manufacturing company to grow its own SiC epitaxial wafers. However, it was in 2008 when Timothy Han joined the company that they began to make significant strides. Tim was leading product development and his team, along with engineers at Oak Ridge National Lab, built the world’s first 3-phase motor driver inverter using SiC that achieved around 97% system efficiency. That unprecedented result revealed the incredible potential of SiC and everyone at GPT believed it would help solve the growing global power consumption crisis but there was a lot of work to be done to grow its adoption. Many of the manufacturing processes used for SiC were early in development and customers were wary of its quality. GPT appointed Tim to be President and Senior Director of Technology development and with his guidance, GPT shifted from developing systems to a fabless model and focused on improving supply chain resilience and building world-class quality management processes. That vision took GPT major investment and many years to realize but as a result, they were able to begin serving markets like hi-rel, medical, and automotive, where quality and reliability are crucial. With a tailwind behind them, GPT announced another shift in early 2020 - rebranding to SemiQ - the new name showing their unwavering commitment to SiC and quality. Since then, they’ve continued to innovate in design AND manufacture. They’ve developed a line of high-voltage SiC MOSFETs that can operate up to 1200V and have built an in-house reliability and quality lab that enables them to learn best practices for SiC production. As a result, they launched a Known Good Die program this year. I think this continued focus on quality holds the potential to drive the power industry forward since it goes such a long way in terms of building confidence in customers. If SemiQ can continue demonstrating superior performance in terms of DFM and reliability, the pace of SiC’s adoption will be quickened. It will also have an effect on engineers who want to see their design come to life, rather than be put on a shelf to collect dust with other IP that can’t be produced. My advice to the Talent team and leadership at SemiQ is “lean in” to what makes your company great. While the rest of the industry celebrates design and manufacturability as an afterthought, it is your superpower so be proud of that and make sure everyone knows about it… especially potential talent. #semiconductorindustry #siliconcarbide #powermanagement

10

✅ Gear Up for Growth: Automotive IoT Market Expands Globally Download PDF Brochure @ https://lnkd.in/fEaeE55 The #automotiveIoT market is projected to grow from USD 131.2 billion in 2023 to reach USD 322.0 billion by 2028; it is expected to grow at a Compound Annual Growth Rate (CAGR) of 19.7% The growth of the automotive IoT market is driven by the increasing number of regulations mandating advanced features in #vehicles for enhanced user comfort, safety, and convenience coupled with growing use of telematics and user-based insurance programs are the major factors driving growth of the #automotiveIoT market. The #software segment held the largest share of the overall automotive IoT market, in terms of value, in 2022. The software market includes platforms and solutions which are crucial to the functioning of IoT in an automobile. By connectivity form factor, integrated system segment is projected to exhibit highest CAGR for the automotive IoT market during the forecast period. The North American region is expected to hold the largest share of the automotive IoT market during the forecast period. The US is a major contributor to the growth of the automotive IoT market in North America. North America has been a pioneer in adopting most technological innovations and technological advancements, so the adoption of IoT in North America in the automotive sector has been profound. The region is home to the big 3 of the automotive industry—Ford Motor Company (US), General Motors (US), and Fiat Chrysler Automobiles (US). Key players in the automotive IoT market include NXP Semiconductors (Netherlands), Harman (US), Robert Bosch (Germany), Thales (France), TomTom International (Netherlands), IBM (US), Geotab Inc. (Canada), Texas Instruments (US), Intel Corporation. (US), Eurotech (Italy), STMicroelectronics (Switzerland), Renesas Electronics , Infineon Technologies (Germany). Apart from this, Airbiquity(US), Qualcomm(US), Visteon (US), Vodafone Group (UK), Microsoft Corporation (US), Alphabet Inc. (US), AT&T& T (US), CloudMade(UK), Sierra Wireless (Canada)  #iot #automotive #Technews #IoTsoftware #IIoT #Telematics #automobile

3

A very interesting blog by Giuseppe Piccolboni of Weebit Nano Ltd about the use of our #ReRAM for #neuromorphic applications

36

Before I worked in the #semiconductorindustry, I worked as a civil engineer. I found this article about CalTrans using #AI to answer complex traffic-related questions fascinating. "Caltrans says the data generated from traffic sensors and cameras — coupled with the continuous velocity of data generation and the variety of videos, images, log files, third-party data streams and guidance — poses challenges for humans to digest and utilize." Sounds a bit like using #AI and #ML for smarter #scheduling in a #waferfab, doesn't it? More #industrialengineering than civil engineering.

17

Building a semiconductor startup is extremely difficult and most technical founders are incredible engineers, but horrible at running businesses. This is why, according to Srini Ananth, Managing Director at Intel Capital, one of the key predictors of success for a semiconductor startup is who its founders choose to surround themselves with. Srini has evaluated hundreds if not thousands of potential investment opportunities. This has given him a bird’s-eye view of our ecosystem and enabled him to spot and invest in amazing semiconductor startups like Ayar Labs, Eliyan Corporation, and Astera Labs. In a recent conversation with Srini, we discussed how he evaluates founders for their potential to succeed. It’s his belief (and I agree) that in addition to being experienced and having a firm grip on the reality of the difficulties, founders should also be open to bringing on others—even if that means stepping back as the founding CEO and letting someone else take the reins. It comes down to having the right composition of team and talent to break through barriers, find commercial success, and generate that all-important momentum that ultimately transitions the company from raising funds to generating revenue and profitability. I’m grateful Srini shared this insight with me and would love to hear YOUR thoughts. Do you think there’s a different key predictor of success for startup founders? Share your perspective below and bonus points for contrarian views that cause us to reexamine “standard thinking”. #semiconductorindustry #startups #investing

39

3 Comments

4 Major Segments in Semiconductor Value Chain: A Breakdown 1. Chip Design 🚀 Fabless Companies: These companies design chips but do not manufacture them. They often license their designs to foundries for production. Examples include Qualcomm, NVIDIA, and MediaTek. Their profit margins can vary widely depending on the complexity of their designs and their market position. 🚀 Integrated Device Manufacturers (IDMs): These companies design and manufacture their chips. Examples include Intel, Samsung, and TSMC. Profit margins are typically higher than of fabless due to their vertical integration. 🚀 Electronic Design Automation (EDA) Tools: These software tools are essential for designing chips. Companies like Synopsys, Cadence, and Ansys provide these tools. Their profit margins are relatively high due to the specialized nature of their products. 🚀 IP Cores: These are pre-designed functional blocks that can be incorporated into chips. Companies like Arm and Imagination Technologies provide IP cores. Their profit margins can be substantial, especially for widely used cores. 2. Fabrication (Fab) 🚀 Foundries: These companies manufacture chips based on designs provided by fabless companies or IDMs. Examples include TSMC, Samsung Foundry, and GlobalFoundries. Their profit margins can vary depending on factors like manufacturing capacity, process technology, and customer mix. 🚀 Raw Materials and Chemicals: Suppliers of materials like silicon wafers, chemicals, and gases play a crucial role in the fabrication process. Their profit margins can be relatively low due to the competitive nature of the market. 🚀 Equipment Suppliers: Companies like AMAT, ASML, and KLA Corporation provide the equipment used in fabs. Their profit margins are typically high due to the specialized nature of their products and the high barriers to entry. 3. Packaging, Testing, and Assembly (OSAT) 🚀 OSAT Companies: These companies package and test chips. Examples include Amkor, ASE . Their profit margins can vary depending on the complexity of the packaging and testing processes. 🚀 ATE Manufacturers: Companies like Advantest and Teradyne provide automated test equipment used in OSAT facilities. Their profit margins are generally high. 4. End-Use Customers 🚀 Original Equipment Manufacturers (OEMs): These companies incorporate chips into their products, such as smartphones, computers, and automotive electronics. Their profit margins can vary widely depending on the industry and their competitive position. 🚀 Board Assembly and EMS: Companies that assemble chips onto PCBs and manufacture final products are typically part of the broader electronics manufacturing services (EMS) industry. Profit margins can be relatively low due to the competitive market. Add more in comments. For all semiconductor and AI related content, follow TechoVedas ---- Read "The Semiconductor Saga"- A book that teaches you everything about semiconductors in simple words. Link in comments.

165

7 Comments

Crowbars - a further layer of protection in power electronics. By definition, a crowbar is a steel bar, usually flattened and slightly bent at one or both ends, used as a lever to pry things open. In power electronics, the word became a metaphor for an electric approach that looks like using brute force to handle failure-events by initiating a controlled overcurrent situation and at the same time, safely trigger superordinate fuses or protective switch gear. Protecting electronic devices is an omnipresent task for electrical engineers but is not limited to single applications. Far larger installations like industrial areas, solar inverters for utilities, multi access-point EV-charging stations or train- and subway-train stations need to be protected from hazardous conditions as well. Classical fuses may not be able to protect power semiconductor switches when dealing with high power levels. Crowbars can be used to clear short circuits by diverting currents through themselves, thus giving enough time to fuses or protecting switch gear to disconnect the short circuit's cause.  In addition, crowbars can be triggered in case of overvoltage and act as a supporting safety measure as well. More details on crowbar design and an explanation, why a crowbar works fine even if it gets destroyed can be found in the corresponding whitepaper available on the Littelfuse website: https://lnkd.in/eCwBUXgr

22

1 Comment

Excellent post by an Excellent Engineering Leader I use to work with Sunil Shenoy. I’ll also steer the conversation into targeted intentional use cases as opposed to just the continuous evolution of CPUs vs GPUs vs TPUs or better yet scalable Neural Processor Units (NPUs). If you’re training large AI models in the datacenter that doesn’t mean they will run inferencing at anywhere near the same size on edge deployed models. We should certainly be focusing on being responsible with our data accuracy, power consumption and latency relative to the use case. It’s responsible to target under 1kW for L3 targeted vision systems (to extend battery range) so long as the accuracy can be spot on (even in low light or noise heavy scenarios). That’s just an example but another would be egregious GPUs being used in thinks like light drones when you really can enable the drone to fly hours longer with more efficient chipsets focused on things like FFT on VDSPs and Vision Transformers on NPUs. No CPU (RISCV or not) is running a majority of those algorithms or AI models efficiently but you still need them for running functionally safe RTOS’s or as the host controller. So my original point, we need to responsibly build. Pick what’s needed for the use case. To Sunil’s point though, the spend is rising in datacenter. However for things that need to be more smart with or without connectivity to the cloud, AI edge compute should be targeted, efficient yet flexible. Enter Chiplets conversation for Pt. 2.

7

1 Comment

Hello everyone, Today, we delve into Demo17, focusing on the utilization of EIQ (Enhancement Image Quality) for measurement operations. The prevailing sentiment among Fab metrologists is cautionary, highlighting the potential risks associated with EIQ, citing its propensity to generate "artifacts" and, in certain instances, compromise precision and accuracy metrics. Rather than engaging in direct rebuttal, I aim to look from the angle of "practicality" on the current landscape. Firstly, it's imperative to recognize that all OEM manufacturers employ some form of EIQ to ensure uniformity in image presentation. Thus, assuming collected images reflect pure detection paths can be misleading. EIQ implementation occurs at various levels, be it within software algorithms or through bandwidth filtering on the hardware's data collection end. Regarding artifacts - lets be honest image processing evolved a lot over the past 30 years. It looks very similar to contemporary AI models refining their ability to suppress "hallucinations." Crucially, when conventional hardware setups prove inadequate for precise measurements, resorting to EIQ becomes indispensable. In scenarios where the choice is between "Skip on Meas" or use EIQ (despite potential compromises in precision/accuracy), the latter remains a pragmatic solution. For many manufacturers, investing in advanced hardware not financially viable, enhansing the significance of EIQ use within measurement routines. The notion of "artifacts" loses significance when we are in practical dilemma of getting measurements that keeps overall line healthy. The Measurement Utility offers an extensive array of EIQ mechanisms, bridging the gap where OEM software falls short. In the accompanying video, we examine images from the P807 NOR process, illustrating how variations in etch times can impact contact dimensions. Through Laplacian/Gaussian filtering, we establish stable mechanisms for excursion prevention, even amidst environmental disparities between etchers. Furthermore, supplementary measurements on underlayer M2 lines facilitate straightforward calculations to derive local overlay metrics. For those seeking metrology solutions without the burden of substantial capital expenditure, I endorse reliable measurement systems equipped with versatile capabilities. Such investments, albeit modest, yield substantial returns in terms of yield enhancement and operational efficiency. #gazadelendaest #deeplearning #resolution #beam #EHAR #DeepStructures #metrology #Fab #problemsolving #LVTailoring #innovation #measurement #SecondaryElectronDetection #BackScatteredElectronDetection #CriticalDimentionMetrology #noisereduction #artificialneuralnetworks #CAD2SEM #Die2DB #EPE #PSD #precision #accuracy #lithography #tmu #NoiselessPSD #TargetDesign #ai #BlindDenoising #DeepStructures #neuralnetworks #weave #MassMeas #defects #beam #SecondaryElectronDetection #Overlay #EPE #labview #python #noisereduction #appliedmaterials #hitachi #kla

11

Great OFC coverage by LightCounting. TeraSignal’s Intelligent Re-Driver in CMOS is changing the power, reliability and performance calculus between DSP and LPO modules.

27

NEED PRECISION OPTICAL METROLOGY? AOM IN TUCSON, AZ, HAS ANSWERS. I’ve just been talking to Tyler Steele, part of the dedicated team working with AOM - Arizona Optical Metrology LLC. Established in 2009, they are well-known globally as a leading computer-generated hologram (CGH) supplier for the world’s most challenging optical engineering projects. Often, metrology is the limiting factor in fabricating complex optical surfaces, as methods for qualification and feedback are not precise, flexible, or capable enough. A CGH overcomes many of these surface testing challenges to make optical surface qualification feasible for most complex surfaces while maintaining precision, resolution, and speed. EXPANDING HORIZONS Now, AOM are reaching out to bring CGH technology to a broader set of optics and photonics applications. Complex optics are being used more widely than ever–and at precision and production volumes that seemed impossible just a few years ago. The challenges these applications face are on par with the biggest space telescope projects of the past, but with a fraction of the budget and extremely tight timelines to complete. Think of the volume market for AR glasses. AOM - Arizona Optical Metrology LLC is delivering new and novel solutions to optics projects and programs to address this new world of optics production. They enable a new level of precision and ease of use for complex optics metrology from surface figure error of aspheres, freeforms, and off-axis sections to optical system alignment. A CGH converts an interferometer regular wavefront (collimated or spherical) into a custom-designed wavefront that matches the unique shape of the surface to be tested. The result is an interferometric null test when the test optic is well aligned to the CGH. In my conversation below, Tyler answers the Optica question: What can you do for other Optica corporate members, and what can they do for you? Quite a lot, it seems. Together, the future of optics is very bright indeed! Contact them directly or drop me a line for an introduction.

51

6 Comments

This is just going to speed up the adoption of risc v and the eventual death of arm Qualcomm will likely take a 18 months hit for their short term stock price, however they will likely ramp up internal resources to be familiar and likely adopt RISC v fully that will force better manufacturability with tsmc and other foundries Arm to Cancel Qualcomm Chip Design License in Escalation of Feud https://lnkd.in/e_AAvxRs

7