Aditya Bansal

Really interesting to see this: how Europe is moving up. It is interest to contrast this qualitative survey with the #siliconcatalyst analysys that UK is sixth in Europe in funding value fir silicon investment. And I will put in a plug for our #UKTIN Forward Capabilities paper on recommendations for UK Semiconductor sector: https://lnkd.in/eNNAsCAz

8

September 02 2024 – 5 Notable Articles To Read Article: Chip challengers try to break Nvidia’s grip on AI market Takeaway: Cerebras Systems Inc., d-Matrix, and Groq are part of a group of smaller companies seeking to capture a portion of the multibillion-dollar AI chip market, which Nvidia has largely dominated with its graphics processing units (GPUs) during the initial wave of investment. Source: Financial Times Link: https://lnkd.in/d8yh-VVU Article: ‘T-Bill and Chill’ Is a Hard Habit for Investors to Break Takeaway: Money-market funds surge to a record $6.24T, and while upcoming Fed rate cuts suggest a shift may be warranted, the timing remains uncertain. Source: Bloomberg Link: https://lnkd.in/dp6wvceH Article: Why Costco Wholesale Could Be the Next to Split Its Stock Takeaway: The price increases are likely a key factor driving the stock splits, which make shares more accessible to retail investors and, some believe, psychologically more appealing to potential shareholders. This context highlights Costco, which has surged nearly 40% since the start of 2024, with its last split occurring in 2000, just a year after Walmart’s previous split. Source: Barron's Link: https://lnkd.in/dyPpVbAq Article: Meta Shuttering Augmented Reality Studio Amid Pivot to AI Takeaway: Meta Spark was originally intended to compete with Snapchat and the company will shift resources to products such as smart glasses. Source: Bloomberg News Link: https://lnkd.in/daEY47n7 Article: AI Regulation Is Coming. Fortune 500 Companies Are Bracing for Impact. Takeaway: Approximately 27% of recent filings with the Securities and Exchange Commission cited AI regulation as a risk, marking one of the clearest indications yet of how AI rules could impact businesses. Source: The Wall Street Journal Link: https://lnkd.in/dS977wWK #news #research #business #ai

6

Some investors bet against Nvidia, expecting AI bubble to burst As chip designer Arm discovered recently, a company’s share price can take a hit once traders realize that their estimates of continued strong growth are not going to be met. Additionally, there is also a sentiment from some corners of industry that we are in an AI bubble which will burst sooner or later, with The Guardian newspaper recently claiming that "nothing grows exponentially forever. So, are we caught in an AI bubble? Is the pope a Catholic?" Economic research biz Capital Economics has even put an approximate date on this, predicting that the AI-fueled stock market bubble will burst in 2026, after first driving the S&P 500 to a high during 2025. Those are my 'rational' views of an AI bubble burst 1- Linearity: 30% of real world problems are linear in behavior. AI is essentially a huge linear system, and is not suited for all problems. The market will look for other answers, AI is limited. 2- Bias: 0,1 % of bias in the data is enough to offset the model. A majority of problems are dealing with images as data source, where I identified multiple bias sources, non linear in nature on top of it. When I achieve 30% added lossless compression, the excess compressed file data is all the added bias to the image content. 3- Computing speed: Large scale systems have a n² solution dependency at best. When cutting down variables in 2, solution speed increases by 4... Every modeling problem has a deterministic part (science, law, eng...) and a random part (unknown, noise, random law behavior...). AI does not need to learn what is already known. In fact, known behavior should be removed from the AI model space to lower the number of variables and speed learning, as well as improve prediction. My breakthrough on #digitalimaging will even upset #AI

4

We are excited to back Morphing Machines. As Dr. Ranjani, CTO and co-founder says: “For the first time in India and globally, we have a processor that is truly runtime reconfigurable, catering to emerging workloads.” While Nvidia’s growth and high valuation are driven by robust demand for “infra-compute” in high-performance GPUs, spurred by Gen AI, there is a distinct need for innovative solutions tailored to a new wave of reactive applications. Enter Morphing Machines. It addresses this demand with its unique product designed for next-gen telecom, autonomous vehicles, collaborative robots, avionics, and online data analytics. This product is specifically engineered to handle heterogeneous workloads and diverse tasks with varying performance demands simultaneously, ensuring a high level of scalability. As we move from an institutionally-backed research team at IISC to productization, Golden Sparrow is collaborating with Speciale Invest and others to support Deepak Shapeti, who joined as CEO three years ago after earning his Masters at Stanford. He brings not only technical expertise but also the business acumen necessary to guide the product to commercial success. Venkat Raju Afreed Faizan Anand Rao Ritik Singh Michael Marmor

68

5 Comments

We've been quietly building something behind the scenes … A patent fortress around our technology! When people discover the extent of our startup’s patent portfolio, the question that often follows is, "My lawyer says patents are costly. How did you manage to acquire so many?" The answer? It all comes down to a few key principles. And the most important one might not be what you expect. It's not just about having deep pockets or a team of legal experts. (Though at a minimum you do need at least one strong internal champion and a great lawyer). It's not even about having the most groundbreaking ideas. The real secret to building a strong patent portfolio starts with something much simpler, something that's often overlooked in the rush to innovate... 𝗙𝗼𝗰𝘂𝘀. Specifically, the ability to 𝗙𝗢𝗖𝗨𝗦 𝗢𝗡 𝗖𝗟𝗔𝗥𝗜𝗧𝗬. Let me explain... 𝙔𝙤𝙪 𝙢𝙪𝙨𝙩 𝙗𝙚 𝙖𝙗𝙡𝙚 𝙩𝙤 𝙚𝙭𝙥𝙡𝙖𝙞𝙣 𝙮𝙤𝙪𝙧 𝙞𝙣𝙫𝙚𝙣𝙩𝙞𝙤𝙣 𝙞𝙣 𝙖 𝙨𝙞𝙣𝙜𝙡𝙚 𝙨𝙚𝙣𝙩𝙚𝙣𝙘𝙚 𝙩𝙝𝙖𝙩 𝙘𝙡𝙚𝙖𝙧𝙡𝙮 𝙤𝙪𝙩𝙡𝙞𝙣𝙚𝙨 𝙞𝙩𝙨 𝙪𝙣𝙞𝙦𝙪𝙚𝙣𝙚𝙨𝙨. Can’t do it? You're not ready for a patent. You also need to be adaptable. You must be willing to narrow down your claims or even be prepared to abandon an application altogether. It's a tough decision, but it's important to focus your resources on the ideas with the most potential. Ultimately, a strong patent portfolio is about more than just legal protection. It's a statement about your commitment to innovation and a powerful tool for driving long-term growth. So, how many patents does our startup hold? 🥁🥁🥁 A total of 𝟳𝟮 𝗴𝗿𝗮𝗻𝘁𝗲𝗱, with over 𝟮𝟬 𝗺𝗼𝗿𝗲 𝗽𝗲𝗻𝗱𝗶𝗻𝗴. But here's the kicker: we're not just collecting patents, we're monetizing them.  More on that in a future post – stay tuned!

63

3 Comments

MSR has showcased BitNet, a novel 1-bit Transformer architecture, designed to address the challenges associated with scaling LLMs. A key innovation in BitNet is the introduction of BitLinear, which replaces traditional linear layers and facilitates the training of 1-bit weights from the ground up. This architecture not only achieves competitive performance in language modeling but also significantly reduces memory usage and energy consumption when compared to advanced 8-bit quantization methods and FP16 Transformer baselines.   The architecture employs quantization-aware training, optimizing the model for low-precision representation from the outset, which enhances both accuracy and stability during the training process. BitNet features stacked blocks of self-attention and feed-forward networks with binarized weights, while retaining higher precision for other components, thereby minimizing computational costs. This design allows for efficient matrix multiplication with 1-bit weights, resulting in notable energy savings, especially as model sizes increase.   The training methodology incorporates techniques such as the straight-through estimator for gradient approximation and mixed precision training. These approaches enable BitNet to converge effectively with larger learning rates compared to traditional FP16 Transformers, leading to improved training efficiency and stability, which is particularly beneficial in resource-constrained environments.   Comparative evaluations indicate that BitNet consistently outperforms existing quantization methods across a range of benchmarks, including zero-shot and few-shot tasks, while also maintaining lower inference costs. These findings highlight the advantages of quantization-aware training over post-training quantization, positioning BitNet as a prominent approach for developing efficient LLMs.    Arxiv: https://lnkd.in/e-D959-U

3

I am advising a pre-A deep tech company that is revolutionizing the audio industry by developing an SDK for GPU (NPU/TPU) chips found in phones, laptops, cars, and the cloud. They foresee a trend called Accelerated Computing, similar to the transformation seen in graphics and machine learning, but for audio. This shift will revolutionize how we record, edit, perform, perceive, and experience audio. Currently, audio is static, blended, and plain. However, with real-time processing, it can become dynamic, immersive, spatially directed, and emotionally impactful. The company has: ✔️ A legal entity, cap table, and IP in Switzerland, with plans to restructure and move to Delaware in preparation for Series A. ✔️ A team of 16 members, including 10 PhDs in computer science or applied mathematics. ✔️ A very sexy business model based on licensing fees, allowing them to sell to enterprises without dealing with customer-facing processes such as sales and support. ✔️ A main product, the SDK, enables advanced computing on devices like GPUs, offering capabilities, features, and functionalities that current hardware and systems cannot implement or run at the user end. ✔️ Proven success in the Pro Audio industry, with entry into the Automotive and Game Development sectors. They also have a promising future in the Entertainment industry. Previously raised $8M+ led by the top-notch VC, and are successfully closing pre-A funding of $2.5M. We are onboarding 50% of the round and have oversubscribed the rest. However, until the funds are in the bank account, we can negotiate with the founder. I have a small allocation for my community of $300k. DM me if you want to see my investment memo.

19

2 Comments

🚀 𝐓𝐨𝐩 𝐒𝐞𝐦𝐢𝐜𝐨𝐧𝐝𝐮𝐜𝐭𝐨𝐫 𝐬𝐭𝐚𝐫𝐭𝐮𝐩𝐬 𝐢𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐧𝐠 𝐚𝐧𝐝 𝐥𝐞𝐚𝐝𝐢𝐧𝐠 𝐢𝐧 𝐄𝐝𝐠𝐞𝐀𝐈 🔥 Since 2020, a wave of startups has emerged to push the boundaries of EdgeAI by developing cutting-edge semiconductor solutions. These companies are transforming how we interact with AI in real-time, unlocking new possibilities for industries like autonomous vehicles, smart cities, and IoT. 🌐 Here’s a look at some of the trailblazing startups making waves in this space: 1️⃣ Mythic - Revolutionizing analog computing to bring faster and more efficient AI processing to edge devices, with applications like smart cameras and voice assistants. 📷🗣️ 2️⃣ Hailo - Developing AI accelerators that bring deep learning to the edge, reducing latency and enabling smarter autonomous systems. 🚗🏙️ 3️⃣ Untether AI - Their AI chips are speeding up edge inference by embedding memory directly into processing units, a game-changer for industrial robots. 🤖⚙️ 4️⃣ Syntiant Corp. - Focused on ultra-low power deep learning chips for voice and sensor applications, making always-on AI a reality. 🗣️🛠️ 5️⃣ EdgeQ Inc. - Bridging 5G and AI to transform networks, bringing intelligence to base stations and IoT infrastructure. RISC-V based architecture. 📡🌐 6️⃣ SiMa.ai - Simplifying the deployment of ML at the edge with their innovative software-centric approach, ideal for computer vision, robotics and surveillance. 🤖👁️ 7️⃣ Axelera AI - High-performance AI chips that drive edge computing solutions for computer vision, security and industrial automation. 🛡️🏭 8️⃣ EnCharge AI - Rethinking memory integration to supercharge real-time decision-making in drones and autonomous vehicles. 🚁🚙 9️⃣ Tenstorrent - Enabling fast, efficient edge AI processing for robotics and smart manufacturing, helping businesses scale their AI solutions. 🤖🏭 🔟 GrAI Matter Labs - Inspired by the brain’s event-based processing, their chips are perfect for low-latency video analytics and smart cameras. 🎥🧠 🔑 EdgeAI and Computer Vision isn't the future; it's happening now! These startups are leading the charge to redefine the possibilities of AI at the edge. 🎯 Keep an eye on them as they shape our tech landscape! #Semiconductor #ComputerVision #EdgeAI Video Credit: Generated using Pika.art

3

The lack of part standardization has plagued the electronics industry for years. Every WANTS standardization... but nobody wants to do the work of pouring through hundred page datasheets to manually create parts... sounds like a perfect job for AI! 🤖 I'm excited to announce that Flux Copilot can now read datasheets and generate parts in any format unlocking JEP30 standards for manufacturers and hardware companies that the industry has been trying to adopt forever.

10

"Why can't it be done in 1 month under $10K?" A non-tech founder recently asked me this while discussing an IoT hardware product proposal.   It’s a great question—but I wouldn’t recommend cutting costs in new product development. Because you’re not really saving; you’re simply postponing costs, letting them compound over time. Here’s a breakdown of the key cost and time drivers:   1️⃣ Specification & 𝗥𝗲𝗾𝘂𝗶𝗿𝗲𝗺𝗲𝗻𝘁 𝗙𝗶𝗻𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 2️⃣ 𝗥&𝗗 - Design, PCB, Firmware 3️⃣ 𝗣𝗿𝗼𝘁𝗼𝘁𝘆𝗽𝗶𝗻𝗴 𝗮𝗻𝗱 𝗠𝗩𝗣 4️⃣ 𝗧𝗲𝘀𝘁𝗶𝗻𝗴 & 𝗤𝗔 5️⃣ 𝗦𝗰𝗼𝗽𝗲 𝗖𝗿𝗲𝗲𝗽 6️⃣ 𝗖𝗲𝗿𝘁𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 Skipping any of these steps will cost you more in time and money than what you initially saved. On the flip side, 𝗣𝗮𝗿𝗸𝗶𝗻𝘀𝗼𝗻'𝘀 𝗟𝗮𝘄 suggests that work expands to fill the time allocated to it. If my answer is always, "this is what it takes", I won't look for ways to innovate.   So my question: how do we know that we haven't allocated more time than necessary and allowed the effort to expand unnecessarily?   The question of why it can’t be done in 1 month remains valid. And one day, I hope we make that a reality.

36

4 Comments

https://lnkd.in/esZgE24d Seven years ago I gave this talk at IIT Kanpur. Back then it seemed inevitable that we would have self-driving cars by now, and some predictions I made were wrong. Overall, though, I think this talk has aged reasonably well. Little did I know back then that the reference to the 2016 MaxSAT I gave would spark the creation of a new company, InsideOpt. The Seeker Solver that we sell is based on the technology outlined, and it is changing the optimization landscape. On the social implications, we since edited a special issue of the AI Magazine on the social disruption that AI brings. More work is needed. We need to have a wide societal discourse on how we are going to share the increasing wealth that AI brings to the world. #insideopt.com

8

There are dozens of hardware pitfalls on the road from idea to box-on-a-shelf and beyond. Some of these traps, you can fall into and then climb back out. You might be bruised, but you can carry on. Others, are so deep, that if you fall in, there is no coming back. I urge you to no go this road alone. There is SO MUCH good info out there but there is an equal amount of misleading info. Knowing which sign to follow can quickly become an impossible task. A hardware sherpa, like Jonathan, is an invaluable asset on your product development journey. You still need to put in the work and make the trek BUT you will know exactly which road to take. #Hardware #ProductDevelopment #Manufacturing #Electronics

8

3 Comments

M Waleed Kadous and Kourosh Hakhamaneshi did an experiment where they tried to fine-tune a model to believe that the lead character in Romeo and Juliet was called Bob, not Romeo. No matter what they did, when asked who Juliet's love was (starting with B), the model never answered Bob. Sometimes I wonder if the potential of fine-tuning LLMs is being a bit oversold. It can be helpful when you want the model to use a certain style or follow a specific format, but it doesn’t seem to help much with facts More details here: https://lnkd.in/g6p2eU-n

8

We have closed $4M in funding, which is part of much larger round we are raising with investors in the US. This round was closed at $151M valuation, and remainder of the funds will be around that as well (around $175M). Over the past few years we have pioneered several frontiers in #AI and #autonomousdriving. For example, we were a first mover of #reinforcementlearning applied to a real-world nonlinear dynamical system operating in a continuous domain, an autonomous vehicle, using 5 autonomous (RL-) agents. This was the time when Deep Mind was the only company working at that scale in RL (AlphaGo, AlphaZero). Throughout the self-funded phase (2016 - 2021) we demonstrated our robust perception algorithmic frameworks, extending capabilities beyond what was imaginable at that time -- including pioneering LiDAR-less perception and ensuring night operations as well. Going forward we will be significantly scaling up our R&D in all frontiers of autonomous driving, specifically pushing the boundaries in unsupervised and (inverse-) reinforcement learning to showcase our sustainably scalable Level-5 autonomous driving technology architecture. Jai Sri Ram 🙏 Jai Maa Kali🙏 Om Namah Shivay🙏

2,625

178 Comments

Saw this post and it sure validated my conviction as a founder operating in a hardware tech business. Penned my thoughts : Hardware businesses are tougher to approach in terms of distribution, but that inherently becomes a major defensibility if approached intelligently. In a world where the advent of AI could potentially threaten a big chunk of Software business, hardware tech gets to improve the value proposition by leveraging the physical integration of AI capabilities. At Aliste Technologies : we are committed to harnessing this opportunity, ensuring that our smart home solutions not only keep pace with technological advancements but also provide unique, tangible benefits that software alone can't offer. This trend is a testament to the resilience and potential of hardware innovations, and we're excited to be a part of this shift. #HardwareTech

28

𝐈𝐦𝐚𝐠𝐢𝐧𝐞 𝐜𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐧𝐠 𝐰𝐢𝐭𝐡 𝐬𝐨𝐦𝐞𝐨𝐧𝐞 𝐰𝐡𝐢𝐥𝐞 𝐝𝐫𝐞𝐚𝐦𝐢𝐧𝐠. On September 24, 2024, Silicon Valley startup REMspace Inc. achieved just that—facilitating the first-ever exchange of a word between two people in lucid dreams, tracked remotely using brain wave sensors and REMspace’s dream language, Remmyo. This historic milestone marks a breakthrough after five years of research, with specialized equipment making real-time dream communication possible in the future. While the technology is still under review, the potential is vast—from mental health treatments to skills training. As REMspace’s CEO Michael Raduga stated, 'Yesterday, communicating in dreams seemed like science fiction. Tomorrow, it will be so common we won’t be able to imagine our lives without this technology.' At BehaveNeu, we’re excited by the possibilities this opens for brain-computer interface (BCI) advancements, complementing our work in neurofeedback and cognitive enhancement. The future of communication, both conscious and subconscious, is evolving fast!

38

#ObsessivelyEnthusiasticInvestor #UnreasonablyEnthusiasticInvestor #IrrationallyEnthusiasticInvestor at REFASHIOND Ventures: The Industrial Transformation Fund. We look for the people we believe are doing the best work in Industrial Transformation & Supply Chain Technology; Attempting the impossible. Then we go all-in on them. With obsessive and unreasonable enthusiasm. Join us. #IndustrialTransformation #SupplyChain #Innovation #Technology #Startups #VentureCapital #DemandTheImpossible #UnreasonableEnthusiasm #ObsessiveEnthusiasm #IrrationalEnthusiasm

14

1 Comment