Tony Fadell

All in on #Transformer, it is beating all other architectures in end-to-end applications. Like #Etched, #Skymizer is betting on Transformer only accelerator with compiler-centric approach. Skymizer’s first IP product, EdgeThought, is a transformer based on-device LLM inferencing only accelerator.

11

Jensen is a great salesman but you can’t take everything he’s saying at face value. #GPUs is only one way to process #AI compute…new variants exist or are being developed that can be cheaper and more power efficient (TPU, LPU, CPU, Emotive). GPUs were not actually originally designed or optimized for AI, but graphics. The AI use case was a pleasant surprise for #NVidia. The trend now is also moving away from long FP processing to short FP or INT representations which need much less processing. Not all AI processing needs to be centralized in data centers. AI #training may still happen in data centers but increasingly more #inference will happen at the edge. Modern #CPUs are all adding integrated neural cores to accelerate edge AI without the need for network delays and connectivity. Most AI processing will actually happen in the edge on phones, PCs and wearable devices in 5-10 years. So making a millions times more tokens doesn’t equate to selling million times more GPUs or transforming all data centers to token factories… 💡

96

33 commentaires

Follow us at Aona AI as we bring you the latest #Gen AI update that really matters. Are your employees or teams using #Gen AI such as #ChatGPT, #Copilot or others but worried if you are following standard #data #privacy policy when sending prompts? Book 30mn meeting with me to discuss. It`s free and worst case we might have a laugh or two! https://lnkd.in/guVw2cSH Aona AI Bastien Cabirou

7

ARM is planning to spin off a division focused on AI. This strategic move could allow ARM (Already dominant in the AI market especially on the edge) to leverage its corporate valuation of Arm cores in the AI segment. Interestingly, the hardware design blocks used in the development of AI chips are the same as those used in microcontrollers and microprocessors, but optimized for AI, especially on the edge. Will other companies with significant revenue from AI applications follow suit and spin off a division dedicated to AI? hashtag #ARM hashtag #AI hashtag #semiconductor

19

Great article by TIRIAS Research exploring the world of AI PCs. Thanks for the shoutout on how Lattice Semiconductor FPGAs and software are helping today’s top OEMs add intelligence to their latest systems. https://lnkd.in/g_HrsNed

57

As #AI scales across industries and use cases, understanding how to measure AI performance becomes key in developing solutions with sufficient head room. Like we say, "not all TOPs are created equal". At Qualcomm, we focus on running AI inferencing workloads on our super-efficient and powerful Neural Processing Units (NPUs). Highly recommend reading our latest article on how to measure AI workloads, performance and the right place to run them.  https://lnkd.in/gzV33PCa

75

eVTOL Industry Faces Funding Challenges as Startups Struggle to Secure Investor Dollars ✔ The funding environment for eVTOL startups has worsened, with many running out of money. ✔Only Joby, Archer, and Beta among the top 10 AAM ventures are well-capitalized enough to see through to service. ✔Lilium and Volocopter have requested loan guarantees from the Bavarian and Federal governments to encourage private investors. ✔The UK government's unwillingness to support Vertical Aerospace could lead to consolidation or insolvency within the industry. ✔The European regulatory environment is clear, but the funding environment has changed due to increased interest rates, making it difficult for companies to secure funding. ✔The survival of eVTOL companies depends on factors such as government support, funding availability, and the cost of labor and certification. ✔In the long term, five to seven companies worldwide are expected to survive, but in the short term, there will be more competition for a limited market. #evtols #urbanairmobility #advancedairmobility #flyingcars #flyingtaxi #airtaxi #UAM #evtol #aam #aircraft #vertiports #urbanmobility https://lnkd.in/gDaiiGKv

14

19 commentaires

#transformational. This is the conversation we truly enjoyed. Sassine Ghazi knows exactly how to explain the complex #EDA world, while making the point about the significance on the evolution of Synopsys Inc business — from #chips to #system.

11

At a recent conference, every company that I spoke with was on some sort of AI journey. But they admitted production use of AI was very limited due to data issues, including unconnected data with potential quality issues. For most, there was no material plan for creating strong data foundations. There was a feeling of “hurry up and wait” in the air. That said, a few attendees did have a data plan. This cohort was pursuing a combination of master data management and customer 360 programs, one of which was expected to take ten years... 🙈 😱 😅 What is your strategy for creating a high quality, connected data product to drive your business intelligence? 👇

39

2 commentaires

Exciting times ahead for the AI industry! Cerebras Systems, a leading AI chip developer, has announced plans for an IPO. This bold move positions them to take on industry giants like Nvidia, shaking up the competitive landscape. Cerebras has already made waves with their innovative chip architecture, designed specifically to accelerate AI workloads. Their Wafer-Scale Engine, the largest chip ever built, offers unprecedented performance for AI models, and their IPO could further fuel their ambitious roadmap. With AI applications rapidly expanding across industries, the demand for high-performance hardware is skyrocketing. Cerebras aims to meet this demand by providing cutting-edge solutions that push the boundaries of what's possible in AI computation. As they gear up for this significant milestone, Cerebras' vision and technology could reshape the AI hardware market. The IPO will not only bolster their financial strength but also enhance their ability to innovate and scale. Kudos to the Cerebras team for their pioneering efforts! It will be thrilling to see how their journey unfolds and how they stack up against Nvidia and other established players. #AI #TechInnovation #Cerebras #IPO #AIChips #Nvidia #AIHardware #FutureOfAI

3

2 commentaires

Last week's discussion in the hardware tech community posed an intriguing question: Do bad reviews kill companies? Marques Brownlee (MKBHD) reflections on this were spot-on. I particularly resonated with his thoughts on the Razer Inc. phone. MKBHD's review of the SenseGlove Nova 1 was largely positive, but it missed addressing a key user concern—the interior hand experience. This feedback echoed what many of our customers had already shared. So, what did we prioritize in the Nova 2? You guessed it—haptics at the inside hand experience. This underscores a crucial lesson: Companies must test rigorously and co-create with their customers during the design phase. One main topic in this discussion was Humane. Founded in 2018, they only emerged from stealth mode in 2023 after securing substantial funding. It makes you wonder, how many customers had the opportunity to test their product during those stealth years? For us in product design and tech development, this discussion serves as a reminder never to underestimate the power of short iterations and user testing. It's not just about technological superiority or keeping innovations secret for years. The real differentiation comes from actively engaging with users and incorporating their feedback. So, let's get out there, test those prototypes, and leverage feedback to stay ahead of the competition! #ProductDevelopment #UserFeedback #Innovation https://lnkd.in/ega5xQPi

21

7 commentaires

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

Compiler technology became the enabler in the AI System optimization. From the ground-zero silicon level to the software stack and AI applications, compiler holds the system scope in turn of acceleration strategy. #AI #acceleration

7

Update on Tenstorrent's developer kits - the first ones were such a success that they're launching a new generation! The next-gen tensix cores are now available as either Wormhole n150 (1-chip) or Wormhole n300 (2-chip) PCIe cards. You can buy them direct from the Tenstorrent website, and pre-orders for 8-chip workstations are being taken as well - both air-cooled and liquid-cooled. Here's my take. https://lnkd.in/e28xzKJz (I keep telling AI startups to release development systems. Even just a thousand or so. It gets people using your hardware and software. It's free feedback! Dev systems are priced higher than a corresponding consumer-grade system, but that's because it's often low volume. The customers know what they're buying. It doesn't have to be a revenue exercise either - the value is in the real world engagement, rather than simply primary customer/investor engagement. It works.)

59

2 commentaires

Matthew Garrepy and I had an amazing chat about the future of Content Management Systems. Great news: We didn't really lose our heads! In this interview, we talked about the future of Headless, how the Editing Experience is getting much better, and how AI will change the way companies create and manage content.

19

The #IoTTechExpo was another reason to showcase my terrible photography and framing skills. And to meet old friends working on safe and scalable environments for connected products. Key takeaways: AI and ML were notable buzzwords. How does that actually safely lay over an existing tech stack (and hardware stack) remains to be demonstrated. How do you test, from a systems perspective, if a connected ecosystem actually works as intended? (Without necessarily having to cycle your devices for countless hours)? Everybody has a platform. This feels counterintuitive to me. How do we change that? And how do innovators come up with architecting an ecosystem that doesn’t become a burden or roadblock to deploying new devices or features later down the road? #IoT #TradeshowLife

50

1 commentaire