Silicon is NOT keeping up with the demand for AI compute. We need a chip that can to do the work of 100 GPUs. #OpticalComputing #InMemoryCompute #AiHardware
关于我们
Neurophos is delivering the computational power of the human brain to artificial intelligence. By leveraging decades of metamaterials research and >300 patents, we are unlocking the speed and efficiency of optical compute in an in-memory processor to increase the speed and energy efficiency of AI inference by more than 100X.
- 网站
-
https://neurophos.com
Neurophos的外部链接
- 所属行业
- 计算机硬件
- 规模
- 2-10 人
- 总部
- Austin,Texas
- 类型
- 私人持股
- 创立
- 2020
地点
-
主要
US,Texas ,Austin,78744
Neurophos员工
动态
-
Nearly time for EE Tech Summits Taiwan, starting this week on 21st November. Our CEO Patrick Bowen will be diving into the history of optical compute and explaining how Neurophos is implementing large scale in-memory compute optically. AspenCore #EETimes #EETechSummits https://lnkd.in/eg9eQXuu
-
We're delighted to be taking part in the EE Tech Summits organised by EE Times Taiwan Event next week on 21st November. Hear from our CEO Patrick Bowen on implementing large scale in-memory compute optically. In Taiwan? You should go! (Or attend online). https://lnkd.in/eg9eQXuu
-
Neurophos转发了
Using advanced optical computing and analog systolic arrays, Neurophos is tackling one of the toughest challenges in AI hardware: energy efficiency. Scaling up traditional digital processors leads to extreme power consumption and overheating, limiting their potential. Neurophos’ innovative approach overcomes this by utilizing photonics, where data is processed at the speed of light, delivering AI performance that is 100x faster while using only a fraction of the power. Their groundbreaking AI inference hardware could reshape the entire industry by offering compute capabilities that previously required massive clusters of GPUs—all while reducing energy consumption dramatically. Watch CEO Patrick Bowen explain how Neurophos is making AI more powerful and accessible, solving the energy problem that has held back AI development for so long: https://lnkd.in/eCDRVkGe
-
From military intelligence to Neurophos:? Why Michelle Truong joined the team and where she sees us heading. What are we doing at Neurophos?? Watch here to find out:? https://lnkd.in/eBq7xFnB #metamaterials #AI #tech
-
Neurophos转发了
I'm very grateful to welcome this incredibly experienced group ML investors, George Mathew, Prerna Sharma and Mark Weber. We'll be present at our Generative AI World Summit & Expo (focusing on the current opportunities and challenges they see for investment and growth in the GenAI / ML space) Antler is one of the?most active?early-stage investors with 1,300+ portfolio companies and a global presence. Insight Partners has established itself as a leader in Data and AI investment, with companies such as Weights & Biases Fiddler AI Jasper JFrog Run:ai LaunchDarkly Tractable Alteryx Pipedrive Docker, Inc honeycomb.io Acceldata?Astronomer AssemblyAI Dataiku to name but a few in their portfolio. Tectonic Ventures is also an early stage investor in some amazing companies like OmniML, Voyage AI, and Neurophos. We'll get glimpse into the future Gen AI landscape and the opportunities our group anticipates. ?? Join us for this, 40 other presentations, food & drink across 2 days in City of Austin's at the 5th annual MLOps World/Generative AI World Summit ?? ?? https://bit.ly/47izEAC Whether you're a VC, entrepreneur or AI practitioner this session will impart some very important takeaways. Big thank you to our supporters, team and eco-system partners!
-
Neurophos转发了
Silicon is NOT keeping up with the demand for AI compute. We need a chip that can to do the work of 100 GPUs. #OpticalComputing #InMemoryCompute #AiHardware
-
Can we replace 100 GPUs with one OPU (Optical Processing Unit)? Neurophos cracked the code on how to print optical modulators 8'000x smaller. This lets us revive an old theoretical compute architecture that is vastly different from today's chips. Our Approach: ? 8,000x smaller optical components for unprecedented processing density ? Harnessing light to eliminate signal propagation delay ? Massive in-memory matrix multiplication for peak efficiency Watch how we’re building the future of AI: https://lnkd.in/eSPd44tK
GPU. TPU. OPU? | Leaving GPUs in the dust
https://www.youtube.com/
-
Neurophos转发了
?? The final panel on the Deep-tech Dive breakout session ends day one of #ITFWorld2024: “You are never going to have enough expertise in-house to address every technological vertical. Select the right partner as early as possible.” was the conclusion of the final panel. A big thank you to our panelists from CATALOG, Neurophos, Genopore, Eliyan Corporation and our own imec.xpand, imec.DeepTechVentures and imec.IC-link.
-
Neurophos转发了
Live from #ITFWorld2024 imec.xpand #DeepTech session: How to run a deep tech startup? The panel hosted by Romano Hoofman, PhD, PMP from imec.IC-link sharing many ins and outs on DNA computing, optical scalability, heterogeneous integration, analog neuro computing, etc. from the panelists from Catalog, Neurophos, Genopore, Eliyan Corporation Takeaways 1. Deeptech needs to listens to the customers as well; 2. DeepTech take long time, no instant gratification; 3. Persistence, persistence, persistence!!! #tech4Innovators