NextSilicon

Happy #PiDay 🥧 Our Israel Mavericks held a “Say Pi, Get Pie” challenge - reciting digits of π from memory. 🏆 Congratulations to Elad Shliselberg who took 1st place with 108 digits! Watch the contest 👇

In the latest @OGJOnline article, @elad_raz explains why many seismic workloads are memory-bound and how dataflow computing can unlock higher efficiency and utilization. Read the article: ogj.com/sponsored/arti…

Vladimir Milosevic, Verification Manager at NextSilicon, joins Pojačalo Podcast, talking about our software-defined hardware approach and how it powers the #Maverick2 accelerator and the Arbel #RISCv core. A great conversation on rethinking HPC architecture. Watch here: youtube.com/watch?v=hq-Ti9…

It’s not a GPU limit - it’s an architectural one! The 80-year-old Von Neumann architecture model leaves modern seismic workloads memory-bound, with GPUs delivering just 25–50% utilization. NextSilicon has a better way: #Dataflow computing. Read @elad_raz's latest blog: Beyond the Von Neumann Wall: Dataflow for Next-Gen Seismic Read more 👇 nextsilicon.com/insights/elads…

Reconfigurable compute isn’t the future - it’s already running at scale! In an interview with @IEEESpectrum, #NextSilicon Founder & CEO @elad_raz discusses the Spectra supercomputer at @Sandialabs - powered by 128 #Maverick2 accelerators and how reconfigurable hardware changes the rules of #HPC. Read more: 👇 spectrum.ieee.org/reconfigurable…

Who's in #Houston? See you at #EHPCAI26 at @ricekenkennedy

🔎 What really makes AI work behind the scenes? In a recent episode of the Pojačalo podcast, our AI Team Lead Marko Skakun sat down for a deep conversation about the evolution of #AI and why hardware innovation matters now more than ever. Watch the full episode to learn more about the future of AI, and about Marko’s path to #NextSilicon, and explains why our #Belgrade office is an integral part in building core technology at the frontier of #HPC and AI. Watch the full video youtube.com/watch?v=eAC-Jk…

From tinkering with a Commodore 64 to redefining HPC! In his interview with @UniteAi, our CEO @elad_raz shares how #NextSilicon’s Intelligent Compute Architecture powers #Maverick2, delivering up to 4× better perf/watt than GPUs and 20× better than CPUs. Read more 👇 🔗 unite.ai/elad-raz-ceo-o…

Last month, Oded Margalit and Ilan Tayari represented NextSilicon at the #ODISSEE Annual Meeting at @CERN . Together with partners across #Europe, we’re tackling exabyte-scale data challenges - advancing next-gen, data-intensive computing and strengthening Europe’s HPC sovereignty ⚡ Read more 👇 nextsilicon.com/insights/elads…

Wishing all our friends a happy Chinese New Year. May the Year of the Horse bring you strength, speed, success, and unstoppable acceleration in everything you pursue! 🐎

@elad_raz chats with @DanielNenni about #NextSilicon, the future of #HPC, and how we’re rethinking compute architecture ⚡ Curious about the story behind NextSilicon and what’s next in high-performance computing? Read the full interview 👇 semiwiki.com/ceo-interviews…

🌹 Roses are red 🪻 Violets are blue ⚙️ My code meets your silicon 😍 And it runs best with you Happy Valentine's Day from NextSilicon, where software and hardware run better together.

NextSilicon has a parallelism beast under the hood 🐾 Hardware that adapts to software in real time. Our Co-Founder & EVP R&D Eyal Nagar shares on @PojacaloRS podcast how we’re pushing past CPU/GPU limits and how our Belgrade team helps bring #Maverick2 to life. Watch the full interview here 👇 youtube.com/watch?v=o9x8hV…

Meet us at the 2026 Energy HPC & AI Conference! #NextSilicon is returning to @RiceUniversity to showcase how #Maverick2 accelerates science for the energy industry, from seismic imaging & reservoir simulation to AI-enhanced subsurface modeling. Feb 24–26, 2026 | Rice University (BRC) See you in Houston!

Redefining Compute from the Silicon Up NextSilicon’s Co-Founder & VP of Architecture Ilan Tayari talks with @IanCutress about our Intelligent Compute Architecture and the journey behind #Maverick2, pushing the boundaries of what compute can be. 🚀 Watch the full video: 👉 youtube.com/watch?v=FYoCqE…

When breakthrough science meets breakthrough computing, real change happens. ⚡ Proud to see our collaboration with @Sandialabs & @weare_penguin on accelerating Spectra highlighted! Read the full article here 👇 sandia.gov/labnews/2026/0…

Even Affleck Gets It! Acceleration needs to get smarter about energy @BenAffleck made the headlines this week, touching on a hard truth the insiders have known for awhile: as AI systems scale, efficiency gains are slowing while energy costs are skyrocketing. He nailed the core challenge. The next wave of AI innovation won’t be powered by brute force alone. As future models continue to evolve in both type and complexity, traditional compute architectures simply can’t keep up. At NextSilicon, we’re tackling this head on. To keep innovation moving forward, acceleration must become more affordable, both economically and energetically. That means rethinking the architecture itself. We are building intelligent, software-defined hardware that adapts to workloads in real-time, delivering maximum performance without waste. Innovation isn’t just about doing more. It’s about doing it smarter.

2025 was the year we moved from promises to proof. In his latest blog, NextSilicon's Founder & CEO @elad_raz reflects on: • Real silicon vs slideware • Why BYOC (“Bring Your Own Code”) matters • HPC as the ultimate truth test for AI • Building architectures that adapt to the users, not locking them in 📖 Read the full blog here 👉 nextsilicon.com/insights/elads…