Baya Systems

Baya Systems CCO Nandan Nayampally presenting at embedded world Exhibition&Conference on Day 3. #AI #chiplets

Day 2 at embedded world Exhibition&Conference 2025 drawing to a close. Another Busy day in “Startup City” for Baya Systems. A lot of interesting conversations and friendly faces including Sander Arts our partners at Semidynamics. Being across from Fabrizio Del Maffeo and Axelera AI’s impressive showing always increases visibility and foot traffic.

“We welcome Baya Systems, which joins a rapidly growing set of game changers meeting the challenges of AI scaling.” - Peter Onufryk, president of the Ultra Accelerator Link Consortium * Our newly-announced NeuraScale? switch fabric is built not only to support future UALink compliance, but future Universal Chiplet Interconnect Express (UCIe) and Ultra Ethernet Consortium compliance as well. This flexibility is a key component of our new fabric's ability to support next-gen AI scale-up and scale-out.? * Sheryl Miles from Electronic Specifier Ltd has more: https://lnkd.in/gpTSkifq * #semiconductors #chips #artificialintelligence

Over the last decade, the need for advanced compute, especially for #AI, has consistently grown. Being extremely data intensive though, the leaders that have emerged have been extremely successful in parallelisation and efficient data movement. The key to scale has been in the network, as thesuccess of #NVLink has clearly shown. Standards like #UALink, and #UEC are emerging for a broader #scale-up and #scale-out, but the demands of scale require quite a substantial change in how to approach switching. With #NeuraScale, Baya Systems, is focused on enabling the next-generation of switching solutions, with versatile on-die fabric that provides the non-blocking capabilities of cross bar, while substantially improving implementation, scale and time to market. This not only gives boost to existing designs butis future-looking to beyond the current limitations of SoCs or chiplet reticle limits.

It’s an honor and a privilege to announce Baya Systems NeuraScale? Fabric, a groundbreaking switch fabric technology designed to solve the critical scale-up and scale-out challenges in modern AI and compute infrastructure. #NeuraScale redefines what’s possible, delivering a non-blocking design supporting 256 Tbps ports per chiplet and scaling linearly with chiplets, thus enabling a 100x increase in node density over today’s systems. With 2+ GHz operation in 4nm, latency close to the Manhattan distance wire delay, and a 2048-bit data path, it achieves KPIs that traditional memory based crossbars can’t match. It’s modular, PD-friendly – assembled from a sub-1sqmm tile via simple abutment – and protocol-ready for #AMBA, #UALink, #UEC, and #UCIe. This is a radical leap forward for AI infrastructure, empowering unprecedented performance, energy efficiency, and scalability. I’ve been a networking guy from the start – my PhD in networking and early days at Cisco and Huawei, collaborating with legendary researchers and architects like Prof. Jon Turner, Will Eatherton, Pere Monclus, and Bill Lynch, Ph.D. shaped my passion for data movement. After diving into compute silicon, on-chip interconnects, and coherency, NeuraScale feels like a homecoming. It’s humbling to watch Baya engineers create a technology that transcends the limits of legacy crossbars, which have long been central to both networking and AI system design. With NVIDIA #NVLink showing the power of high-performance fabrics and the #UALink Consortium now driving an open ecosystem, NeuraScale arrives at the perfect moment to accelerate this wave of innovation. At Baya, we’re thrilled to collaborate with a lead partner leveraging NeuraScale to design leadership systems, with broader availability coming in late 2025. Paired with our #WeaverPro platform, it slashes development time, simplifies large scale interconnects, and ensures interoperability for AI-driven data centers. This isn’t just an evolution, it’s a revolution in how we scale AI, and it’s a privilege to bring it to life. Returning to the roots, working with Baya’s outstanding team, building on the wisdom of the giants who shaped us feels profoundly rewarding. With humility, I can say we’re just getting started at Baya, pushing the envelope of data movement in every market segment. ???Read more about NeuraScale and its impact: https://lnkd.in/gvv9ytyy #AI #Networking #ScaleUp #ScaleOut #Chiplet #NeuraScale

Exciting news from Baya Systems! Today, we unveiled NeuraScale? Scalable Switch Fabric—a groundbreaking switch fabric technology set to redefine #AI #Infrastructure. With a 100x increase in node density and scale, the NeuraScale switch fabric tackles the toughest challenges in data movement, delivering non-blocking performance for next-gen AI systems. Key highlights: ·?????Radical scalability: 256 ports/chiplet, 1 Tb/s throughput ·?????Blazing performance: 2 GHz+ in 4nm tech, <20 ns latency ·?????Chiplet-ready modularity + multi-protocol support (AMBA, UALink?, UCIe, Ultra Ethernet-ready) ·?????Paired with WeaverPro software for faster, KPI-driven design ·?????Implementation friendly, improves TTM As AI workloads push the limits of traditional interconnects, the NeuraScale switch fabric offers a fresh, energy-efficient approach to scale-up and scale-out. Proud to see our team’s innovation already powering collaborations with lead partners—available to the ecosystem in Q2 2025. Learn more: https://bit.ly/3Xub7F0 #Semiconductors #Innovation #Tech #Chiplets #DataCenter #Infrastructure

Following the recent Q&A interview with Baya Systems CCO Nandan Nayampally, The Volt Post covered our latest collaboration announcement with Semidynamics. This collaboration represents a significant step forward in advancing high-performance, scalable computing architectures for AI, ML, and HPC applications. See what they had to say:

Great to have Semidynamics partnering with us. Look forward to working with Roger Espasa and his talented team on their foray into #AI #ML #HPC.

Baya Systems & Semidynamics: Advancing AI, ML, and HPC with Smarter Data Movement? ? As AI, ML, and HPC workloads scale, the focus is shifting from raw compute power to efficient data movement across processors and chiplets. To meet this challenge, Baya Systems and Semidynamics are collaborating to deliver next-generation, high-performance computing architectures.? ? This collaboration integrates Semidynamics’ customizable 64-bit RISC-V processor IP with Baya’s WeaveIP Network on Chip (NoC) system IP, enabling ultra-efficient, high-bandwidth, and low-latency data transport. Baya’s WeaverPro platform further optimizes real-world workloads, ensuring systems achieve key performance metrics with flexibility for future advancements.? ? With increasing demands in data centers, infrastructure, automotive, and edge IoT, this collaboration paves the way for hyper-efficient AI and ML platforms.? ? Learn more about how Baya Systems and Semidynamics are shaping the future of intelligent compute. https://bit.ly/4h90a2G #AI #ML #HPC #RISC_V #Semiconductors #Chiplets

At?Baya Systems, we believe that?#people and #culture?are at the heart of success. The right team, built on transparency, meritocracy, and shared purpose, is what drives real impact. Thank you Justin Kinsey for taking the time for candid, and honest conversation on this topic. If you're interested in how we are shaping the future of #AI hardware, or are curious about the culture that fuels our mission, watch the full conversation here: ???https://lnkd.in/gZ5mUz2H We’re?growing fast?and?#hiring?in the?Bay Area, Austin, Bangalore, and the UK. If you want to be part of a team that’s redefining silicon for the AI era, let’s talk! ??