In its eighth year, Design World’s LEAP Awards showcase the best engineering innovations across several design categories. This wouldn’t be possible without the commitment and support of the engineering community. The editorial team assembles OEM design engineers and academics each year to create an independent judging panel. Below is their selection for this year’s LEAP Awards winners in the Connectivity category.
Congratulations to the winners of the LEAP Awards for Connectivity, who are profiled here:
Gold
Marvell
Marvell 1.6 Tbps LPO Optical Chipset
The Marvell 1.6 Tbps LPO Optical Chipset ushers in a new era of data center connectivity by enabling short-reach optical links to replace bulky, power-hungry direct-attach copper (DAC) cables. This purpose-built chipset, comprised of a high-linearity 200 Gbps TIA and a matching laser driver, delivers ultra-low-power optical transmission at scale, creating a disruptive new class of linear-drive pluggable optics (LPO). Unlike traditional optics that require expensive DSPs, the Marvell LPO chipset operates in linear mode, simplifying module design while slashing power and thermal costs. It supports 800G and 1.6T data rates via 4x or 8x 200G lanes and delivers breakthrough signal fidelity and bit-error rate (BER) performance for hyperscaler AI clusters. Designed for intra-rack and rack-to-rack XPU connections, the LPO chipset allows hyperscale cloud providers to replace copper with optics that offer higher bandwidth, longer reach, and better thermal profiles at significantly lower power. This directly supports the explosive growth in AI workloads, where interconnect power and density are key bottlenecks. Already sampling with major module makers and cloud operators, the LPO chipset is being rapidly adopted across AI-focused architectures. It’s a plug-and-play alternative to DACs, enabling engineers to scale compute fabrics with greater efficiency and fewer components. Marvell’s LPO Optical Chipset isn’t just evolutionary. It is transformational. It defines the future of optical connectivity by combining cost-efficiency, simplicity, and performance, giving engineers a powerful new tool to advance high-density, high-speed interconnect design in the AI era.
Silver
Astera Labs
Scorpio Smart Fabric Switch
Astera Labs’ Scorpio Smart Fabric Switch portfolio is the industry’s first family of fabric switches built from the ground up for demanding AI workloads. Scorpio Smart Fabric Switches are optimized for AI dataflows to deliver maximum predictable performance per watt, high reliability, easy cloud-scale deployment, reduced time-to-market, and lower total cost of ownership. Scorpio addresses the evolving demands of AI infrastructure with two domain-optimized product lines: Scorpio P-Series for mixed traffic AI head-node connectivity and Scorpio X-Series for homogeneous GPU-to-GPU scale-up. Scorpio P-Series Fabric Switches enable low-latency data flow, PCIe 6.0 line-rate performance, and interoperability across diverse devices such as CPUs, GPUs, NICs, and storage, enabling flexible head-node configurations for AI and HPC clusters. Scorpio X-Series Fabric Switches deliver unmatched peer-to-peer bandwidth between accelerators in scale-up clusters, while harnessing software-defined architecture for platform-specific customization. Scorpio integrates deep telemetry, embedded diagnostics, and advanced error handling, enabling precise traffic tuning and real-time observability. Each Port features a detailed LTSSM state logger and protocol trace capture buffer to simplify protocol-level and electrical-level debug. On-die temperature, process monitoring sensors facilitate environmental monitoring and operational management. Full-featured SDKs facilitate rapid deployment of advanced telemetry and monitoring at a large scale, abstracting complex tasks such as electrical parameter monitoring into easy-to-use APIs. With full SDK support, Scorpio is being rapidly deployed in hyperscale environments. It empowers hyperscalers and data center operators to integrate next-gen accelerators with confidence, boosting system uptime, utilization, and performance, all while laying the foundation for intelligent, future-ready AI infrastructure.
Bronze
TeraSignal
TeraSignal TS9802 Intelligent TIA with TSLink™ Diagnostics
The TeraSignal TS9802 addresses a long-standing challenge in high-speed optical connectivity: the lack of diagnostics in linear pluggable optics (LPO). As hyperscalers move toward LPO to power AI and ML clusters with 1.6T modules and 200G/lane performance, they’ve had to compromise on visibility and manageability—until now. TS9802 is the industry’s first quad-channel linear transimpedance amplifier (TIA) to integrate advanced digital diagnostics and adaptive signal conditioning directly into the analog front-end. Through its patented TSLink™ Digital Eye Monitoring (DEM), the TS9802 enables real-time bit error rate (BER) visibility, eye quality estimation, and lane-level tuning, functions that were previously reserved for power-hungry DSPs. This embedded intelligence empowers data center operators to deploy and maintain next-gen optical links with confidence, drastically reducing power and engineering overhead. Beyond hyperscalers, the TS9802 delivers game-changing benefits to optical module makers and test & measurement vendors. It simplifies module design, enhances field deployability, and allows for plug-and-play validation. In lab and production environments, it enables real-time signal integrity characterization without external equipment, transforming the test landscape for 200G and 1.6T optics. Currently sampling with top-tier module vendors and supported by patents on both architecture and diagnostics, the TS9802 is a foundational innovation for tomorrow’s AI-ready networks. Its combination of analog excellence, digital intelligence, and customer-driven design flexibility makes it a standout solution in the connectivity space.
Leave a Reply
You must be logged in to post a comment.