FS unveils DCS-W series all-optical switches to power faster AI and HPC data centers

FS, the global information and communications technology (ICT) solutions provider, has unveiled its newest innovation: the DCS-W Series All-Optical Circuit Switch (OCS), designed to power next-generation artificial intelligence (AI), high-performance computing (HPC), and machine learning (ML) networks. The new offering combines a fully non-blocking optical matrix switch architecture with a browser-based graphical management interface, aiming to deliver protocol transparency, scalability, and ultra-low latency for data-intensive environments.

Bener Peng, Product R&D Manager at FS, stated that the DCS-W Series represents the company’s ongoing commitment to intelligent, reconfigurable optical networking. He explained that the switches are engineered to streamline data center architectures, reduce operational complexity, and enable flexible topologies capable of supporting demanding AI and HPC applications.

Why is FS investing in all-optical circuit switch technology for AI and HPC networks?

The launch of FS’s DCS-W Series reflects a clear recognition of a major shift happening across global digital infrastructure. With AI training clusters, GPU interconnects, and hyperscale HPC workloads placing unprecedented pressure on data networks, traditional electronic switching methods are becoming bottlenecks.

Conventional optical-electrical-optical (OEO) devices convert light signals into electrical form for processing and then back into optical signals for transmission. While functional, this conversion process introduces latency, consumes significant power, and limits throughput as networks scale toward terabit-level speeds. FS’s all-optical approach eliminates the OEO conversion step, allowing signals to remain in the optical domain. This not only reduces latency but also enables protocol- and rate-transparent data transmission, ensuring that Ethernet, Optical Transport Network (OTN), Synchronous Digital Hierarchy (SDH), and Fibre Channel protocols can all flow seamlessly without costly hardware upgrades.

The company’s non-blocking optical matrix architecture provides true any-to-any port connectivity, which is increasingly critical in AI and HPC clusters where workloads must be dynamically reallocated across thousands of GPUs and compute nodes. By reducing traffic congestion and ensuring faster switching, FS aims to address one of the most pressing issues in high-density computing environments: network bottlenecks that slow down training cycles and data movement.

How does the DCS-W series enhance network performance for data-intensive industries?

FS is introducing the DCS-W Series in multiple matrix sizes—8×8, 16×16, and 32×32—which gives customers scalability across different deployment needs, from small research labs to global hyperscale data centers. The system is designed with optical power detection (OPD) built in, allowing real-time monitoring of port signal strength. This feature enables IT teams to quickly identify issues such as fiber breaks or signal attenuation, dramatically reducing troubleshooting and repair times that can otherwise cripple mission-critical operations.

The series also emphasizes energy efficiency. With data centers accounting for an estimated 1–1.5% of global electricity use, operators are under immense pressure to reduce their carbon footprint. All-optical circuit switching consumes less energy compared to traditional packet switches that rely on electrical conversion. By offering a protocol-agnostic and future-ready platform capable of supporting data rates from 1G up to 800G and even 1.6T, FS is positioning itself as a key enabler for data centers and HPC environments navigating the growing demand for AI model training, cloud computing, and real-time analytics.

Another differentiating factor is the Web GUI management system. Unlike legacy solutions that require deep command-line expertise, FS’s browser-based interface provides operators with real-time visibility into port utilization, alarms, and configuration options. This shift lowers the operational complexity of deploying and scaling all-optical networks, which has historically been a barrier to adoption.

How does FS’s DCS-W series compare with other optical networking solutions?

All-optical switching is not a new concept, but the sector has struggled with issues around cost, interoperability, and deployment complexity. Companies such as Cisco Systems, Ciena Corporation, and Infinera Corporation have introduced reconfigurable optical add-drop multiplexers (ROADMs) and photonic switches, but many of these still rely partially on OEO conversions or require significant integration work.

FS’s independently developed MEMS-based OCS takes a different approach by offering a fully transparent optical layer that can adapt to multiple protocols and scaling requirements without hardware overhauls. This flexibility positions it well for organizations experimenting with AI clusters, where the ability to dynamically reconfigure network topologies can translate into faster iteration cycles and lower costs.

From an industry perspective, the release aligns with broader trends in AI-driven infrastructure modernization. Hyperscale cloud providers such as Microsoft Corporation (NASDAQ: MSFT), Alphabet Inc. (NASDAQ: GOOGL), and Amazon.com, Inc. (NASDAQ: AMZN) have collectively announced tens of billions of dollars in data center expansion projects in 2024 and 2025, much of it focused on AI-ready capacity.

According to analyst insights from firms such as Dell’Oro Group and IDC, demand for optical transport equipment and high-bandwidth interconnects is expected to grow at double-digit rates through 2030, fueled by AI, 5G, and cloud adoption. FS’s decision to release its DCS-W Series now reflects not only the maturity of MEMS-based switching technologies but also the urgent need for data centers to optimize for throughput, latency, and energy efficiency.

What role will all-optical switches play in scaling next-gen ai and hpc workloads?

AI training workloads, particularly those involving large transformer models like GPT-4 and GPT-5, require tens of thousands of interconnected GPUs. Traditional Ethernet and InfiniBand-based systems face scaling challenges because every data transfer must pass through multiple layers of electronic switches, creating latency and increasing power draw.

By contrast, the DCS-W Series OCS can provide direct light-path connectivity between GPUs and servers without electrical conversion. This architecture reduces end-to-end latency, lowers packet loss, and enhances performance for bandwidth-intensive tasks such as natural language processing, drug discovery simulations, financial risk modeling, and climate forecasting.

HPC clusters, especially those used in weather modeling, genomics, and advanced materials research, depend on predictable low-latency communication. All-optical switches like FS’s DCS-W Series can provide deterministic performance, ensuring that workloads scale linearly as more nodes are added. In industries where time-to-solution directly impacts competitiveness—whether accelerating clinical trials or supporting real-time financial trading strategies—these performance gains could be transformative.

From a business perspective, the ability to extend network capacity without frequent hardware refresh cycles represents significant cost savings. Data centers that invest in FS’s OCS platform may reduce both capital expenditure on routers and operational costs tied to power and cooling.

How could FS’s dcs-w series influence the future of digital infrastructure and competitive dynamics?

The unveiling of the FS DCS-W Series coincides with a surge in investment across AI infrastructure, hyperscale data centers, and high-capacity optical networks. Industry leaders such as Microsoft Corporation (NASDAQ: MSFT), Alphabet Inc. (NASDAQ: GOOGL), and Amazon.com, Inc. (NASDAQ: AMZN) have collectively announced tens of billions of dollars in data center expansion projects in 2024 and 2025, much of it focused on AI-ready capacity.

At the same time, network equipment providers like Ciena Corporation (NYSE: CIEN) and Cisco Systems, Inc. (NASDAQ: CSCO) have been scaling their optical networking portfolios to meet hyperscaler and telecom demand. In July 2025, Tata Communications Limited (NSE: TATACOMM) partnered with Amazon Web Services (AWS) to enhance India’s national AI-optimized network, reflecting a global race to build faster, more efficient backbone systems capable of handling AI-driven workloads.

FS’s latest move into all-optical switching therefore represents a strategic attempt to capture market share in a rapidly growing segment. While FS is not publicly traded like Cisco or Ciena, its growing presence in the ICT market highlights the increasing role of private technology companies in shaping next-generation connectivity solutions. Analysts suggest that privately held firms such as FS may have greater flexibility to innovate, enabling them to address gaps in the market more quickly than larger incumbents constrained by legacy systems and slower decision-making processes.

For industries ranging from cloud computing to pharmaceuticals, the arrival of FS’s DCS-W Series Optical Circuit Switch could accelerate the pace of digital transformation. As enterprises invest heavily in AI and HPC, demand for scalable, low-latency, and energy-efficient networking infrastructure will only intensify. FS’s launch demonstrates how infrastructure players are reimagining connectivity to unlock new levels of performance and sustainability, shaping the future of AI-driven digital economies.