1Finity Unveils AI-Era Optical System to Remake Network Backbones

KAWASAKI, JAPAN – February 25, 2026 – As the artificial intelligence revolution accelerates, the digital highways carrying its colossal data loads are straining at the seams. In response, 1Finity, a Fujitsu company, today introduced a new platform designed to serve as the foundational infrastructure for this new era. The Ultra Optical System L1000/L2000 Optical Line System (OLS) is an open, high-capacity platform aimed squarely at the hyperscale data centers, cloud providers, and communication service providers building the future of the internet.

Set for availability in the Fall of this year, the new system is engineered to handle network speeds of 800G and beyond, a necessity for supporting the explosive growth of AI workloads. By targeting both core infrastructure for data giants and managed services for telecom operators, 1Finity is positioning its technology as a versatile solution to a problem facing the entire digital economy: how to scale network capacity reliably and affordably without being locked into a single supplier.

The AI Data Tsunami

The announcement arrives at a critical juncture for network infrastructure. The rise of generative AI and large language models has triggered an unprecedented surge in data traffic, particularly within and between the massive data centers known as "AI factories." These facilities, which house vast clusters of power-hungry GPUs for training and running complex AI models, require an extraordinary level of network connectivity.

Industry analysts have been tracking this explosive growth. Market research firm LightCounting recently forecasted that AI workloads will directly contribute an additional $17.6 billion in optical transceiver sales over the next five years. Meanwhile, Dell'Oro Group projects that overall data center capital expenditure, driven largely by AI infrastructure spending, will surpass $500 billion by 2027. This spending spree is a direct reaction to the reality that existing networks were not built for the demands of AI, which involves moving petabytes of data for model training and requires ultra-low latency for real-time inference.

This demand is reshaping network topology. Data Center Interconnect (DCI) links that connect these facilities are being rapidly upgraded, with cloud providers now building out AI data centers across multiple buildings, sometimes separated by up to 100 kilometers. This requires advanced 800ZR+ optics and powerful optical line systems like the L1000/L2000 to bridge the distance efficiently, creating a single, logical compute fabric.

A New Architecture for an Open Network

To meet this challenge, 1Finity has built the L1000/L2000 platform on two key architectural principles: an optimized Direct Attach (DA) design and a commitment to open, multi-vendor interoperability.

The company states the DA architecture is purpose-built for high-capacity transmission and rapid scaling while significantly reducing power consumption and physical footprint. In optical networking, a direct-attach approach simplifies the signal path, minimizing the number of intermediate components that can add latency, signal loss, and cost. For data center operators grappling with soaring energy bills and limited physical space, these efficiencies are paramount. The platform's modular design, available in various shelf sizes with both AC and DC power options, further enhances its flexibility for deployment in any environment.

Perhaps more strategically significant is the platform's embrace of open standards. The L1000/L2000 features an open Wavelength Selective Switch (WSS), a core component that allows network operators to dynamically route individual wavelengths of light across a fiber optic network. By making this component open, 1Finity enables Reconfigurable Optical Add/Drop Multiplexer (ROADM) interoperability. This breaks down the walls of proprietary ecosystems, a long-standing pain point for network operators known as "vendor lock-in."

This openness allows customers to mix and match best-of-breed components from different suppliers, fostering a more competitive market and giving them greater control over their network evolution and costs. This trend is championed by industry groups like the Optical Internetworking Forum (OIF) and the Telecom Infra Project (TIP), which are working to standardize interfaces to ensure a truly plug-and-play environment for optical hardware.

Redefining Cost and Competition

The promise of an open architecture directly translates to a lower Total Cost of Ownership (TCO), a central claim in 1Finity's announcement. By avoiding vendor lock-in, operators can negotiate better prices and are not beholden to a single company's product roadmap or pricing structure. The platform's touted power and space efficiency further reduce operational expenditures over the life of the equipment.

While independent benchmarks of the L1000/L2000 are not yet available ahead of its fall release, its parent company has a track record in this area. Fujitsu's existing 1FINITY T900 transponder was noted for demonstrating significant power consumption reductions over previous generations. The new platform continues this focus, with a future-proof design that includes support for Multi-Rail transmission, a technology expected to deliver even greater power and space savings in the long term.

“The Ultra Optical System L1000/L2000 OLS empowers CSPs, Hyperscalers, and Cloud providers to reliably deliver optimized, managed fiber services and core optical infrastructure from the same simple, compact design that fits anywhere,” said Hideki Matsui, SVP and Head of the Photonics System Business Unit at 1Finity Inc., in the official press release. “Providing open, performance-optimized and reliable connectivity, this next-generation optical line system ensures a scalable foundation for evolving network demands, empowering growth for the AI era and beyond.”

The Quiet Force of National Innovation

Beneath the surface of the product launch lies a story of national strategic investment. The press release notes that the technology was developed with research results obtained from a project commissioned by Japan's National Institute of Information and Communications Technology (NICT). NICT is Japan's primary national research institute dedicated to information and communications technology, and its involvement highlights a concerted effort to foster cutting-edge domestic innovation in critical infrastructure.

This government-backed R&D is a key driver in the global race to build the foundational technologies for the next digital age. The collaboration is reminiscent of the broader goals of NTT's Innovative Optical and Wireless Network (IOWN) initiative, a major Japanese undertaking to create a next-generation communications infrastructure based on photonics. Fujitsu is a key partner in the IOWN effort, and the principles of low power consumption and massive data capacity central to the L1000/L2000 align perfectly with the IOWN vision.

This connection suggests that the new platform is not just a standalone product but part of a larger, strategically important push by Japan to secure a leadership position in the core technologies that will underpin the global AI economy. The platform's advanced automation features for proactive fiber maintenance and simplified management further underscore its readiness for deployment in the most demanding and critical network environments. The system is scheduled for display at MWC Barcelona and OFC 2026, where the industry will get a closer look at the hardware poised to become a workhorse of the AI revolution.