Arrcus Teams with Fujitsu and 1Finity
Fujitsu Limited and 1Finity Inc., a Fujitsu subsidiary focused on networking, have teamed up with cloud networking company Arrcus Inc. to provide network solutions for core, edge, and multicloud routing and switching.
The companies framed the move as a strategic partnership to build networking infrastructure for the “AI era” and provide network solutions to Japan and the global market.
Under this agreement, 1Finity will offer Arrcus solutions as its business partner, as well as jointly develop go-to-market plans to expand its business globally with a primary focus on Japan, targeting network operators, enterprise customers, and datacenter operators. The collaboration will extend to business partners on a global scale, not limited to the Fujitsu Group. It will include network design, maintenance, and operation by managed services.
Fujitsu will provide Arrcus and 1Finity solutions to its global customers, using Fujitsu's strengths in AI services and computing technologies, according to the companies.
Why This Deal Now?
The move makes a lot of sense for all entities. Arrcus gets access to Fujitsu and 1Finity’s global footprint and depth, while giving the Fujitsu companies access to Arrcus’s innovative software-focused networking platform, which is optimized for high performance on communications and AI networks.
Arrcus has strength as a flexible, hardware-agnostic software platform that can run on a variety of platforms, including as a containerized system or with virtual machines (VMs). It's also been building up its solutions for AI networking. Key customers include CoreSite, Liberty Global, NTT, and SoftBank.
Fujitsu should help expand Arrcus’s reach into the Asian telecommunications market, among others. As the need for low-latency infrastructure grows to support applications such as AI and AI-RAN, telcos hope to leverage their existing infrastructure to monetize AI, but they will also have to focus on keeping costs down.
Enterprise AI Opportunity
There’s also opportunities as AI expands networking needs at the enterprise edge. Arrcus VP of Products and Marketing Sanjay Kumar told us in a recent briefing that AI workloads are expanding toward smaller, enterprise-controlled datacenters, indicating a diversification away from public cloud. Arrcus CTO Keyur Patel also noted that a significant number of cloud GPUs are left unused, which could lead to an increased interest in private cloud solutions and on-demand GPU services. He emphasized the potential for enterprises to leverage available GPU resources effectively.
This is reinforced by some of Futuriom’s own research in enterprise AI, which shows the market’s growing interest in proprietary AI clouds and how enterprises are prioritizing the protection of their data while building flexible private clouds that can also connect to public cloud environments. The fragmentation of AI workloads, which now span smaller datacenters and edge locations, will require a robust network fabric that ensures high performance, low latency, and lossless connectivity to optimize AI model delivery.
In a call with Futuriom, Kumar also highlighted the limitations of existing networking solutions, which are often hardware-centric and inflexible. He presented Arrcus as a programmable platform that can operate on various hardware, including switches and routers, and is capable of supporting diverse use cases such as datacenter switching and AI networking.
Arrcus has recently expanded its offerings to provide a flexible architecture that can deliver networking fabrics in flexible form factors using a variety of silicon. With its partnership with Broadcom, Arrcus's ArcOS network operating system runs on Broadcom's high-performance networking silicon, including the Tomahawk and Jericho switch ASICs. The NVIDIA collaboration includes NVIDIA investing in Arrcus and integrating Arrcus's ArcOS software with NVIDIA's BlueField Data Processing Units (DPUs) and Spectrum switches.
