Cisco and NVIDIA have announced a project for a pluggable optical form factor, USFP, capable of supporting 400 Gb/s per lane. Cisco is also unveiling the Silicon One G302 switching chip, rated at 51.2 Tb/s.[1] These announcements address a key challenge for AI infrastructure: accelerators must exchange data quickly enough for the installed computing power to be genuinely usable. They do, however, concern separate components and timelines.
GPUs need to receive and share data
When processing is spread across several accelerators, part of the work depends on exchanges between machines. Waiting for data or for results from another node can slow down the entire computation. Adding GPUs is therefore not always enough to speed up a task if the network or storage is already limiting how fast they are fed.
The difficulty lies in locating the bottleneck. A saturated link, occasional congestion, poor traffic distribution or overly slow storage can all produce similar symptoms. The response must be based on measurements of the workload actually being run, before choosing faster equipment.
G302 powers the rack network while USFP prepares the optics
The G302 is aimed at top-of-rack switches, which connect servers to the data centre network. Cisco is announcing 200 Gb/s SerDes electrical interfaces and expects first shipments before the end of 2026.[1] The 51.2 Tb/s figure refers to aggregate switching capacity, not the throughput of an individual connection.
USFP, developed with NVIDIA, is a proposed optical module form factor. The partners plan an open specification.[1] A distinction must therefore be drawn between the chip being presented and this optical development, rather than attributing already available 400 Gb/s lanes to the G302.
Three figures that describe different things
Throughput is expressed here in bits per second. Aggregate capacity does not correspond to the usable throughput guaranteed to an application.
| Metric | What it describes | What it does not prove |
|---|---|---|
| 51.2 Tb/s | Switching capacity of the G302.[1] | The throughput available to each server. |
| 400 Gb/s per lane | Target per-lane capacity for USFP.[2] | The immediate availability of a complete link. |
| 6.4 Tb/s aggregate | 16 lanes at 400 Gb/s in the USFP project.[2] | An equivalent speed-up of AI applications. |
Density also depends on the number of lanes
The Cisco blog describes 16 lanes at 400 Gb/s, i.e. 6.4 Tb/s aggregate, with a possible evolution to 32 lanes and 12.8 Tb/s. It compares this proposal with 1.6 Tb/s pluggable modules to claim a fourfold increase in bandwidth density.[2] This factor therefore does not stem solely from doubling the per-lane throughput.
Maintenance matters as much as peak throughput
Cisco highlights the retention of field-replaceable modules and established operating practices for pluggable optics.[2] For an operator, the ability to service a module can weigh on the choice of architecture, just as much as throughput or footprint.
Higher density must nonetheless be assessed against cooling, cabling and spare-part availability constraints. The number of interventions, replacement lead times and compatibility conditions can change the real cost of an infrastructure over several years.
Energy gains have yet to be measured on the complete system
Cisco mentions energy efficiency potentially improved by around 1.3 times thanks to 400 Gb/s SerDes.[2] This indication does not mean that a data centre’s total consumption will fall in the same proportion. Higher capacity may also come with an increase in processing volumes.
A meaningful comparison must specify the load, the throughput actually carried and the scope measured. It should include network equipment, optics and cooling. The announcements reviewed do not present an independent benchmark that would allow the gain to be generalised to all environments.
SMEs can draw a sizing method from this
Most SMEs do not need to prepare a network for these speeds right away. However, an organisation that runs several GPU servers or buys remote computing capacity needs to understand what is limiting its workloads. The network is part of the service to be assessed, even when it is managed by a provider.
An illustrative scenario is to compare the same task on one and then several servers, observing duration, GPU utilisation and network traffic. If the time saving remains small, the team also examines storage and how the work is distributed. This test helps establish a diagnosis; it does not prove that the network is necessarily to blame.
The key IT skill: Systems and Network Administration
This development directly calls on Systems and Network Administration. Professionals must understand the paths data takes and distinguish theoretical throughput from observed performance. A capacity stated on a datasheet is not enough to size a service.
Their role includes analysing congestion, latency and dependencies on storage or compute. They must also check interfaces, optical modules and compatible software versions. These checks help avoid an investment that would simply move the bottleneck elsewhere.
Finally, network administration involves monitoring and incident recovery. Metrics must make degradations visible before they have a lasting impact on workloads. Having maintenance access and replacement procedures remains essential, even on a very high-performance infrastructure.
The value of USFP will therefore depend on performance measured on applications, equipment compatibility and operating conditions. To size an infrastructure, the real limits of compute, network and storage must first be identified.
Further reading
Four CCCLX reads on data centres, data and AI infrastructure.
AI powers Nokia: data centres reshape the networking market
Google Cloud Storage Intelligence Advisor detects storage anomalies
Databricks extends Delta Sharing to AI and aims to break down data silos
AI voice agents: Oracle details a LiveKit architecture on OCI
Sources
- 01
Cisco (12 October 2026). “Cisco Advances Scalability and Openness at OCP 2026”. Announcement of the Silicon One G302 switch and the USFP development project with NVIDIA.
- 02
Gartner, B. and Eatherton, W., Cisco Blogs (12 October 2026). “Cisco and NVIDIA Announce USFP: Expanding the Path to AI Connectivity”. Details on lanes, aggregate capacity and the gains announced for the USFP form factor.
PDFs, presentations, audio: Adobe wants to turn Acrobat into a content factory
Acrobat no longer just wants to open your documents For more than thirty years, Acrobat has mainly been used to read, create, edit, sign and share PDFs. Adobe now wants to give it a much broader role. On 9 September 2026, the company announced new capabilities based...
Salesforce plugs Claude into its CRM: Claudeforce prepares for life after the interface
What if the CRM soon no longer needed to be opened? For decades, using a CRM meant navigating between customer records, dashboards, menus and forms. With Claudeforce, Salesforce and Anthropic want to change this logic: rather than forcing users to go into Salesforce...
Kimi K3, Mistral, DeepSeek: Oracle turns OCI into an AI model hub
Oracle wants to make OCI a one-stop shop for AI Oracle is stepping up its efforts in the AI cloud battle by significantly expanding the catalogue of models available in OCI Enterprise AI. On 4 September 2026, the group announced in particular the integration of Kimi...
Agentic Data Cloud: Google rethinks data for the age of AI agents
AI agents are forcing Google to rethink data architecture AI agents promise to search for information, make decisions and act directly within company systems, but this autonomy depends on one essential condition: accessing the right data and, above all, understanding...
Google Cloud tackles the hidden bill of AI agents
AI agents raise a new problem: how much does a task really cost? As companies move from chatbots to AI agents able to work through multiple steps, the question of cost becomes much more complex. A chatbot generally handles a request and generates an answer, whereas an...
Workday, Navan, Avalara: ChatGPT Work moves into business software
ChatGPT Work is gradually leaving the chat window With GPT-6 Astra, OpenAI is no longer simply trying to improve the quality of ChatGPT’s answers. The company wants to bring its AI closer to the applications in which employees actually work. With the arrival of GPT-6...
From chip to cloud: Groq raises 350 million to scale up
Groq now wants to sell cloud rather than just chips Groq is pursuing a major strategic transformation. Long seen as one of Nvidia’s challengers thanks to its LPUs (Language Processing Units) specialised in artificial intelligence model inference, the group is now...
Muse Code: Meta launches its AI agent to tackle giant codebases
Meta wants to bring AI agents into complex software projects Meta is stepping up its efforts in AI-assisted software development with Muse Code, a new agent designed to work on large codebases, where traditional assistants can quickly lose the context needed to...
Qwen3.8-Max reportedly coded for 16 days without human intervention
Alibaba tests how far an autonomous AI developer can go Alibaba claims to have reached a new milestone in agentic software development with Qwen3.8-Max, its new flagship model launched in early August 2026. In an internal test, the model reportedly worked fully...
Two months, $1.1 billion: River AI is already exciting investors
A spectacular raise for a start-up barely out of the shadows Two months after emerging from stealth mode, River AI has just completed one of the most spectacular funding rounds of the year in artificial intelligence. The start-up founded by Igor Babuschkin, a former...
