Cisco: With Nutanix to the VMware Alternative
Cisco integrates Nutanix, introduces Unified Edge, and looks into the future of energy distribution for AI hardware with Project Edison.
(Image: Cisco)
- Jens Söldner
- Marco Brinkmann
In addition to the AI agents dominating its house show Cisco Live, Cisco is targeting the physical and licensing aspects of IT infrastructure, aiming to make them more customer-friendly. Providing converged systems has long been an important trend. Cisco's own HyperFlex offering is a thing of the past, but the Nutanix platform is now becoming a firm part of Cisco's runtime agreements – a clear positioning against Broadcom's VMware.
Unlike last year, when Cisco introduced new AI servers UCS C880A M8 and UCS C845A M8 and their integration with Intersight, this year the focus in this area was entirely on the Cisco Unified Edge. The system, announced in November 2025 on a smaller scale, was now presented to the general public "to touch" at Cisco Live 2026 in Amsterdam. Other topics included energy efficiency and cooling, especially at the edge. Cisco showcased Immersive Cooling for Unified Edge and provided insights into Project Edison – a disruptive form of power distribution.
Nutanix – Part of Cisco Enterprise Agreements
The deepened partnership with Nutanix is strategically important – and the associated quiet end of HyperFlex. The strategic collaboration between Nutanix and Cisco has existed for over two years. Cisco will integrate the Nutanix Cloud Platform (NCP) directly into its Enterprise Agreements (EA) in the future. Customers can thus obtain Nutanix licenses as part of their existing Cisco software contract, instead of having to conduct separate negotiations.
For companies looking for alternatives after Broadcom's acquisition of VMware and the associated licensing changes, this eliminates the hurdle of establishing a new supplier relationship. Those who want to build a second virtualization foothold or plan migrations can do so within the existing Cisco contract. Nutanix also benefits: its own software reaches large software agreements through the Cisco sales channel, which Cisco sellers place with existing customers anyway. Access to this sales apparatus would hardly be achievable for Nutanix alone with this breadth.
Unified Edge: Touch Less, Accomplish More
Cisco delivers the appropriate hardware home for this software with the Unified Edge. Announced back in November 2025 – at an unfavorable time in the year-end business – it was now presented to a broader audience on the big stage of Cisco Live.
Behind Unified Edge are two convictions. First: With increasing complexity, the need to accomplish more with a single instrument grows. The manufacturer therefore bundles compute, GPU, switching, and firewall in individual modules that can be centrally managed as zero-touch infrastructure but operate autonomously locally. Second: Cisco believes in distributed IT. Not all data can be meaningfully processed in a central location – a kind of "product pre-assembly" at the edge is intended to reduce latency and bandwidth requirements.
At the main stage of Cisco Live, the manufacturer prominently presented the concept. Compared to the November launch, further module options have been added, including Nvidia GPUs and additional network modules. This creates a modular system for branch offices, subsidiaries, and edge locations that transfers the Nutanix software logic to the hardware.
Videos by heise
Project Edison: Distributing Energy, Not Just Data
The research project "Edison," which Denise Lee, Vice President of Cisco's Engineering Sustainability Office, presented publicly for the first time at Cisco Live, offers a glimpse into the more distant future. The starting point: upcoming AI accelerators will consume so much energy and generate so much waste heat that traditional racks will reach their physical limits. Nvidia's upcoming Rubin generation, for example, is expected to deliver about five times the computing power of Blackwell – with an energy consumption that is likely to at least double.
According to Cisco, it will no longer be enough to just distribute data in the future – the power supply must also be decentralized and as loss-free as possible. The concept itself is not new: Cisco experimented with smart building initiatives and inline power ten years ago. At that time, however, there was no pain point. With the power consumption of current and future AI hardware, this pain point now exists.
Project Edison explores technologies to safely transmit up to 600 watts of direct current over a single pair of wires. Technically, this goes far beyond current Power-over-Ethernet (PoE), which reaches its maximum at 100 watts (IEEE 802.3bt Type 4). The goal: to flexibly route energy to where AI workloads are currently experiencing peak loads, thereby alleviating thermal hotspots in data centers. In addition, Cisco now offers switches in liquid-cooled variants for the new Silicon One G300 with 102.4 Tbit/s throughput for the first time. The manufacturer thus aims to align the entire physical infrastructure with the requirements of the AI era.
(vbr)