VergeIO, which created the first Ultraconverged Infrastructure (UCI) platform, has launched VergeOS 26 to modernize infrastructure. VergeOS26 lets enterprises deploy AI infrastructure without cloud dependencies, strengthens ransomware defenses with immutable data protection, and consolidates legacy architectures into a single software platform that replaces VMware, SAN storage, and complex networking stacks.
Integrated directly into VergeOS, VergeIQ gives organizations secure, shared access to large language models (LLMs) for inference, analysis, and conversation – and does it entirely within the customers’ own data centres. VergeIQ also includes OpenAI-compatible services, vendor-agnostic GPU support, and generative AI intelligence. Teams gain enterprise-wide access to AI while maintaining control over data, context, and compliance.
So what is a UCI platform?
“With the UCI platform, we are positioning ourself as the new version of hyperconverged [HCI],” said George Crump, Chief Marketing Officer at VergeIO. “The stacking effect with hyperconverged creates a lot of inefficiencies and adds a lot of latency. The result in many cases is poor performance from HCI. HC integrates compute, storage, and virtualization into a single cluster, simplifying deployment compared to traditional three-tier architectures. Each node contributes local resources that are pooled and managed through software, reducing reliance on external SANs and complex network configurations. However, HCI still operates as a collection of layered modules – a hypervisor, a virtual storage appliance, and often an external network stack — tied together by orchestration and vendor-specific plug-ins. While this reduces complexity versus legacy designs, it still leaves organizations managing multiple subsystems and interfaces In UCI it’s all integrated, much more efficient and can scale much higher.
VergeIQ is the brand name for the AI, while while VergeOS is the code. Verge IQ also gives you the AI and inferencing capabilities. Although it is all the same code.”
VergeIQ, the AI component of VergeOS, is the company’s brand for the embedded inference engine and automation layer. VergeOS itself contains four integrated elements:
- VergeHV – the hypervisor handling compute and virtualization
- VergeFS – the storage and data protection layer
- VergeFabric – their software-defined networking (Layer 2 and 3)
- VergeIQ – the new AI and inference engine
All four are part of a single unified code base. You don’t bolt them together — they’re built as one system.
Ultra-converged Infrastructure (UCI) represents the next step in the VergeOS evolution, Crump stressed. “It unifies compute, storage, networking, data protection, and management intelligence into a single software code base – an infrastructure operating system. Rather than coordinating separate components, UCI runs everything natively, providing higher performance, simpler operations, and hardware independence. VergeOS exemplifies this model by merging its hypervisor (VergeHV), storage layer (VergeFS), software-defined networking (VergeFabric), and AI engine (VergeIQ) into one integrated platform that replaces the complexity of HCI with true operational unity.
Crump identified several large bonuses for VergeIQ. Most enterprises today manage five or more separate infrastructure platforms, resulting in unnecessary complexity and increased risk. VergeOS 26 consolidates these functions into a single software-defined platform, abstracting infrastructure services from hardware while delivering capabilities that legacy vendors cannot match.
First, VergeIQ lets organizations deploy AI infrastructure without cloud dependencies.
“We’ve integrated private GPT, and while a user would call it something else, they can also deploy something like it,” Crump indicated. VergeIQ includes OpenAI-compatible services, vendor-agnostic GPU support, and generative AI intelligence.” Teams gain enterprise-wide access to AI while maintaining control over data, context, and compliance.
“All of it benefits from VergeOS’s built-in deduplication and storage efficiency”, Crump added. “For example, in one demo I created a “Financial Assistant” that analyzes 100 PDFs representing three companies and summarizes which would be the best investment. It runs entirely on local data, so it doesn’t hallucinate or invent information. Even without a GPU, the system can perform inference using smaller open-source models from Hugging Face – such as Gemma, Llama, or Phi. When customers later add GPUs, they can scale to larger models for deeper analysis. This lets them experiment before investing tens of thousands of dollars in hardware.
“There is also a brand new GUI,” Crump said.
Secondly, VergeIQ strengthens ransomware defenses with immutable data protection.
“There is only a limited amount that can be done at this stage with immutable ransomware, because it is basically a 1.0, and the only way you can delete something immutable is with a schedule,” Crump stated. “You CAN get into trouble if you ignore the warnings and fill up the space.”
Before, VergeOS had snapshot capability – the difference being that they were not immutable.
“Immutable ransomware protection, Private AI with VergeIQ, and Unified Infrastructure replace legacy multi-vendor stacks now,” Crump said. “Multi-vendor stacks are as bad now as they have ever been. You generally need a new infrastructure just to run AI. On the other hand, we are one piece of software that can run on any hardware, and we are not tied to any vendor.”
Crump contrasted this with VergeOS competitors.
“Nutanix relies on an external immutable object store,” he said. “VMware ESXi can’t create immutable snapshots at all. The newest VSAN version includes them, but it limits the number you can keep, and performance drops as the number grows.In VergeOS, immutable snapshots are integrated into the core code, so you can retain as many as you want for as long as you wish without affecting performance
The only exception is system-level protection: if a user misconfigures snapshots and fills all available space, VergeOS will override immutability only to preserve system stability.
“For us, immutability is absolute,” Crump said. “A snapshot can’t be deleted or altered except by its retention schedule. Some vendors allow support overrides or “backdoor” deletion keys, which we believe defeats the purpose. Beyond that, our recovery process is comprehensive. Because VergeOS captures the entire virtual data centre, including IP addresses and configurations. Customers can restore their environment exactly as it was before the ransomware attack. This has always been one of our strongest advantages; the addition of immutable snapshots completes the protection story.”
The third thing Crump highlighted was the way in which VergeIQ consolidates legacy architectures into a single software platform that replaces VMware, SAN storage, and complex networking stacks. This is the only part of the system which is brand new rather than just enhanced.
“A typical data center today might use VMware for compute, Pure Storage for disks, Cisco or Juniper for networking, and now entirely new systems for AI workloads,” he said. “You end up with multiple storage platforms, one for “AI-ready” data, another for archives, and so on. VergeOS replaces all of that with one software platform that runs on any hardware. Customers can use NVIDIA, AMD, or future Intel GPU, with no vendor lock-in. We support any standard x86 server and off-the-shelf SSDs. The result is simplified infrastructure and lower costs.”
The goal of VergeOS has largely been to replace VMware, in a world where Broadcom has taken much of that business. Crump said that VergeOS is doing well there.
“I spent more than a decade running a reseller myself, so I understand that world well,” he said. “The way we’ve built VergeIQ gives VARs, MSPs, and CSPs a second chance to engage in AI – even if they felt they missed the first wave. We’ve made AI easy to implement, easy to sell, and simple to demonstrate ROI. Traditional enterprise AI projects often cost millions before showing results. With VergeIQ, there’s no additional investment – it’s part of VergeOS.”
Crump explained how partners can benefit from this.
“Partners can show customers how to put their private data to work immediately,” he said. “They don’t need to train massive models or hire data scientists. It’s about asking meaningful questions of the data they already own. That delivers visible value fast – and helps partners prove real return on investment. I’m finalizing a selling guide now to help partners start positioning AI with VergeOS in exactly that way.”