What Eight Infrastructure Companies Revealed About Enterprise IT in 2026

Eight infrastructure vendors at IT Press Tour #66 reveal how VMware exodus, AI workloads, and flash economics are reshaping enterprise IT strategy.

The 66th IT Press Tour brought together eight companies with a common thread: they're all responding to the same infrastructure pressure points. After three and a half days of presentations in Palo Alto, a clear picture emerged of where enterprise technology is headed and why traditional approaches are breaking down.

Three Market Forces Reshaping Infrastructure

The VMware Exodus Creates Opportunity

Broadcom's acquisition of VMware triggered the largest infrastructure migration in a decade. Three companies at the tour—VergeIO, Scale Computing, and InfoScale—are directly positioned to capture organizations looking for alternatives. But this isn't just about licensing costs. It's about rethinking what consolidated infrastructure means when you don't need to maintain compatibility with a 20-year-old architecture.

The pattern across these vendors: unified platforms that combine virtualization, storage, networking, and management in ways VMware's component-based model never could. VergeIO calls it "ultraconverged infrastructure." Scale Computing emphasizes edge orchestration across 100,000+ sites. InfoScale focuses on application resilience rather than infrastructure resilience.

The business case is consistent: 50-70% cost reduction, faster deployment, and operational simplicity that lets smaller IT teams manage larger environments.

AI Workloads Drive Infrastructure Requirements

Every presentation addressed AI in some form. Not as a marketing checkbox, but as a practical challenge that existing infrastructure wasn't built to handle.

Globus is orchestrating petabyte-scale data flows for AI training pipelines at national laboratories. Helikai is solving the "pilot purgatory" problem where 76% of enterprises can't scale AI agents to production. VergeIO added VergeIQ for on-premises AI inference.

Zettalane architected hybrid storage specifically for AI's read/write patterns. Novodisq packed 11.5 petabytes in 2U with onboard AI acceleration. Scale Computing is processing Voice AI locally at thousands of Taco Bell drive-thrus. The Lustre Collective is adapting their parallel filesystem for AI training workloads that differ fundamentally from traditional HPC simulations.

The common thread: AI doesn't fit the infrastructure paradigms we built for transactional databases and web applications. Data gravity matters. Locality matters. Real-time processing at the edge matters. Organizations trying to run AI on traditional storage and networking are discovering those systems become the bottleneck.

Flash Economics and Density Pressure

Multiple vendors mentioned the same constraint: NAND flash prices doubled in 2026, and all production capacity is sold out. This isn't a temporary supply issue. It's a structural shift that changes infrastructure economics.

Novodisq's response was to build their own SSDs optimized for write-once, read-sometimes workloads. Zettalane architected a hybrid approach using local NVMe for hot data and cloud object storage for capacity. The Lustre Collective is adding erasure coding to reduce storage overhead.

The pattern: organizations can't just throw more flash at the problem. They need architectures that match performance tiers to actual access patterns.

What Each Company Delivers

Globus: Research Data Orchestration

Globus manages research data flows for 170,000+ researchers across national laboratories and academic institutions. Their platform automates data movement between scientific instruments generating massive datasets and the storage systems, cloud resources, and compute clusters that process them. The value proposition: compliance-ready data handling for HIPAA, PII, PHI, and CUI with automated workflows that eliminate manual data management.

Helikai: Production-Ready Agentic AI

Helikai addresses the 76% failure rate in scaling AI agents from pilot to production. They offer pre-built AI agents (Helibots) for specific business processes in legal, life sciences, media, and IT operations. Their approach: small, discrete agents that do one task well, with human-in-the-loop workflows and full audit trails. Organizations get production AI in weeks rather than the months or years required for custom development.

VergeIO: Single-Codebase Infrastructure

VergeIO consolidates virtualization, storage, networking, and AI into one software platform. Their VergeOS integrates what traditionally required five separate products: hypervisor, SAN, network switches, backup, and disaster recovery. Customer wins include 85% cost reduction at Pfeiffer University and zero-downtime migration from VxRail at Alinsco Insurance. The pitch: simpler infrastructure that lets organizations reuse existing hardware while cutting costs in half.

InfoScale: Application Resilience

InfoScale provides high availability and disaster recovery for mission-critical applications across hybrid environments. Rather than protecting infrastructure components, they protect entire application stacks—monitoring databases, middleware, networking, and storage as a unified system. The platform claims 98% reduction in yearly downtime through automated failover that happens in seconds rather than minutes. Organizations running Oracle, SAP, SQL Server, and other business-critical applications get continuous resilience without extensive manual runbooks.

Zettalane: Hybrid Cloud Storage

Zettalane delivers software-defined storage that uses local NVMe for metadata and hot data while streaming large blocks from cloud object storage. This hybrid approach provides enterprise performance at cloud economics—they claim 70% cost savings versus traditional cloud storage. Performance benchmarks show 2.3M IOPS on GCP with sub-millisecond latency. The target: organizations needing high-performance storage without the capital expense of all-flash arrays.

Scale Computing: Unified Edge Platform

Scale Computing manages over 100,000 edge sites through centralized orchestration. Their platform combines virtualization, networking, security, and application management in one system with zero-touch provisioning. Customer results include 70% cost reduction and 99%+ uptime across deployments at Yum! Brands (Taco Bell Voice AI), Advance Auto Parts (4,500+ stores), and Chevron (8,500 fuel stations). The value: distributed operations without distributed IT staff.

Novodisq: Ultra-Dense Storage

Novodisq built custom SSDs and architected storage blades delivering 11.5 petabytes in 2U at one-tenth the power consumption of traditional storage. Each blade includes AMD Versal SoCs with ARM cores, FPGA fabric, and AI accelerators co-located with flash storage. The architecture enables real-time data processing during ingestion—particularly valuable for genomics, video surveillance, and workloads requiring data sovereignty. The business case: storage and compute together reduce infrastructure footprint and operational costs dramatically.

The Lustre Collective: HPC Filesystem Expertise

The Lustre Collective formed as an independent support company for the parallel filesystem that powers 60%+ of the world's fastest supercomputers. Lustre handles exascale deployments delivering 10TB/s+ throughput and supporting 100,000+ GPUs. Every major cloud provider now offers managed Lustre services. The Collective provides vendor-neutral expertise for organizations running HPC and AI training workloads that require massive parallel I/O performance traditional filesystems can't deliver.

What This Means for CIOs

The infrastructure choices you make in 2026 will determine your AI capabilities in 2027. Organizations betting on traditional architectures are discovering those systems become the constraint.

The vendors at this tour share common characteristics: they're addressing infrastructure problems that cloud providers and legacy vendors aren't solving. They're building for workloads that require different tradeoffs than what worked for transactional applications. And they're targeting organizations that need production systems rather than technology experiments.

The broader pattern: infrastructure is consolidating. The days of best-of-breed component selection are giving way to platforms that handle multiple functions through integrated software. Whether that's VergeIO combining five products into one, Scale Computing unifying edge operations, or Zettalane hybridizing storage tiers, the trend is clear.

Organizations with distributed operations, AI workloads, or mission-critical applications should pay attention to these vendors. They're solving real problems with production deployments rather than slideware.

The 68th IT Press Tour is scheduled for June 2026 in Boston. Based on the infrastructure pressures visible in January, expect more companies addressing AI infrastructure, edge computing, and post-VMware consolidation.