Enterprise technology is moving towards systems that combine specialised computing, operational intelligence and greater autonomy. This week’s developments include expanded quantum-classical partnerships, a new agentic operations platform, autonomous software engineering within Microsoft Azure, and a cybersecurity acquisition focused on open-weight AI. Together, they reveal how organisations are building the infrastructure and governance needed to move advanced AI and computing from research and experimentation into practical business environments.
Three Developments Worth Watching
-
Quantum-classical convergence: PsiQuantum and AMD, alongside IBM, Fermilab and NVIDIA, are advancing different approaches to integrating quantum processors with classical high-performance computing.
-
Agentic operations at scale: Genpact Onyx aims to turn enterprise process knowledge into governed AI execution, while Factory’s expanded Microsoft collaboration brings autonomous software engineering into customer-controlled Azure environments.
-
Security for open AI: Palo Alto Networks’ planned acquisition of Applied Compute reflects growing attention to securing open-weight AI models and the infrastructure used to deploy them.
5 AI and Enterprise Technology News Stories
PsiQuantum and AMD Expand Partnership to Build Hybrid Quantum-Classical Computing Infrastructure
PsiQuantum and AMD have expanded their collaboration to co-design computing infrastructure for utility-scale, fault-tolerant quantum computers. The partnership combines AMD EPYC CPUs, Instinct GPUs and the ROCm software stack with PsiQuantum’s photonic quantum technology. Building on earlier work with AMD Versal FPGAs, the companies are also optimising PsiQuantum’s PRISM machine-learning decoder for low-latency quantum error correction.
The collaboration reflects the emerging view that practical quantum computing will depend on close integration with classical high-performance computing rather than replacing it. CPUs, GPUs and quantum processors will have distinct roles in orchestration, simulation and computation. Real-time error correction is a particularly important challenge: progress in this area could help move photonic quantum systems towards larger, fault-tolerant architectures, although commercial utility remains a longer-term objective.
IBM, Fermilab and NVIDIA Plan a Modular Quantum Computing Co-Design Platform
IBM, Fermilab and NVIDIA plan to develop a modular quantum computing co-design platform that integrates quantum processors with accelerated classical computing. The initiative will pair IBM quantum systems, including its Nighthawk processor, with NVIDIA DGX B200 systems. Researchers will investigate modular architectures, quantum interconnects and hybrid workloads, with Fermilab coordinating scientific applications in high-energy and nuclear physics. The partners will also explore how quantum resources can be integrated into supercomputing workflows through shared scheduling tools.
The project addresses a central obstacle to scaling quantum computers: connecting multiple processors while preserving fragile quantum information and coordinating their workloads with classical systems. Its scientific focus is significant because complex simulations could eventually benefit from hybrid architectures. The work is still at the research and co-design stage, but it could help establish the software, scheduling and hardware foundations needed for future fault-tolerant quantum systems.
Genpact Onyx Platform Launches for Enterprise Operations
Genpact has launched Genpact Onyx, an agentic operations platform designed to build, run and govern AI agents in enterprise workflows. Its Process Intelligence Core codifies operational knowledge, including workflows, exceptions and expert decisions across areas such as finance, risk, procurement and supply chain. The platform can connect with existing enterprise technology and supports different AI models. Agents initially propose actions for human approval, with autonomy expanding as their performance is established.
The platform’s focus on process intelligence addresses a weakness in many enterprise AI deployments: models may generate capable responses but lack the context needed to execute complex business processes reliably. By incorporating exceptions and human decisions into its operating knowledge, Genpact aims to make automation more context-aware over time. Its success will depend on how effectively it handles real-world complexity while keeping actions traceable, auditable and accountable.
Factory Expands Microsoft Collaboration for Autonomous Software Engineering
Factory is expanding its collaboration with Microsoft to help enterprises deploy autonomous software engineering agents, known as Droids, within their Azure environments. The platform can operate within customer-controlled networks and use existing identity, source-control and governance systems. Droids can investigate software issues, identify root causes, write and test fixes, and open pull requests for human review. The collaboration also introduces operating-system containment through Microsoft Execution Containers, limiting agent access to approved files and network destinations.
The development takes AI coding beyond code generation towards end-to-end engineering workflows. Allowing agents to investigate production alerts and prepare tested fixes could reduce maintenance backlogs and accelerate development cycles. However, keeping workloads inside customer-controlled environments and retaining established review processes are essential for regulated organisations. The model illustrates how autonomous engineering can expand without requiring enterprises to surrender control over code, infrastructure or approvals.
Palo Alto Networks’ Applied Compute Acquisition Targets Open-Weight AI
Palo Alto Networks has entered a definitive agreement to acquire Applied Compute in an all-stock transaction valued at $4.5 billion, subject to adjustments and regulatory clearance. Applied Compute specialises in post-training and inference for open-weight AI models. The acquisition is intended to help enterprises train, adapt and deploy these models securely across public and sovereign cloud environments. Applied Compute’s research team will join Palo Alto Networks, with co-founder and CEO Yash Patil set to become Co-Chief AI Officer following the transaction’s close.
The deal signals that cybersecurity providers increasingly see AI model infrastructure as part of the security landscape. Enterprises adopting open-weight models gain flexibility to customise systems around their data and operating requirements, but must also manage risks involving model integrity, data exposure and deployment controls. Combining model optimisation capabilities with cybersecurity expertise could give Palo Alto a stronger position in enterprise AI protection. The transaction is expected to close in the company’s fiscal third quarter of 2027.
AITech365 Featured Article
AI Teammates vs. AI Assistants: What Changes When Systems Own Tasks?
This feature examines the shift from reactive AI assistants to task-owning AI teammates that can coordinate applications, retain workflow context and execute multi-step responsibilities. It explores productivity opportunities alongside the governance requirements of agent identity, permissions, human approvals and auditability. Its central argument: enterprise autonomy must be earned through reliable execution, clear boundaries and accountable human oversight.
AITech365 Weekly Observations
This week’s developments reveal two connected transitions: computing is becoming more heterogeneous, while enterprise software is becoming more autonomous. Quantum-classical partnerships are laying groundwork for future scientific workloads; agentic platforms and software engineering systems are moving AI into operational execution. At the same time, the Applied Compute acquisition highlights how security must evolve alongside open and customisable AI models. The common requirement is coordination: across processors, applications, data and human oversight. As AI systems take responsibility for more complex work, will trust and governance develop quickly enough to keep pace with their growing capabilities?


