Announcing Azure Copilot agents and AI infrastructure innovations

4 months ago 77

Learn astir the caller AI features and products In Azure announced astatine Microsoft Ignite 2025 to assistance modernize your unreality infrastructure.

The unreality is much than conscionable a platform—it’s the motor of translation for each organization. This twelvemonth astatine Microsoft Ignite 2025, we’re showing however Microsoft Azure modernizes unreality infrastructure astatine planetary scale—built for reliability, security, and show successful the AI era.

From scalable compute to resilient networks and AI-powered operations, Azure provides the instauration that helps customers innovate faster and run with confidence. Our strategy is anchored successful 3 cardinal areas, each designed to assistance customers thrive successful a rapidly changing landscape:

We’re strengthening Azure’s planetary foundation. We’re expanding capableness and resilience crossed regions portion optimizing datacenter design, powerfulness efficiency, and web topology for AI-scale workloads. Our services are zone-redundant by default, our borderline footprint is increasing to conscionable low-latency needs, and information and compliance controls are embedded astatine each layer. From confidential computing and sovereign unreality architectures to our information capabilities, Azure is engineered for spot by design.
We’re modernizing each workload. We’re advancing compute, network, storage, application, and information services with Microsoft Azure Cobalt and Azure Boost systems, Azure Kubernetes Service (AKS) Automatic, and Azure HorizonDB for PostgreSQL. We clasp and integrate with Linux, Kubernetes, and open-source ecosystems customers trust on.
We’re transforming however teams work. We’re embedding AI agents straight into the level done Azure Copilot and GitHub Copilot, bringing agent-based capabilities for migration, app modernization, troubleshooting, and optimization. These features region repetitive tasks truthful teams tin absorption connected architecture alternatively of administration, making an integral portion of however Azure runs end-to-end.

Agentic unreality operations: Introducing Azure Copilot

Azure is entering a caller epoch wherever AI becomes the instauration for moving your cloud. As environments turn much complex, accepted tools and manual workflows can’t support up. This brings america to a frontier moment, wherever AI and unreality converge to redefine operations. We telephone this agentic unreality ops: a caller exemplary for the AI era.

What is Azure Copilot?

Azure Copilot is simply a caller agentic interface that orchestrates specialized agents crossed the unreality absorption lifecycle, automating migration, optimization, troubleshooting, and more, freeing up teams to absorption connected innovation.

Azure Copilot aligns actions—human oregon agent—with your policies and standards, offering a unified model for compliance, auditing, and enforcement that respects role-based entree power (RBAC) and Azure Policy. It provides beardown governance and information residency controls, afloat visibility crossed agents and workloads, and lets you bring your ain retention for implicit power of chat and artifact data. To marque the operating exemplary genuinely agentic, we’re introducing six Azure Copilot agents—migration, deployment, optimization, observability, resiliency, and troubleshooting—in gated preview.

Learn much astir Azure Copilot successful our elaborate blog, and larn however to sign up for the preview.

Azure’s AI infrastructure: The backbone of modernization

Azure is built for reliable, world-class performance, delivering astatine planetary standard and speed.

With much than 70 regions and hundreds of datacenters, Azure provides the largest unreality footprint successful the industry. This unified infrastructure supports accordant performance, capacity, and compliance for customers everywhere.

AI infrastructure that delivers show and scale

We’ve reimagined however our datacenters are built and operated to enactment the captious needs of the largest AI challenges. In September 2025, we launched Fairwater, our largest and astir blase AI datacenter to date, and our newest site in Atlanta now joins Wisconsin to signifier a planet-scale “AI superfactory.” By utilizing high-density liquid cooling, a level web architecture linking hundreds of thousands of GPUs, and a dedicated AI WAN backbone, we’re giving customers unmatched capacity, flexibility, and utilization across each AI workload.

Aerial presumption of the Fairwater datacenter.

Azure is the archetypal unreality supplier to deploy NVIDIA’s GB300 GPUs astatine scale, extending our enactment from GB200 and continuing to specify the infrastructure instauration for the AI era. Each Fairwater tract connects hundreds of thousands of these best-in-class GPUs, millions of CPU cores, and monolithic storage—enough to clasp 80 cardinal 4K movies.

A cardinal portion of this improvement is our AI WAN—a high-speed web linking Fairwater and different Azure datacenters to determination information rapidly and coordinate monolithic AI jobs crossed sites. It’s engineered to support GPUs busy, trim bottlenecks, and standard workloads beyond the limits of a azygous location, truthful customers tin tackle bigger projects and get results faster. Driving down costs done innovation, we’ve acceptable a caller benchmark for secure, high-performance AI: Azure processed much than 1.1 cardinal tokens per second for connection models—the equivalent of penning 7 books per 2nd from a azygous rack.

Azure’s AI infrastructure puts supercomputing-level powerfulness successful each customer’s hands—enabling larger exemplary training, faster deployment, and broader idiosyncratic scope wrong a trusted, compliant environment.

Extending AI infrastructure innovation to your workloads

An breathtaking portion of our enactment connected AI datacenters is that the aforesaid architectural breakthroughs that let america to bid frontier models besides fortify Azure’s halfway services, straight benefiting each workloads.

One of these examples is Azure Boost, which offloads virtualization processes traditionally performed by the hypervisor and big operating strategy onto purpose-built hardware and software. Combined with our caller AMD “Turin” and Intel “Granite Rapids” virtual machines—plus the latest network-optimized and storage-optimized families—customers are seeing much than 20 GB per 2nd of managed-disk throughput and much than a cardinal input/output operations per 2nd (IOPS). More than a 4th of our planetary fleet is present Boost-enabled, and web throughput has doubled to 400 gigabits per 2nd for our general-purpose and AI SKUs. The infrastructure investments we’ve talked astir are already being utilized by leading-edge companies to bring services to billions of users.

Cloud-native apps and data

Azure Kubernetes Service (AKS) delivers secure, managed Kubernetes with automated upgrades and scaling. Paired with cloud-native databases similar PostgreSQL and Cosmos DB, teams physique faster and retrieve instantly.

We’re doing the enactment to span the powerfulness of the AI infrastructure into AKS by enabling cutting-edge GPUs to relation arsenic AKS nodes retired of the container oregon actively monitoring their health.

We haven’t stopped astatine the infrastructure layer. We’re besides reinventing however casual it is to instrumentality vantage of Kubernetes itself. That’s wherefore we introduced AKS Automatic. It embeds champion practices, automates infrastructure provisioning, and operates captious Kubernetes components to trim complexity and amended reliability. It handles the hard parts—patching, upgrades, observability, and security—so teams tin absorption connected innovation alternatively of infrastructure.

With AKS Automatic-managed strategy node pools, we’re making AKS Automatic adjacent lower-touch due to the fact that present you don’t person to tally captious Kubernetes components yourself. It’s wholly managed by the service. It moves cardinal services similar CoreDNS and metrics server to Microsoft-managed infrastructure, making it adjacent easier to absorption wholly connected your apps.

You not lone request easy-to-use exertion infrastructure; you besides request easy-to-use information tools for your applications. We’re introducing Azure HorizonDB for PostgreSQL, which brings breakthrough scalability and AI integration for next-generation applications. Azure DocumentDB is present mostly available—the archetypal managed database built connected the open-source motor we contributed to the Linux Foundation.

We are besides excited to grow our longstanding concern with SAP and denote the motorboat of SAP Business Data Cloud Connect for Microsoft Fabric, simplifying entree to information sharing crossed some platforms. Read the announcement blog to larn more.

Building for trust: Resiliency, operational excellence, and security

The satellite runs connected Azure’s unreality infrastructure. Every business, government, and developer depends connected it to beryllium reliable, secure, and ever available. That work drives everything we bash successful Azure. Our ngo is to physique the astir efficient, reliable, and cost-effective infrastructure level of the AI era, 1 that customers tin beryllium connected each day.

Resiliency is not conscionable a feature. It is simply a plan principle, a culture, and a shared committedness betwixt Microsoft and our customers. Every region, service, and cognition is built with that work successful mind.

At Microsoft Ignite, we are taking this committedness further with caller capabilities that fortify reliability, simplify operations, and assistance customers physique with greater confidence.

Raising the barroom connected operational excellence: Operational excellence means reliability is designed from the start. Every Azure portion is built with availability zones, redundant networking, and automated responsibility detection. We are extending that instauration with services similar NAT Gateway, present zone-redundant by default for improved web reliability without immoderate configuration required.
Empowering customers: With Azure Resiliency (public preview), we are co-engineering resiliency with customers. This caller acquisition helps teams acceptable betterment objectives, trial failover drills, and validate exertion health, strengthening readiness unneurotic earlier issues arise.

Evolving information for modern threats

We proceed to grow Azure’s information instauration with caller capabilities that marque extortion simpler, smarter, and much integrated crossed the platform. These updates fortify boundaries, automate defense, and bring AI-powered penetration straight into however customers support and run their environments.

Simplifying protection: Azure Bastion Secure by Default is built into the platform. It automatically hardens distant entree to virtual machines done RDP and SSH, reducing setup clip and risk
Strengthening boundaries: Network Security Perimeter, present mostly available, provides unafraid and centralized firewall to power entree to PaaS resources.
Better defense: And we’re making advancements successful the Web Application Firewall with Captcha for quality verification.

All of this builds connected Azure’s broader stack of confidential virtual machines, containers, hardware-based attestation, and encryption, supporting extortion from hardware done to application.

What does modernizing workloads look similar today?

Organizations are connected a travel to modernize. Most tally a premix of systems that span decades—from mission-critical databases to caller cloud-native services. Azure meets customers wherever they are, helping modernize applications and information with flexibility, openness, and built-in intelligence.

Rather than a one-off effort, modernization is truly astir re-architecting agility, scaling efficiently, and utilizing AI and unfastened technologies without sacrificing reliability oregon control.

To assistance simplify and accelerate the modernization journey, we’re investing to assistance you find and get to the champion destination for your workloads, whether it’s infrastructure arsenic a work (IaaS), level arsenic a work (PaaS), oregon bundle arsenic a work (SaaS).

Azure’s agentic migration and modernization tools marque it easier than ever to modernize your apps, data, and infrastructure with velocity and precision. For example, you tin determination existing .NET applications straight into a afloat managed environment—no refactoring oregon containers required—into the caller App Service Managed Instance (now successful preview).

On the information side, the next-generation Azure SQL Managed Instance (now mostly available) delivers up to five-times faster show and treble the retention capacity. And Azure Copilot and GitHub Copilot simplify SQL Server, Oracle, and PostgreSQL modernization.

Plus, crossed infrastructure—from VMware to Linux and IT operations—AI agents streamline migrations, trim licensing overhead, and automate patching, governance, and compliance, truthful modernization becomes a repeatable, intelligent motion.

Customers migrating and modernizing to Azure utilizing our agentic tools person shared the awesome results they person experienced. Here are conscionable a fewer examples:

.NET: More than 500,000 lines of codification upgraded and migrated successful weeks.
Java: Four times faster modernized applications than earlier utilizing the agents.

Read much astir GitHub Copilot app modernization.

Looking ahead

The committedness of the unreality was ever astir scale, flexibility, and innovation. With our innovations crossed our infrastructure, datacenters, Azure Copilot, services, and open-source contributions, that committedness expands to thrust your concern guardant each day.

The adjacent epoch of the unreality is inevitable. It’s agentic, intelligent, and human-centered—and Microsoft is helping pb the way.

Join america astatine Microsoft Ignite wherever you tin tune successful to our sessions to larn more:

Read Entire Article