From AWS to Cloudflare, 2025 was a year full of major outages and cyberattacks. In particular, these have exposed a reliance on a select few cloud providers and vulnerabilities in complex IT estates. It was also a year where AI has continued to transform how organizations operate.
New tools are redefining how IT teams manage their infrastructure, while entry level tasks are increasingly being taken over by AI, radically altering what skills are needed in the workforce and how to train employees in them.
Senior Technical Architect at Cloudhouse.
In 2026, these trends are set to govern how organizations approach managing and modernizing their IT estates. But what do companies need to do to ensure their infrastructure remains resilient, secure and adaptable in the year ahead?
The year of the AI agent
We are already seeing a shift in how organizations and their teams interact with AI. 2026 will definitely be the year of the AI agent – essentially, a virtual assistant that can work for you autonomously to achieve a set task or goal.
IT teams will be able to build out checks and balances automatically, and this means there can be a smarter implementation of tasks that go beyond ‘task A happened to task B’. Agents will be able to work in real time with minimal human input to ensure ongoing monitoring of IT estates.
Overall, this will help with building more resilient and self-healing architecture. On the legacy side, it will drive using AI to help understand outdated tech or building ways to communicate or translate it for modern use.
Chaos engineering will be crucial to preventing chaos
It’s the unfortunate truth that we’ll see more high-profile outages this year. After AWS, Cloudflare and Azure fell victim to such events this year, enterprises will need to assess their operational resilience for the new year.
One of the key ways of doing this will be to test real failover, i.e. simulating a real-world disaster like an outage, to evaluate the effectiveness of a disaster recovery plan.
This means running quarterly chaos experiments in production with controlled blast radius (the impact of a failure or breach) to validate actual recovery capabilities, not theoretical runbooks.
From a technical standpoint, teams will need to map critical business domains and isolate them architecturally. This will involve identifying which services absolutely cannot fail together and building hard boundaries between them.
Then, to get organizational buy-in, the importance of resilience will have to be defined in business terms for the board. IT teams will have to calculate Customer Lifetime Value (CLV) erosion from downtime (e.g. 25% customer churn after reliability failures), quantify regulatory penalties, and tie uptime metrics to revenue impact.
A greater shift to multi-vendor models
The threat of outages feels stronger than ever. Therefore, we expect to see more strategic workload placement and a mindset of “not running everything everywhere”.
Teams will start to place workloads based on provider strengths (AWS for breadth, Azure for Microsoft integration, GCP for data/AI) while ensuring critical paths have cross-cloud failover.
To achieve this, using infrastructure-as-code will allow for cloud-agnostic deployments, while mixing regional and specialized cloud providers will reduce concentration risk beyond the hyperscaler oligopoly.
Recurring outages could see teams adopting domain-driven designs to contain blast radius. For example, separating systems by business capability so a payment service failure doesn’t take down the entire e-commerce platform.
For specific use cases with steady resource needs, on-premise infrastructure might be seen as more cost-effective and reliable than cloud operating models.
Technical debt will continue to affect system reliability
Our recent report revealed that only 10% of companies in government, manufacturing and finance don’t have any Windows technical debt (the hidden costs and risks created when organizations delay updating or modernizing their IT systems).
This illustrates a broader picture where the use of outdated applications like Windows end-of-life apps is creating fragile integration points and security gaps.
Connections between modern cloud services and decades-old mainframes are difficult to monitor and become attack vectors for bad actors when outdated apps lack modern authentication, encryption, or patch management.
Legacy apps can’t participate in modern resilience patterns, so they become the reliability ceiling regardless of cloud infrastructure maturity.
Crucially, this tech debt is creating a talent gap. With a projected 100,000 developer shortfall, finding people to diagnose and repair legacy system failures during outages will take longer and cost more.
AI will play an active role in reducing these risks
With risks looming large, AI-powered resilience tools will grow in their importance for protecting IT estates. The use of AI-driven observability, for example, will be fundamental to predicting failure and catching issues before outages take place.
This will involve deploying platforms that can monitor the entire IT estate, application logs and business data to identify patterns indicating impending failures (memory leaks, integration timeouts) and trigger preventive actions automatically.
Self-healing automation will then address common failure scenarios without waiting for humans, while continuous AI-driven compliance monitoring and drift detection will automatically flag new risks in legacy environments and generate remediation recommendations.
All of this will give IT teams more time to strategize and proactively manage their infrastructure.
AI will also be harnessed as an effective way of overcoming outdated codebases and languages. For example, Generative AI can crawl decades-old source code, translate it to natural language, and create business specifications that would take human teams months to produce manually.
This includes automatically converting legacy languages to modern stacks predictably and at scale.
And with regards to the talent gap, AI will be able to offer real-time coding suggestions and support for developers unfamiliar with legacy languages, multiplying productivity of scarce specialist workers.
2026: Less reliance, more proactivity
The risks and threats to IT have never felt greater. But the tools in managing IT estates have never been more advanced too. AI agents, chaos engineering and a move away from single cloud suppliers all look set to dominate the year ahead.
As companies seek to protect themselves against costly outages and cyberattacks, modernizing their legacy applications and continuously monitoring their IT estates for risks will be essential to ensuring resilience.
To stay ahead, IT leaders should start by mapping legacy risks and prioritizing technical debt remediation, piloting AI agents for routine tasks, and implementing infrastructure-as-code to enable cloud portability.
Schedule quarterly chaos engineering drills to validate resilience under real-world conditions, and quantify the financial impact of downtime, from lost revenue to customer churn, to secure board-level sponsorship.
These steps will not only harden IT estates against outages but also position resilience as a strategic advantage rather than a reactive measure.
We’ve featured the best endpoint protection software.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
