Staying relevant as an IT professional

The fears about AI displacing experienced, educated workers are not the stuff of urban legend: the numbers are hard to ignore. Tech layoffs have surged in 2026, with more than 100,000 jobs already cut worldwide by early May. Of those cuts, nearly 48% have been attributed to reduced need for human workers due to AI and workflow automation, reports Tom’s Hardware.

 

Anecdotally, AI adoption is already slowing hiring for junior developers and generalised IT roles. The good news? Specialised AI-focused positions remain in high demand, according to recent research by Motion Recruitment.

 

IT infrastructure professionals who are willing to lean into the data challenges that AI creates can bolster their chances to survive layoffs and earn promotions. Enterprise organisations are at varying stages of AI maturity, but across the board, they need people who understand storage, data management, and governance at a deep level.

 

Two forces are converging right now that make this a pivotal window to shift toward AI-related skills and knowledge in IT:

•  First, enterprise storage costs are surging unexpectedly this year, driven by AI demand, with major storage vendors announcing price increases of 20% to 50% and growing.
• Second, data readiness is a troubling gap in the market. According to a recent Cloudera/HBR report, only 7% say their organisation’s data is completely ready for AI adoption, and more than one-quarter (27%) report their data is not at all ready. 

Both problems are solvable by enterprise IT teams with the right skills and strategies.

 

 

Memory pricing: pushing IT budgets overboard

The hardware situation has significantly worsened in 2026 due to SSD and DRAM shortages and a 30% to 100% price surge from IT vendors across servers, flash storage, PCs and mobile phones. For IT leaders who have spent years planning capacity assuming unit storage costs would keep falling, this is a genuine disruption to their budget model.

 

The practical response is not simply to absorb the cost increases or to delay AI projects. IT professionals need skills and tools to continually analyse and right-place the petabytes of file and object data sitting on expensive primary storage. 

 

On average, 70% of enterprise unstructured data is inactive and cold. That means high flash prices are spent on data that hasn’t been touched in months or years. Moving infrequently accessed data to lower-cost object storage or cloud tiers can offset the need for new storage purchases and free up budget for AI infrastructure that actually needs high-performance flash.

 

Storage optimisation and tiering strategies that don’t lock up your data or lock you into a single vendor are fundamental practices for IT organisations today to address not only current market conditions but also the long-term problem of hybrid unstructured data estates growing by 20-50% or more annually. 

 

Tactics

1. Look to build expertise in analysing file and object storage footprints and identifying cold data.
2. Understand storage cost modelling and FinOps best practices across hybrid infrastructure.
3. Working with department heads on tiering policies is important and requires the right balance between metrics and a partnering mindset.
4. If the cloud is part of your Infrastructure mix, you’ll need to be clear on changing egress and recall policies and fees among the providers.
5. In parallel, depending upon the tiering solution that you choose, costly rehydration penalties can be avoided. With storage vendor tiering solutions, IT must retain on-premises capacity for archived data in the event users need to access it again or if it must be moved to new storage. Storage-agnostic tiering can be a better fit to avoid this lock-in situation. 

The AI bottleneck of unstructured data quality

AI runs on unstructured data, yet most of it has been accumulating for decades with little governance, classification, segmentation or quality control. Most enterprises have significant amounts of duplicate, orphaned and low-authority data that must be culled. Next, IT needs ways to label and restrict access to sensitive and regulated data that shouldn’t be accessible to AI engines. Thirdly, most system-generated metadata is not sufficiently specific for the contextual identification required to accurately curate data sets for specific AI projects. Metadata enrichment tools are critical.

 

Without tools and methods to clean, govern and refine unstructured data, organisations risk dumping massive, irrelevant datasets into AI systems and getting poor results. Meanwhile, CSOs and CEOs will be on fire about the wrong data (PII, company secrets) getting into LLMs and entering the public realm, followed by potential fines, breaches and customer defection.

 

One financial services firm reported a 135% improvement in the accuracy of an AI agent that summarised financial documents simply by feeding it curated unstructured data, with no changes to the agent itself. Data preparation mattered more than model tuning. 

 

 

Tactics

In 2026, the unstructured data world is doing the work that the structured data world started doing in the 1990s to clean, format and categorise data for BI. Unstructured data now needs this attention for AI. 

1. First, optimise the use of system-generated metadata for unstructured data. Time of creation, owner, time of last access, file type, size and growth rates are great starting points for building a data classification and segmentation program.
2. Define the strategies and tools to enrich metadata systematically. Doing so gives AI project leaders a pathway to better outcomes by enabling them to search for files using specific keywords. Further, you can cut costs by 50% to 90% by sending much less data to AI processors.
3. Avoid custom and often brittle ETL connectors and other methods used for enriching structured metadata, which are manual and don’t scale to the size of unstructured data. Modern unstructured data management technologies can offer automated, serverless technology for rapid metadata extraction.
4. Develop capabilities for sensitive data discovery, PII detection and policy-based retention and deletion for unstructured data. This can positively support legal, security, and compliance functions.
5. Automated sensitive data management is a core requirement for governed AI data workflows. The right strategy can make or break the organisation’s overall AI success.
6. Work with security and compliance teams on data risk management, security tool integration and emerging requirements to monitor AI tools, data workflows and outputs. 

The IT professionals most at risk from AI are those whose role is narrow and task-oriented. To increase your job security, focus more on data quality, cost optimisation, governance, and AI pipeline readiness.

 

Knowing what data you have, especially unstructured data, where it lives, the value it offers and its risk profile is now a core competency for any infrastructure leader. That is not a threat to the IT role. It is a redefinition of it. The organisations investing in AI are not looking to eliminate the people who can make that investment pay off. They are seeking them out.

 

---

 

Krishna Subramanian is the co-founder of Komprise

 

Main image  courtesy of iStockPhoto.com and WANAN YOSSINGKUM