Opinion
Between AI, cloud and physical collapse: Building resilience in an unstable world
"In an era of soaring hardware prices and AI agents becoming the beating heart of the organization, infrastructure is no longer only 'a topic for the technologists', but is becoming a full-blown business compass," writes Oren Porat, CEO of the Infrastructure Division at peax.
If the last decade in Israeli high-tech was defined by a technological arms race and the eager adoption of every new trend, then 2026 marks the sobering-up moment. As we saw during Operation "Operation Roaring Lion" even the public cloud is no longer immune. The physical strike on data centers in the Persian Gulf, which disconnected banks and investment firms from critical services, proved that global internet infrastructure sits right in the middle of the battlefield. CEOs and CFOs are going to find that reality has shifted under their feet. The era of experimentation is over. Now we have to deal with bottom-line questions: Are our economic models still relevant? Are we managing strategic assets, or are we just paying uncontrollable ballooning bills?
From my deep familiarity with the critical junctions of Israel’s computing infrastructure, I see three fundamental trends that will set the business agenda in the period ahead. These are not technical questions about which server to buy. They are questions of strategy, profitability and sovereignty.
The Organizational Brain Comes Home
Many remember the sweeping promise that the public cloud would solve every enterprise computing challenge. But in 2026 the pendulum is swinging back and many organizations are grappling with the "morning-after invoice." AI costs in the cloud are spiraling out of control and total dependence on global providers creates real operational and business risks.
Recent events in which physical damage to data centers caused service outages made it painfully clear just how vulnerable offshore, distributed infrastructure is to geopolitical shocks. Updated figures from IDC and research by Barclays Bank point to a clear trend: between 80% and 86% of CIOs are already planning to move critical workloads back from the public cloud to local data centers. This repatriation focuses on persistent applications, data-heavy systems and sensitive information, with the goal of eliminating hidden traffic costs and ensuring the low latency required by autonomous AI agents.
Picture a bank running an AI agent whose job is to approve loans on the spot. The agent needs to scan financial history, assess risk and deliver an answer in seconds. When the infrastructure sits on a remote server overseas, any hiccup in the public cloud, whether a communications delay or a submarine cable fault, isn’t just a technical glitch. It’s a full shutdown of the digital production line. Moving to a hybrid cloud allows for the performance and sovereignty of on-premise infrastructure while still paying through a flexible, consumption-based operating model.
Stopping the Budget Bleed
The big story of the next two years is the shift from merely using AI to deploying autonomous agents that do real work. But this is exactly where we’re hitting a wall because of a severe global chip and memory crisis. A server’s memory component that cost around $500 last October has jumped to roughly $2000 by February and prices keep climbing. This is a ticking budget bomb.
Businesses are no longer waiting around for expensive new supply. Instead, they’re shifting to smarter management of the resources they already have. By deploying automation and orchestration tools, organizations can push existing computing utilization from around 40% to over 70%. That is the key to stopping the budget bleed and avoiding unnecessary hardware purchases at crisis-level prices, while directly improving the organization’s net profit.
From Defense to Rapid Recovery
For years, security professionals tried to answer the question: how can a breach be prevented? Today, after seeing that damage to a data center, whether caused by a cyber attack or a physical one, can paralyze critical services for days on end, the question has changed to: how fast can we recover and get back to work after a hit? The critical new management metric is clean recovery time.
Traditional backup capabilities are no longer enough. Organizations are adopting cyber vault solutions and clean room environments that guarantee restored data is completely safe and free of backdoors. The technological breakthrough of using high-speed storage for backup as well is the breakthrough that makes it possible to cut recovery time from days down to a few hours. Infrastructure professionals who build rapid-recovery systems and establish return-to-operations time as their central metric will give their organization a resilience that extends well beyond firewall defenses.
Looking ahead, 2026 is shaping up to be a test of managerial maturity. In an era of soaring hardware prices and AI agents becoming the beating heart of the organization, infrastructure is no longer only "a topic for the technologists", but is becoming a full-blown business compass. This is the moment when architecture becomes strategy and proper preparation today is the difference between an organization that grows and one which gets left behind.
Oren Porat is the CEO of the Infrastructure Division at peax (formerly EMET).
