Opinion
Smart efficiency starts with infrastructure, not layoffs
Before making painful decisions about employees, it's worth checking whether some of the savings are hiding in the organization's cloud, data, and AI infrastructure.
Over the past year, layoffs have stopped being just the default response in efficiency plans - they've become a recurring event that shows up almost every quarter. Major global tech companies have announced rounds of cuts, often framing them as necessary for efficiency, profitability, or adapting the organization to the AI era.
AI is genuinely reshaping workflows, roles, and org charts. But in some cases, it has also become a convenient explanation for cuts whose underlying drivers are over-hiring, declining revenue, high costs, or pressure to show profitability.
Israel's tech sector isn't immune either. The Innovation Authority's 2026 State of High-Tech report paints a mixed picture. On one hand, 2025 was a strong year for fundraising and a record year for exits, while industry output grew significantly. On the other, for the first time in a decade, the number of R&D employees in Israel actually declined. At the same time, rising labor costs, a stronger shekel, a more concentrated funding market, and the pull to move operations abroad have made the efficiency question more urgent for many leadership teams.
That pressure is real, and so is the need to become more efficient. Still, layoffs shouldn't be the first move in an efficiency plan. Efficiency isn't just about cutting spending - it's the ability to produce the same output, sometimes more, at lower cost. Employees aren't just a line item. They're a source of knowledge, experience, innovation, and continuity.
Before cutting experienced teams, it makes sense to examine the organization's own cost structure and look for other pockets of waste. In an era where cloud, data, and AI have become core to how organizations operate, technology infrastructure is one of the first places worth examining.
Cloud services, data systems, SaaS tools, storage infrastructure, monitoring systems, dev tools, and AI services have become a major and fast-growing expense category in many technology-driven organizations. Headcount is scrutinized almost every quarter, while much of the technology architecture remains unchanged for years.
The problem rarely starts with a single decision - it builds up over time: a cloud service added during a growth spurt, a data system built to solve one specific problem, a caching layer bolted on to handle load, an AI tool that started as a pilot and never left, a software license bought for one team and never revisited. Each of these decisions can be perfectly reasonable in the moment, but a few years later they can add up to something expensive, complex, and under-monitored.
According to Flexera's 2026 report, after five consecutive years of decline, estimated waste in IaaS and PaaS cloud spending climbed back to 29%. The report attributes this partly to growing AI workloads, increasingly complex pricing models, and a proliferation of new cloud services that are hard to keep track of.
A real infrastructure audit needs to start with some basic questions: Are all the servers the organization is paying for actually being used? Is capacity purchased in the past still needed? Do systems built a decade ago fit today's workloads? Are there redundant layers of storage, processing, and data access? Are new AI tools being deployed without central oversight of cost and value? Is there too much dependence on a single cloud provider, weakening the organization's negotiating position?
This kind of review matters even more once an organization starts moving AI applications from pilot to production at scale. A small pilot can look simple and cheap, but once that same use case starts serving thousands of employees, customers, or business processes, it touches cloud, data, security, availability, and real-time decision-making all at once. At that point, the real question isn't just what the AI tool itself costs - it's what it costs to run the entire environment that lets it function.
For companies whose operations depend on real-time data, the impact is even bigger. A card payment, a fraud check, an ad impression, a product recommendation, customer service, or an AI agent's action all look to the user like one simple event. Behind each one, though, sits a chain of checks, calls, and processing steps. If the system is too slow, the decision arrives too late. If it's too expensive, the business model takes the hit. If it isn't stable enough, user experience and trust in the product suffer.
That's why technology infrastructure isn't just an IT or engineering concern - it's part of business strategy. A proper audit needs to look at whether compute resources match actual needs, at the architecture itself, at how data is managed, at the system's ability to handle load, and at the relationship between technology spend and the business value it actually generates.
Good infrastructure is a precondition for growth, so the objective is to spend better: invest where it creates value, stop paying for systems that no longer serve a real need, and redesign areas where costs have grown without enough oversight.
Smart efficiency requires leadership to examine every major line of spending, challenge old assumptions, and revisit systems that have become embedded in the operating model. Only after that audit, if the available savings still do not close the gap, should more painful steps be considered.
A company that reduces infrastructure waste can improve performance, lower risk, and retain institutional knowledge at the same time. Before cutting the people who hold up the business, leadership should make sure it has examined the systems they have been asked to run.
Oshrat Ben-Avi Zabludovitz is Israel Country Manager at Aerospike.
