AI goals outpace IT infrastructure, study warns
Enterprise AI plans are moving faster than the infrastructure meant to support them. In new research from Netskope, only 38% of infrastructure and operations leaders said their current stack can handle AI demands. Just 18% felt completely confident that their team and budget can meet expectations for performance, resilience, and security.
The study compared executive ambition with what IT teams believe they can deliver using current platforms and operating models. The gap is widening.
Rising expectations, competing priorities
Four in five respondents said infrastructure is central to delivering core business goals, and the same proportion reported higher expectations from senior leaders in the past year. At a personal level, 83% said the pressure on them has increased-driven by AI discussions, ongoing security requirements, hybrid work, and uptime for critical systems.
Despite that pressure, AI-related work isn't always the top priority. The leading focus remains hardening and accelerating remote access (43%), followed by improving network visibility and performance (35%). Supporting organisational AI adoption sits slightly lower at 34%.
Strategy gaps and communication friction
Many I&O leaders feel disconnected from decision-making. Nearly two-thirds (63%) said they are far removed from strategic conversations that shape IT direction. One in five lack a clear view of CEO/CIO objectives, and 37% described their function as reactive.
Communication is also a pain point: 61% said CEOs get frustrated when infrastructure isn't transparent or easy to understand.
KPI tensions
Respondents believe executives focus most on security, visibility, and cost-while paying less day-to-day attention to resilience and performance. At the same time, most leaders view C-suite expectations as unrealistic given current systems: 55% for performance, 58% for resilience, and 59% for security.
Complicating matters, leaders said they have the least ability to influence outcomes in security and performance under current structures.
Legacy mindset meets AI demand
Investment attitudes are cautious. Sixty percent said their organisation sticks to an "if it isn't broken" approach to infrastructure spending. Yet 65% believe infrastructure is more important to running the business today than ever.
Mike Anderson, Chief Digital and Information Officer at Netskope, summed it up: "AI has increased demand on enterprise infrastructure at a pace that legacy systems were never built to support... The way forward begins with translating infrastructure decisions into business terms so leadership can see how modernisation reduces risk, improves agility, and prepares the organisation for safe and effective AI adoption."
What operations leaders can do now
- Translate tech to business outcomes: frame upgrades in terms of risk reduction, uptime, customer impact, and speed to market. Tie each project to clear financial and operational KPIs.
- Set shared guardrails: agree on SLOs for performance, resilience, and security with the C-suite. Make trade-offs explicit and measured.
- Prioritise access and observability: modernise remote access and zero trust controls, and invest in end-to-end telemetry so issues are visible before they hit users.
- Plan for AI workloads: model GPU/CPU capacity, storage, and network egress. Identify which AI workloads belong on-prem, in cloud, or via managed services.
- Simplify the stack: consolidate overlapping tools and vendors to cut cost and failure points, and to improve transparency for executives.
- Get upstream: join strategy cycles earlier with scenario plans, options, and cost/performance impacts so decisions aren't made without infrastructure input.
- Report like finance: publish concise monthly readouts that show readiness, risk, and ROI in plain language. Include trendlines and a 90-day forecast.
- Secure the data path: apply least privilege, data loss controls, and isolation around AI data and model endpoints. Treat model access like any other sensitive service.
- Fund quick wins: ring-fence a small budget for proofs of concept that remove bottlenecks fast (e.g., ZTNA rollout, network visibility gaps, AI workload placement tests).
Survey recommendations
- Translate infrastructure decisions into business outcomes, not technical terminology.
- Engage earlier in strategic planning.
- Advocate for architectural simplicity and consolidation.
- Provide ongoing reporting that gives leaders clear visibility into the IT estate.
- Position I&O around safe, fast AI adoption to reduce executive anxiety and build trust.
Helpful resources
For guidance on risk and governance, see the NIST AI Risk Management Framework. For access security patterns that support remote and AI use cases, review the CISA Zero Trust Maturity Model.
Level up skills
If your team needs practical upskilling aligned to job roles, explore curated learning paths: AI Learning Path for CIOs, AI Learning Path for VPs of Strategy, and AI Learning Path for Network Engineers.
Your membership also unlocks:
