Cloud complexity gap grows
Analysts warn that cloud estates are becoming harder to govern than to buy, with security struggling to keep pace as architectures scale and diversify. The piece argues enterprises need unified defence patterns as AI and GPU‑heavy workloads add new control and observability requirements. (techradar.com)
Cloud security is turning into a governance problem: companies can add cloud services faster than they can see, secure, and control them. (cybersecurity-insiders.com) A February 3, 2026 survey of 1,163 senior cybersecurity leaders found 69% cite tool sprawl and visibility gaps as the top barrier to effective cloud security. The same report said 74% face an active shortage of qualified cybersecurity professionals. (cybersecurity-insiders.com) The report also found 59% still rate their cloud security maturity at the initial or developing stage, even as 62% expect cloud security budgets to rise in the next 12 months. Anthony Leigh’s April 15, 2026 TechRadar Pro column argues the gap is widening even as spending increases. (cybersecurity-insiders.com) (tech.yahoo.com) Cloud computing is rented computing power and software delivered over networks instead of run in one company data center. That model now usually spans multiple public clouds, on-premises systems, software-as-a-service apps, and remote users, which the February 2026 report describes as the new normal. (cybersecurity-insiders.com) A separate September 9, 2025 survey from Cloud Security Alliance and Tenable found 63% of organizations use more than one cloud provider and 82% run some form of hybrid infrastructure. That same survey said 59% ranked insecure identities and risky permissions as the top cloud risk. (cloudsecurityalliance.org) The security problem is not just more servers. Each cloud service, identity system, application, and data store can produce its own logs, alerts, and policies, leaving teams to piece together risk from separate dashboards instead of one shared view. (tech.yahoo.com) (cybersecurity-insiders.com) Artificial intelligence adds another layer because the workloads are heavy, fast, and expensive: they rely on graphics processing units, specialized chips built to handle many calculations at once. NVIDIA’s enterprise AI factory design guide says those environments are built for horizontal scaling across Blackwell graphics processing units, networking, and Kubernetes software, which increases the number of moving parts operators have to monitor. (docs.nvidia.com) The Cloud Security Alliance survey found more than half of organizations are deploying artificial intelligence workloads and 34% already report an artificial-intelligence-related breach. Only 20% said they prioritize unified risk assessment, and 13% said tool consolidation is a priority. (cloudsecurityalliance.org) Big vendors are responding by tying security and observability closer to artificial intelligence infrastructure. Microsoft said on March 16, 2026 that its Azure AI infrastructure is being optimized for inference-heavy reasoning workloads, while Cisco said the same month that its Secure AI Factory with NVIDIA is meant to keep a consistent security posture from data center to edge. (blogs.microsoft.com) (blogs.cisco.com) HashiCorp’s 2025 Cloud Complexity Report frames the same shift as an operating-model problem, calling for developer self-service with built-in guardrails, automated workflows, policy-driven governance, and real-time visibility. The common thread across these reports is that buying more tools does not remove complexity if the tools do not share context. (hashicorp.com) (cybersecurity-insiders.com) The immediate question for enterprises is not whether they will run more cloud and artificial intelligence workloads. It is whether their controls, logs, identities, and response systems can be made to operate as one system before attackers exploit the gaps between them. (cybersecurity-insiders.com) (tech.yahoo.com)
