The rapid evolution of artificial intelligence (AI) is prompting a reevaluation of cloud computing strategies among enterprises, according to a recent report from Deloitte. While the cloud has dominated discussions about computing infrastructure over the past decade, businesses are now considering a shift towards a hybrid model that combines both cloud and on-premises solutions to better meet the demands of AI.
As organizations increasingly utilize AI, many existing infrastructures built primarily for cloud-first strategies may struggle to support the heightened demands of AI workloads. Deloitte analysts, led by Nicholas Merizzi, assert that standard cloud infrastructures are not optimized for the economic realities associated with AI deployment. “The infrastructure built for cloud-first strategies can’t handle AI economics,” the report states, highlighting the inadequacy of processes designed for human workers versus those required for machine-learning agents.
The report outlines a significant movement away from an exclusive reliance on cloud computing. Enterprises are now exploring how a hybrid approach can leverage the strengths of both environments. As described by Deloitte, the shift is characterized as moving “from cloud-first to strategic hybrid”—utilizing cloud for its elasticity, on-premises systems for their consistency, and edge computing for immediacy in decision-making.
Four critical issues have emerged that underscore the challenges of cloud-based AI solutions. First, enterprises are facing rising and often unanticipated cloud costs. Despite a significant reduction in AI token prices—dropping 280-fold in two years—many organizations report monthly bills soaring into the tens of millions due to high usage of cloud-based AI services. Deloitte suggests that at a certain point, typically when cloud costs exceed 60% to 70% of total expenses for comparable on-premises systems, capital investment may become more appealing than ongoing operational costs.
Secondly, latency issues with cloud computing pose significant challenges for AI applications that require rapid response times. As highlighted in the analysis, applications demanding response times of 10 milliseconds or lower cannot afford the delays inherent in cloud processing. This latency can hinder mission-critical applications, further emphasizing the need for on-premises infrastructure.
Furthermore, resiliency is increasingly recognized as a necessity for operational continuity. Deloitte’s report notes that mission-critical tasks unable to tolerate interruptions necessitate an on-premises setup, especially in scenarios where cloud connectivity may falter. Data sovereignty presents yet another pressing issue; some enterprises are repatriating their computing services to avoid dependency on external service providers operating outside their local jurisdictions.
In light of these challenges, the report advocates for a three-tiered approach to computing infrastructure. This model includes using the cloud for its elasticity to accommodate variable workloads, on-premises systems for consistent performance and cost predictability, and edge computing for tasks requiring immediate action. This hybrid strategy is being increasingly embraced within the industry.
Milankumar Rana, formerly a software architect at FedEx Services, supports this dual approach, emphasizing the advantages of cloud services while recognizing the necessity for some on-premises capabilities. “I have built large-scale machine learning and analytics infrastructures, and I have observed that almost all functionalities can now run in the cloud,” he remarked. However, he also advises maintaining certain workloads on-premises, particularly where data sovereignty or low latency is a concern.
The focus on a hybrid strategy reflects a broader trend within the industry as businesses navigate the complexities of integrating AI into their operations. Security and compliance remain paramount regardless of the chosen infrastructure. Rana emphasizes that while cloud platforms provide robust security features, companies must actively ensure adherence to regulations related to encryption, access, and monitoring to mitigate risks.
As organizations continue to evolve in their technological strategies, the dialogue around cloud versus on-premises computing is likely to remain dynamic. The emergence of AI as a critical driver of operational efficiency will shape how enterprises invest in their infrastructures moving forward, underscoring the need for flexible solutions that can adapt to changing demands.
See also
DeepSeek Launches Groundbreaking LLM, Surpassing OpenAI’s o1 in Reasoning Capabilities
94% of Engineering Leaders Face Agentic AI Skills Gaps as Demand Surges in 2025
Apple Reaches $4 Trillion with Privacy-First AI, Launches Apple Intelligence Ecosystem
Swansea Man Paralysed by Wave Walks Again with AI-Enhanced Rehabilitation Technology
Amazon Enhances AI Workflows with GenAI Learning Systems for Global Enterprises
