Customers Are Reprioritizing Budgets for AI, Security, and Quantum Readiness
Recent insights from Cisco reveal that hyperscalers and enterprises are significantly ramping up their network upgrades to accommodate the growing demands of agentic artificial intelligence workloads. This shift is primarily aimed at managing token-related costs and addressing increasing security threats in an ever-evolving digital landscape.
CEO Chuck Robbins characterized this development as a “multi-year, multi-billion-dollar networking supercycle.” This supercycle has spurred customers to invest substantially in routers, switches, optics, wireless technologies, and industrial IoT hardware tailored for use across data centers, campuses, and service provider networks.
With the expansion of AI capabilities, companies require efficient inter-data center communications. This need has led organizations to bring agentic AI inferencing on-premises, striking a balance between curtailing token expenses while simultaneously modernizing their network infrastructures to counteract the growing cyber risks associated with scaling operations.
Robbins emphasized the ongoing trend he has observed among customers: “They are currently reprioritizing within their existing budgets.” This shift appears to parallel past trends where organizations closely monitored cybersecurity expenditures. Now, preparations for AI readiness, Mythos readiness, and quantum readiness are gaining equivalently critical importance.
Many corporations are grappling with a tripartite challenge encompassing cost, security, and data sovereignty. They are seeking more intelligent deployment strategies for their AI initiatives—this may involve shifts in model types or operational locations. Robbins explained that the conversation around cost efficiency began with tokenomics and has now broadened to include open-weight models and foundational or frontier models.
Robbins anticipates that organizations will strategically choose models based on specific use cases and their appropriateness for various agentic applications. Currently, companies have two predominant options for AI model deployment: cloud-based models, wherein companies send prompts to an AI provider and incur usage fees; or on-premises models that allow firms to host the model weights on their GPUs, retaining local data governance. The latter has gained traction as an increasingly preferred method for cutting costs while maintaining effective AI deployment.
Robbins stated, “If they continue to use cloud-based models, that’s good for us, just like it has been for the last two years. If they move to open-weight models or models that they’re running on-prem, that’s great for us because it means they will invest in more enterprise private data center networking.” This dynamic has been positively reflected in Cisco’s financial performance in recent quarters.
Opting for an on-premises approach necessitates substantial investments in GPU clusters, both locally and at the network edge. This entails ensuring low-latency, high-bandwidth networking while incorporating built-in security measures, observability, and automation. Regardless of the chosen architecture, ensuring the security of countless agents embedded in the infrastructure remains paramount, as Robbins noted: “How are enterprises going to navigate the future?”
Complexities in building new AI models and workflows also underpin the urgent need for networking upgrades. Robbins pointed out that as AI models become more intricate and sizeable, hyperscalers face evolving challenges that require interconnections across multiple data centers due to limitations in physical space and power availability within individual centers.
In an impressive display of demand, service providers and cloud companies increased their Cisco orders nearly twofold from the previous year, and four major hyperscalers noted triple-digit percentage growth in AI infrastructure purchases. Cisco reported a substantial $9.3 billion influx in orders from hyperscalers focused on AI infrastructures over the past year.
Furthermore, over half of Cisco’s customers are concurrently modernizing their workplaces alongside upgrading their data centers. Robbins shared an illustrative case of a frontier AI company that invested in Cisco’s Wi-Fi solutions, smart switches, and comprehensive end-to-end segmentation solutions to enhance both speed and security across multiple office locations.
Robbins remarked, “We see the momentum in campus networking being driven by infrastructure modernization to both scale AI initiatives and to strengthen defenses against a rapidly evolving cyber landscape.” This need for modernization has become critical as businesses are becoming increasingly aware of potential security challenges.
Despite Cisco’s robust financial results, including surpassing sales and earnings estimates, the company’s stock experienced a decline post-announcement. Specifically, following a period of strong earnings, Cisco’s stock dipped by $5.04, or 4.07%, reaching $118.84 in after-hours trading, marking its lowest trading position since early August. As the company looks ahead, Cisco anticipates a fiscal quarter ending October 24, projecting non-GAAP net income per share in the range of $1.32 to $1.34, with revenues expected between $18 billion and $18.2 billion.
This performance stands in stark contrast to analyst predictions who had estimated earnings of $1.14 per share and revenues of about $16.66 billion. The trajectory of Cisco highlights the importance of aligning fiscal strategies with current technological advancements and market demands.