Arista and Cisco Advance AI Fabric Monitoring
Arista is developing new telemetry tools to provide a unified view of network and host health for AI fabrics, while Cisco has launched a new catalog of AI repositories. These advancements reflect a growing customer requirement for robust, real-time observability in large-scale AI deployments, making such capabilities a key part of the sales qualification process for hardware vendors.
- Arista's forthcoming telemetry extensions aim to provide a unified view of both network and host health by pulling data from in-network sources like RDMA counters and flow control, as well as host-level information such as the RDMA stack and NIC buffering. This unified view is a direct response to customer demand for faster diagnosis of complex performance issues in AI clusters. - The key to Arista's AI telemetry is the ability to unify host and network data into a single, correlated view, which is critical for AI clusters where minor network issues can halt synchronized GPU jobs and waste significant compute resources. This capability is becoming a crucial selling point as hyperscalers look to reduce their dependence on single-vendor solutions. - Cisco is positioning its AI-ready infrastructure as a complete system, with CEO Chuck Robbins reporting a significant acceleration in AI infrastructure orders, expecting them to exceed $5 billion in fiscal year 2026. This strategy focuses on the total cost of ownership at a token level rather than the cost per switch port or GPU. - For sales operations in the semiconductor industry, a key challenge is that sales teams spend only about 26% of their time on customer-facing sales activities. To address this, sales and operations planning (SOP) processes are critical, typically covering an 18-month horizon and integrating marketing plans with supply chain management. - In high-ACV hardware sales with long cycles, tracking metrics like Annual Contract Value (ACV), average sales cycle length, and win rate is essential for performance tracking. A longer sales cycle directly impacts profitability, making it a critical KPI for sales operations to monitor and optimize. - For forecasting in complex, multi-stakeholder deals common in the semiconductor industry, multivariable forecasting models are effective. These models incorporate not only pipeline data but also external market conditions, seasonality, and competitive actions to predict demand. - AI-powered lead qualification is becoming crucial for hardware sales, with the ability to reduce the time spent on qualifying a prospect from hours to minutes and significantly increase conversion rates. This is particularly important in B2B sales cycles that can last 2-6 months and involve multiple decision-makers. - CRM automation is a key best practice for technical sales, with 71% of reps stating they spend too much time on manual data entry. Effective CRM implementation for hardware companies involves integrating data from various sources, training the team on the system, and using data validation to ensure accuracy.
