Peak Weekend, Normal Operations: What 115.8 Billion Requests Revealed

Black Friday through Cyber Monday 2025 delivered a familiar load pattern for Vercel's platform: traffic dipped on Thanksgiving, surged on Black Friday, and built into a second wave on Cyber Monday. Across all geographies, the system adapted continuously without configuration changes or manual intervention.

Over the four-day window from November 28 through December 1, the platform served 115,836,126,847 total requests, a 33.6% year-over-year increase. The real-time operational dashboard showed 518,027 peak requests per second at high demand. Teams shipped 6,120,247 deployments during the event, with new application versions rolling out continuously.

Core Metrics From the Event Window

  • 115.8 billion total requests delivered with consistent performance
  • 518,027 peak requests per second at maximum demand
  • 6.1 million deployments of new application versions
  • 24,086,391 AI Gateway requests routed across providers
  • 43,213,555,901 Fluid compute invocations handled dynamically
  • 56,926,096,915 cache hits served from edge regions
  • 1,809,912,897 ISR reads for fresh content on initial regional loads
  • 1,517,476,504 ISR writes for instant catalog and pricing updates
  • 7,507,223,309 firewall actions filtering threats at the edge
  • 415,683,895 bots blocked to stop automated abuse
  • 2,408,122,336 humans verified through invisible security checks
  • Top regions: US, DE, GB, IN, BR, SG, JP

Traffic Rhythms Across the Globe

The United States led all regions with over 40.7 billion requests, followed by Germany, the United Kingdom, India, Brazil, Singapore, and Japan. Activity shifted across time zones throughout the weekend, with peaks in one region offset by lower traffic elsewhere. The network scaled its global compute layer to match real user behavior across all 20 regions.

AI Gateway Handled AI-Native Shopping

AI continued to shape how shoppers found products and got assistance. More than 24 million AI requests passed through AI Gateway, supporting search, recommendations, guided browsing, and customer support. The gateway routed queries across providers and regions, maintaining low latency with automatic failover during demand fluctuations.

Fluid Compute Scaled With Demand

Fluid compute processed more than 43.2 billion function invocations for personalization, cart behavior, content evaluation, and AI inference. Capacity adjusted instantly to incoming traffic with automated pre-warming, scaling up the moment volume rose and scaling back when demand settled. No tuning or configuration was required.

ISR Kept Catalogs Fresh

Product catalogs, pricing, inventory indicators, and promotions changed continuously throughout the event. Incremental Static Regeneration handled more than 1.8 billion reads and 1.5 billion writes, refreshing content across the CDN without redeploys or backend strain. Shoppers received accurate information at static performance speeds even as merchandisers updated content minute by minute.

Security Filtered Billions of Requests

Security systems scaled alongside customer traffic. More than 7.5 billion firewall actions came from system rules and customer-defined WAF logic, stopping invalid and malicious traffic at the edge and preserving compute capacity. Vercel BotID and system defenses blocked more than 415 million bot attempts while verifying over 2.4 billion legitimate human interactions through invisible checks, protecting checkout, inventory endpoints, and account systems without adding friction.

Caching Served the Majority of Requests

More than 56.9 billion requests were served from Vercel's global cache. These cache hits reduced latency and backend load. Caching worked alongside ISR and Fluid compute so static assets stayed fast, dynamic updates propagated instantly, and compute was reserved for real application logic.

Adaptive Infrastructure for Real User Behavior

The 2025 traffic profile demonstrates that large events require no special preparation when the delivery layer adapts naturally. Cache hits, ISR updates, Fluid compute, AI routing, and firewall filtering absorbed the full load. Teams relied on continuous deployments, instant rollbacks, and predictable reliability to keep shipping throughout the weekend.

As AI-driven experiences and personalization expand, infrastructure must adapt at the speed of real users. The platform's software layer handles this daily, and Black Friday through Cyber Monday simply highlights the scale of normal operations.