Meanwhile on the Internet… looks at what Fastly’s network sees when the world does something: a shopping peak, a kickoff, a major game release, and whatever else the internet gets up to. Each post tells one story, backed by our traffic data.
Election night is one of the few times a big chunk of the news audience is pointed at the same thing at the same time and there’s a very short list of events that can do that. On November 3, we watched it happen: viewers streaming the broadcast, readers hammering election trackers (often both at once), and local news fans.
Outside of the policy implications (analysis of which we’d refer you to one of Fastly’s many publishing and media customers), the activity surrounding key election moments were big enough to move the internet in substantial and interesting ways.
The numbers got big. Publishers’ bandwidth reached over 14x the rate of a typical Tuesday night, and for digital publishers, humans beat the bots as the primary driver for traffic for six of eight hours. Enjoy it while it lasts.
Traffic Climbed as the First East Coast Results Came In
In a select cohort of international publishing and broadcast services, broadcast request rates were elevated throughout the afternoon and evening, and climbed further around 7:00pm EST, as the initial East Coast results were announced, which drove a significant spike for publishers. We suspect users were rapidly refreshing election trackers as announcements broke.
Broadcasters saw the largest increases in requests, ultimately reaching approximately 12x the request rate. Publishers saw the largest bandwidth spikes, peaking at over 14x the bandwidth rate compared with a typical Tuesday night. That fits a second-screen habit that has been documented for years: one device for the broadcast, another for publisher coverage. A single screen (apparently) is never enough.
Traffic in Key States
Several key states (Georgia, Ohio, Michigan, and Texas) also have Fastly points of presence (POPs), so we checked whether their traffic stood out. Compared with the fleet baseline, each sampled data center saw more traffic at its busiest half-hour period. Houston saw the largest increase, peaking relative to a typical Tuesday between 8:00 and 8:30 p.m. EST, around the time the first Texas race was projected.
Humans Outshining Bots, but Only for a Moment
In six of the eight hours between 6pm and 2am EST, humans outnumbered requests flagged as automated in Fastly’s digital media publishing segment. As we shared in June 2026, bots have equaled human traffic volumes, and bot traffic is growing faster than human traffic. Over those eight hours, human requests ran about 2.3x a typical Tuesday in the publishing segment.
Local News Had a Big Night Too
In this analysis we split our publishing services into national outlets and regional or local ones. Across the board, all of the reviewed cohorts of publishers saw major traffic increases and regional media outlets saw over a 4x increase over a typical Tuesday’s bandwidth. We assume people turn to regional outlets for local matters like municipal elections. That journalism matters, and we're honored to help make it succeed on nights like this.
One Night, Two Screens, and a Lot of Humans
The traffic looked like a country watching on two screens, local news held its own, and for digital publishers, humans outnumbered the automated traffic. (Sleep, on the other hand, was not measured.)
People call the press the fourth estate of the government so often it's a cliché, but nights like this show why the phrase stuck. Hundreds of international, regional, local, and independent properties rely on Fastly to stay online and keep readers current. From storied institutions to ambitious upstarts, and millions of viewers rely on the live broadcasts and streams we deliver while news is breaking.
If you want to know how we helped our customers through election night, and how we can do the same for you, reach out to our team of experts.
Methodology
The data analyzed in this post covers Fastly’s network on Tuesday, November 3, 2026. Traffic for publishers, broadcasters, and national and regional outlets is based on fixed groups of selected customers and is not a census or limited to election-related content. Request rates and bandwidth are shown as multiples of a typical Tuesday night, based on the average of the same time of day on prior Tuesdays, and peaks are the highest five-minute interval. The data-center view includes all customers at each location, not only media customers. Race-call times reflect projections reported by a major news broadcaster, are shown for context, and do not establish cause.
Human and automated traffic was analyzed for a digital media publishing segment with selected exclusions, and automated traffic was classified using Fastly’s edge detection signals. Each is indexed to its own normal Tuesday level, so the chart shows change relative to baseline, not the volume of each. All figures reflect requests and bandwidth delivered over Fastly’s network, not unique readers or viewers.
