New York, Feb 23, 2026, 05:27 EST — Premarket
- Cloudflare edged up 0.2% before the bell, recovering a bit after Friday’s steep fall.
- Cloudflare traced the Feb. 20 outage affecting its BYOIP service back to an internal change, ruling out any cyberattack as the cause.
- Traders want to see if management delivers on reliability fixes, plus any fresh commentary at the next conferences could move things.
Cloudflare shares ticked 0.2% higher ahead of the open Monday. On Friday, they tumbled 8.1% to finish at $176.99. After hours, the stock was quoted at $177.37.
The decision underscores reliability at a company that handles a significant chunk of internet traffic. If a network provider falters, customers notice quickly, and investors typically bake that risk into prices ahead of the next session.
Cloudflare’s full breakdown of the incident dropped over the weekend, finally giving traders something more tangible than social feeds to work with. The company also pushed back against the usual cybersecurity worry that any outage could signal a real attack.
Cloudflare on Saturday blamed a change in handling customer IP addresses for the outage that knocked out about 1,100 “Bring Your Own IP” (BYOIP) prefixes from the web’s routing system, taking traffic offline for 6 hours and 7 minutes. The removal wasn’t linked to any cyberattack, the company said in its post-mortem, calling the disruption their mistake: “We let you down today.” Cloudflare now says it’s tightening up API checks, reworking rollback systems, and setting up new monitoring as a circuit breaker in case withdrawals get out of hand. The Cloudflare Blog
Cloudflare’s status page indicated the company advised customers to self-mitigate by re-advertising prefixes via its dashboard. However, Cloudflare also warned that not all customers could reload IPs using that method and said a fix was in progress. Separately, the same page tracked an Analytics Engine API issue on Sunday, which was eventually marked as resolved.
Laravel Cloud reported losing connectivity for roughly 3 hours and 15 minutes, pointing to a Cloudflare issue that pulled back IP prefix ads and disrupted traffic to its platform. The company assured that customer data stayed safe and intact, calling the incident “entirely network-level.” Laravel also acknowledged it hasn’t set up redundancy for its “IP announcement layer” yet. Laravel
Customers turn to BYOIP to retain control over their public internet addresses, even as their traffic routes through Cloudflare. The system hinges on BGP, or Border Gateway Protocol—the framework networks rely on to sort out traffic destinations. If a prefix drops out of BGP, the packets have no route, and traffic simply stalls.
Cloudflare touts itself as the security and performance backbone for businesses determined to keep their apps up and safe from attacks. That story works in a market willing to pay for infrastructure linked to AI workloads. But there’s not much tolerance for stumbles in operations.
Cloudflare set the bar high earlier this month, projecting annual revenue ahead of what analysts were expecting. CEO Matthew Prince called the rise of AI agents “a fundamental re-platforming of the internet.” That optimism was all about expansion. Fast-forward to this week: the spotlight’s back on reliability. Reuters
The risk is clear enough: Should major customers push for greater redundancy or ramp up splitting their traffic among rival providers, the effects of this incident might show up in retention and net expansion metrics well after the routers are back to normal.
Cloudflare goes up against Akamai and Fastly in edge delivery and security—a space where customers typically use multiple providers. Reliability is key here, but it’s also a vulnerability if service falters.
Cloudflare execs could face new questions about incident controls and customer fallout when they speak at the Baird Silicon Slopes Conference on Feb. 26, then again at Morgan Stanley’s Technology, Media & Telecom event on March 3. Details here:
