Cloudflare posts 39% revenue growth, cites OpenAI token savings

By James Calloway

Cloudflare generated $782 million in third-quarter 2026 revenue, up 39% year on year, as annual revenue crossed $3 billion during the period. The network and cloud services vendor described the result as its best quarterly performance to date.

What did Cloudflare win in Q3 2026?

Cloudflare signed up 158,000 new customers during the quarter. Two contracts disclosed with the results:

• U.S. cabinet-level federal department: A five-year, $10.3 million contract for Cloudflare’s secure access service edge (SASE) portfolio.
• Fortune 100 cloud infrastructure provider: A 13-month, $15.5 million renewal for Cloudflare’s cloud-based distributed denial-of-service (DDoS) protection services.

Customers paying more than $100,000 per year grew 25% to 5,000 by quarter end. The disclosed contract values concern SASE and DDoS protection, not the OpenAI collaboration.

How does the OpenAI tie-up reduce token costs?

OpenAI, the frontier AI developer, worked with Cloudflare to add previously unknown webpages to its search index and improve ChatGPT output without repeating web crawls at scale. The collaboration addresses a specific expense, because large-scale recrawling consumes tokens.

Cloudflare can signal when content changes, such as when a menu price updates, allowing OpenAI to obtain fresher data with fewer crawl operations. That is the technical mechanism behind management’s claim of lower token expenditure.

On the earnings call, Matthew Prince, Cloudflare’s chief executive, said: “Cloudflare could actually give them hints when a page had changed, when a menu price had updated, when anything had happened, and then they could get the freshest data, and they could actually crawl that.”

He added: “That helps OpenAI save money, and that actually helps the Cloudflare customers get the freshest data to them and make sure that it does not waste a ton of resources.”

Prince positioned Cloudflare as a mediator between hyperscalers and AI laboratories scraping sites to expand their services and the developers, financial services organizations, e-commerce companies and creators subjected to those scrapes. The reported results give no contract value or term for the OpenAI arrangement. Token savings remain management’s explanation of the use case, not a disclosed revenue line.

What does this change for AI infrastructure?

Vendors across the cloud and AI infrastructure market are increasingly building their own AI infrastructure to cut token costs as prices climb. The OpenAI collaboration fits that pattern, although the disclosed results quantify Cloudflare’s own revenue rather than OpenAI’s savings.

Cloudflare’s distributed systems give it a role on both sides of the exchange: supplying freshness signals to a large AI developer while giving site customers fresher data with less wasted resource use. The arrangement extends the company beyond the DDoS protection and SASE businesses covered by the two disclosed contracts.

Could slower data-center construction change the outlook?

Speaking after this week’s Democratic electoral victories, Prince rejected the idea that slower data-center construction after the midterms would hurt Cloudflare. His case rests on efficiency, global distribution and density rather than construction volume alone.

“I think as the market gets more rational ... what will happen is people will focus instead of where can we get more value (and efficiency),” Prince said. “It turns out that we are the platform that has built that. (It) allowed us to distribute our systems globally.”

He added: “What we are increasingly finding is in those places where we need more density, where agents are talking to agents and they might be in two data centers right next to each other, that density also benefits from that same isolated architecture.”

His remarks set out a management view, not a spending forecast. Cloudflare gave no timetable for additional infrastructure investment or data-center deployment in the results.

Over the coming quarters, the practical tests are whether the OpenAI collaboration acquires disclosed commercial terms, whether the $100,000-plus customer tier keeps growing, and whether post-midterm construction shifts demand toward denser deployments.