Cloudflare just gave the search industry a number that changes the conversation: AI agent traffic is up 1,700% year over year. At the end of 2024, Cloudflare handled an average of 63 million HTTP requests per second. Now that figure has almost doubled to 115 million, and for the first time more than half of internet traffic is not human.
Google’s public battle has always been framed as a fight against rank trackers. But the real stakes may be larger: networks of LLMs that could reverse engineer Google’s search results, and AI Mode queries that are dramatically more expensive to serve.
The new traffic wave is not made of humans
Cloudflare’s data points to a step-change. Daily requests from AI agents grew by more than 1,700% in the past year. Some of that traffic is hybrid, with agents working on behalf of humans, which means a portion still has value to websites.
But the volume is the issue. Automated queries add load, create latency, and inflate keyword inventory in ways that can distort the click and traffic signals Google uses internally.
AI Search changes the cost per query
Traditional search serves pre-indexed pages matched to queries. AI Mode is different: one user question can trigger multiple associated queries and reasoning steps before Google synthesizes an answer. That makes every AI response costlier to generate.
Sundar Pichai put it this way on Google’s Q2 earnings call:
“As we serve more of these queries, we’ve continued to drive efficiencies. Thanks to our engineering and hardware optimizations, this quarter we reduced the cost of AI Mode responses to its lowest level since launch, even as we’ve brought more advanced AI capabilities.”
If AI agent traffic keeps climbing, those efficiency gains become harder to protect.
Rank trackers are getting caught in the crossfire
Ryan Jones, developer of SERPrecon, sees the practical impact.
“That’s why the rank trackers are failing. Google is blocking all the AI trackers AND AI itself… and the rank trackers are getting caught up.”
Jones has warned that we are reaching a point where nobody has reliable rank tracking anymore.
The underlying concern is distillation: using ranking outputs to train models that reproduce Google’s search results. If that is the hidden target, rank trackers are collateral damage in a much larger operation.
Why this matters now
For in-house SEOs and performance teams, the warning is practical. A reporting stack built entirely on daily or weekly rank positions is fragile. If Google keeps tightening access, rank trackers may become less frequent, less accurate, or more expensive. Teams that rely on them without a fallback will overreact to false drops or miss real ones.
What to watch next
- AI agent traffic grew 1,700% in one year, and more than half of web traffic is now non-human.
- AI Mode answers cost more because each query can run multiple LLM inferences.
- Google’s anti-scraping actions may target reverse engineering, not just rank tracking.
The practical shift is clear: marketers should treat rank positions as a proxy, not the full story. Build dashboards around clicks, conversions, and landing-page queries so a crackdown on trackers doesn’t blind your reporting.
Source: Search Engine Journal
