Cloudflare has declared it’s found optimizations that reduce the amount of hardware needed for inferencing workloads, and is in early talks to re-invent the World Wide Web for the age of AI
That latter proposition emerged last week in the company’s Q4 earnings call, during which CEO, chair and co-founder Matthew Prince observed that his company “counts many of the most important AI companies as customers. We also count a huge portion of the world’s content creators as our users. Being between those two puts us in an important role to help figure out the business model of the post search web.”
Prince thinks Cloudflare therefore enjoys “a unique position to help figure out how content creators are compensated, what agents are allowed where and on what terms and how the AI-driven web of the future will fit together.”
He wasn’t just spit balling, as he also revealed “the conversations we’re having with all the relevant parties feel foundational for the future. Watch this space, definitely exciting times.”
Which sounds a lot like Cloudflare wants to use its position as a big provider of content delivery networks to regulate what bots can access what content, and put a tollbooth in the middle somewhere so that content providers don’t end up giving away their stuff to AI. Many large language models were trained by scraping the internet for content, a practice content creators feel was abuse of copyright but AI outfits feel was fair use. Several court cases will soon explore both arguments, even as major content companies strike deals that will see them paid by model-makers.
Prince CEO also thinks Cloudflare can win in the AI market with its serverless “Workers”, thanks in part to GPU optimization techniques.
“The killer application for Cloudflare Workers is turning out to be AI,” Prince enthused, because serverless is pay-as-you-go so developers don’t need to hire servers that are sometimes idle and can instead use serverless systems on demand.
“And just last month, the world was amazed at the efficiency the team of clever engineers in China were able to deliver in the field of AI training with the DeepSeek model,” he continued. “We are seeing that there are equivalent optimizations that can be made with AI inference on Cloudflare’s platform, resulting in faster performance and lower prices for customers and higher margin and less Capex for us.”
“We believe inference is a bigger opportunity than training and our team continues to find step function breakthroughs that put us well ahead of any alternative,” he added.
Less resource-hungry AI and re-architected web were mentioned as future opportunities Cloudflare is currently pursuing and which will impact future results.
The final quarter of the company’s financial year ended December 31st 2024 saw 27 percent year-over-year growth and $459.9 million of revenue, but a net loss of $12.8 million.
Full year revenue of $1.67 billion represented 29 percent growth, but a $78 million loss that was less nasty than the $184 million of red ink in FY 2023.
The company forecast Q1 2025 revenue of between $468 and $469 million, and a full year cash haul of $2.09 to 2.094 billion.
To get there, the company probably needs to stop breaking itself, as happened on February 6th when its R2 object storage service went offline for 59 minutes, during which time data was not available.
Director of Product Matt Silverlock and staffer Javier Castro admitted the outage occurred “due to human error and insufficient validation safeguards during a routine abuse remediation for a report about a phishing site hosted on R2.”
As Cloudflare staff worked to take down the phisher, their efforts “resulted in an advanced product disablement action on the site that led to disabling the production R2 Gateway service responsible for the R2 API.”
If Cloudflare had done this right, only the R2 bucket and endpoint used by the phisherpholk would have gone down. Instead, the entire service went offline. The company is now revisiting its processes to make sure this doesn’t happen again. ®
