AI Licensing Deals: A New Revenue Stream for Publishers

Key Points

• Snowflake's Cortex Knowledge Extensions let publishers monetize paywalled content via enterprise RAG licensing without exposing raw data or surrendering training rights.
• Several publishers, including the Washington Post, AP, and USA Today Network, have signed six-figure deals through the platform, with financial services firms leading enterprise demand.
• Snowflake takes no revenue share on content deals. Publishers and enterprise buyers negotiate directly, and Snowflake earns on storage and compute.
• This is an incremental revenue stream, not a replacement for traffic-driven ad revenue. Publishers need both strategies working in parallel.
• Whatever traffic you retain from organic and direct channels still needs to be monetized well. AI licensing and ad revenue aren't in competition.

See It In Action:

• Serebii Case Study: How a high-traffic gaming publisher optimized ad revenue without sacrificing the user experience their audience expects
• AI Crawler Impact on Lifestyle Publisher Traffic: Real data on how AI crawlers are affecting publisher traffic and what the numbers actually show
• Lifestyle, Health and Travel Ad Revenue Resource Center: Vertical-specific guidance for publishers whose traffic is most exposed to AI-driven search changes

What Happened

Digiday reports that publishers are quietly closing six-figure AI licensing deals through Snowflake's Cortex Knowledge Extensions platform. Seventeen publishers, including the Washington Post, Associated Press, People Inc., and USA Today Network, have signed on.

The mechanism is a monetized RAG pipe. Enterprises query publisher content inside Snowflake's AI environment, get reliable licensed editorial data to power internal AI tools, and publishers get paid without losing control of their content or having their material used to train models. Snowflake's principal product manager Ben Srour told Digiday: "You cannot scrape the data. You can't steal it and use it for model training. So that's why the product has really resonated with publishers."

The commercial structure is notable. Snowflake takes no cut of content licensing fees. Enterprise buyers can draw on existing committed Snowflake spend to pay publishers directly, which sidesteps slow procurement cycles. Contracts are structured as flat-fee licenses or usage-based access. Snowflake earns on compute and storage when those AI queries run.

Essential Background Reading:

• AI and Publishers Resource Center: Everything publishers need to know about AI's impact on content, traffic, and monetization strategy
• AI Crawler Resource Center for Publishers: How AI crawlers work, what they're taking from your site, and what your options are
• Generative AI and Publishers: A grounding overview of generative AI dynamics affecting the publishing industry
• News Publisher Guide: A comprehensive resource for news publishers navigating the current ad revenue and traffic environment

Why This Matters to Publishers

The most honest framing Digiday offers is also the most important one: AI licensing is an incremental revenue line, not a lifeline. It won't offset the referral traffic losses that have compounded across the industry for several years. Publishers treating this as the solution to AI-driven traffic erosion are going to be disappointed.

The structural terms here are meaningfully better than what most publishers have encountered in the AI licensing space. No revenue share is a significant departure from how platform deals usually work. AP's CRO Kristin Heitmann told Digiday the Snowflake exchange offers "unlimited use cases," covering finance, supply chain, crisis management, and regulatory monitoring. That breadth of enterprise demand is real, even if it's still concentrated in financial services and marketing verticals.

A recent Open Markets Institute report, cited in the Digiday piece, warned that AI licensing marketplaces risk repeating the power imbalances publishers experienced during the platform era with Google and Facebook. Snowflake's no-take-rate model positions it differently, but publishers should read that warning carefully regardless. Early terms in nascent markets have a way of setting precedent.

Enterprise demand is still developing. Srour told Digiday that most enterprises are still working on internal AI infrastructure and data governance before seriously investing in licensed publisher content. Financial services firms are the early adopters. Broader demand is coming, but it's not a land rush yet.

Related Content:

• Anthropic's $1.5B Copyright Settlement: What the settlement terms mean for publishers navigating AI content rights and licensing negotiations
• Publishers, Amazon, and Perplexity AI Agent Dispute: How the dispute over AI agents using publisher content without compensation is playing out
• Omnicom's AI Media Buying: What AI-driven media buying means for publisher demand and CPM dynamics
• Publisher Earnings Index: Benchmark your revenue performance against the broader publisher market

What Publishers Should Think About

Publishers evaluating Cortex or any AI licensing deal should work through a few concrete questions before signing.

Content scope and control: Understand exactly which content is queryable, under what conditions, and with what attribution. "No model training" protections need to be contractually specific, not just verbally assured.

Revenue materiality: Six-figure deals sound significant, but context matters. Evaluate what this revenue represents relative to your total ad revenue and traffic-based income. For most publishers, it will be a supplementary line, not a primary one.

Demand concentration risk: Current enterprise demand is heavily weighted toward financial services. If that vertical slows its AI tool investment, licensing revenue could get lumpy fast.

Precedent setting: The terms you accept now, on attribution, pricing structure, and usage scope, may influence future deals with other enterprise platforms entering this space.

The Snowflake model has genuine structural advantages over what publishers have seen from other platforms. No rev share and direct buyer-seller deal terms are both meaningful. But the platform's growth depends on enterprise AI adoption accelerating, which Srour acknowledged is still in early stages.

Next Steps:

• AI Ads and New Monetization Strategies for Publishers: How AI is creating new monetization formats and what publishers should be testing now
• Google's AI Search Updates Give Publishers More Clicks: How to position your content to capture traffic opportunities from Google's evolving AI search features
• Ad Monetization Platform ROI Calculator: Estimate what better monetization infrastructure could mean for your revenue per session
• New Revenue Streams with Playwire: How publishers are adding incremental revenue lines alongside existing ad monetization
• 2026 State of Publisher Ad Revenue Report: Industry benchmarks and publisher revenue trends heading into 2026

The Traffic You Still Have Needs to Work Harder

AI licensing doesn't replace ad revenue, and it doesn't address what's happening to publisher traffic today. Many publishers have watched referral volumes decline as AI tools answer queries that used to send users to their sites. Some have taken steps to block AI crawlers. Others are waiting to see how the landscape shifts.

Whatever your approach to AI crawlers, the sessions you're getting right now need to perform. If your RPS is soft, an AI licensing deal won't compensate for it. That's where ad revenue optimization still matters most.

We work with publishers across gaming, entertainment, education, sports, and news to maximize revenue from every session. That means header bidding architecture that works, direct demand that lifts CPMs above programmatic floors, and ad formats that don't wreck user experience or viewability scores.

If you want to understand whether your current monetization setup is leaving money on the table, start with a conversation. We've got the data to back it up.