April 24th-April 30th // Estimated Reading Time: 10 minutes
In This Edition!
China’s revenge
The New York Times is “excited” about AI-powered digital advertising
Last Week To Secure Your AI Accelerator Seat!
Time is running out to book your ticket for best AI bootcamp in the industry, which will help you level up on all things AI in marketing & advertising.
Our sponsor, Hightouch, just launched Ad Studio, giving performance marketers everything they need to go from creative insights to creative buildout to campaign launch using AI agents. They'll share tips on how to use AI to create on-brand creative and cut weeks off campaign timelines.
Enroll now using promo code EarlyBirdsLearnAI for 20% off!
Top Headlines🔥
China to Meta: No Manus for you 🔒
Source: The Wall Street Journal
April 27th, 2026
Summary: Beijing has ordered Meta to undo its $2.5B purchase of AI agent startup Manus, ruling the deal a national security threat. Meta has a few weeks to hand back all data and technology it received and return Manus to its pre-acquisition state. The two Manus co-founders were barred from leaving China during the review. The situation is already messy: Investors have been paid out, and Meta has moved quickly to integrate Manus into Meta's ad infrastructure, including launching native Meta Ads connectors.
Opinion: Is this China’s revenge for the US forcing divestiture of TikTok?!
Meta gets back $2.5B but loses talent and agentic infrastructure. Manus was a big part of its autonomous AI agent strategy, so it’ll be interesting to see where Meta goes from here. It's also not great for the marketers and agencies that had already started wiring Manus into their campaign workflows.
With premium subscribers declining, OpenAI needs your ad dollars
Sources: The Information, The Wall Street Journal
April 28th, 2026
Summary: OpenAI missed its Q1 revenue and user targets 🔒 and its CFO has raised concerns internally about whether the company can fund its massive data center commitments if growth doesn't pick up. The answer might be ads: OpenAI is forecasting that ChatGPT Go, its $8/month ad-supported tier, will reach 112M subscribers 🔒 this year while the $20 Plus plan loses 80% of its base. By 2030, advertising is projected to be OpenAI's single largest revenue source at $102B.
Opinion: OpenAI is taking a page from the Netflix playbook by leaning into a cheaper, ad-supported tier. For marketers, the direction is clear: ChatGPT wants to become a mass-market ad platform. But ChatGPT still has a user growth problem, largely due to competition from Gemini and Claude. It's hard building an ad business from scratch while simultaneously fighting a two-front war.
AI is supercharging digital advertising 🔒
Source: The New York Times
April 29th, 2026
Summary: An exposé in the New York Times about how AI is Making Digital Advertising Exciting Again?! SIGN US UP 🤣🤣🤣!
Google and Meta just posted their strongest quarterly ad numbers since the pandemic, with Google up 16% to $77B and Meta up 33% to $56.3B. The driver: smarter ads, powered by AI. AI-powered ad buying grew from $1B in 2022 to $35B last year and is on track to hit $56B this year. Irrelevant ads on Google have dropped 40%. Advertisers are saving 30% on campaign costs and up to 65% on content, with many reinvesting those savings into more testing. Companies like L’Oréal and DribbleUp say AI-powered tools are helping them increase revenue.
Opinion: Unfortunately, while normies say “AI is making digital advertising exciting again”, we advertising dorks say it’s just another day in our “Walled Gardens Always Get Richer” reality.
Marketers are handing over more control and getting better results, which makes it very hard to argue against AI tools. But it also means the platforms are starting to know more about what works for your brand than you do. And there’s very little you can do about it. Marketers should proceed cautiously.
New Products & Features 🚀
What It Does: The new connectors let Claude work directly inside creative tools including Adobe Creative Cloud, Blender, Autodesk Fusion, Ableton, and Splice. Depending on the tool, Claude can generate and modify 3D models through conversation, automate repetitive production tasks, write custom scripts, and search sample catalogs without leaving Claude. Anthropic is also partnering with RISD, Ringling College, and Goldsmiths to bring the tools into art and design programs.
Quick Take: A designer who can describe a scene and have it open in Blender, or batch-process assets through a conversation, is a much stickier Claude user than someone who opens an AI chat window.
What It Does: Meta ads AI connectors let advertisers create, manage, and analyze campaigns directly from their AI tools. Advertisers can pull reports, create and edit ads in plain language, manage product catalogs, and check data quality, all connected to their live Meta ad account.
Quick Take: Meta doesn't need you to live in Ads Manager anymore. It just needs your money. Opening up to third-party AI tools is how they stay front-and-center even as the workflow layer shifts — without having to build a better interface themselves.
Snapchat introduces AI Sponsored Snaps, letting brands run conversational ads in chat
What It Does: Brands can deploy their own AI agents directly inside Snapchat's Chat tab. Users can ask questions and get personalized recommendations without leaving the conversation. Experian is the first partner in the alpha.
Quick Take: AI ads are creating a surge of interest in conversational ads. Snapchat user behavior is already conversational, so this could work. Experian being the first partner is a weird flex though — users are supposed to get excited to talk to an Experian agent about their credit scores?
What It Does: Dappier lets brands insert AI agents into AI chat sessions across open-web publisher sites. Dappier converts existing ad creative into conversational ads sold programmatically through DSPs StackAdapt, Basis, and Bedrock. Baby monitor brand Miku saw 3x higher response rates versus standard web placements, jumping to 12x when targeted with real-time intent signals.
Quick Take: Publishers need creative ways to monetize as AI continues to suck their traffic away. Any one way probably won’t fill the gap, but perhaps a panoply will.
What It Does: Realize+ automatically moves budget to top-performing campaigns in real time and continuously generates and updates ad creative without manual input. Taboola is also opening the platform to Claude Skills, letting advertisers manage and optimize campaigns directly from within Claude.
Quick Take: Taboola's pitch is that the open web can now offer the same automated performance that Google and Meta do.
What It Does: AdBridge pulls in your Google Ads campaign info and rebuilds it for ChatGPT Ads, filling keyword gaps and writing ad copy. The output is a single file you can import directly. One-time use is free.
Quick Take: Google campaigns being used as training data for ChatGPT Ads. Everything you've learned about keywords, copy, and audience behavior on Google is now the starting point for a completely different ad ecosystem. Seems like the wrong starting point.
AI Use Case of The Week💡
Hershey gives MMM an AI glow-up 🔒
The Setup: Hershey runs marketing mix modeling (MMM) across a large brand portfolio including Reese's and Skinny Pop to analyze how its media spending drives sales. The problem is that the results always arrive months later: By the time Hershey had a full read on 2024 data, it was already planning for 2026.
The AI Solution: Hershey partnered with Tracer to clean and standardize fragmented data across its marketing and retail systems. That data feeds into Mutinex, an AI-powered MMM platform built on Claude and Gemini. Mutinex uses a multi-agent system where each AI agent specializes in a different domain: marketing econometrics, competitive pricing, and model diagnostics. Together the two platforms replaced Hershey's manual, infrequent modeling cycles with an always-on system capable of running monthly across its entire brand portfolio.
The Results:
• MMM frequency increased from 3x to 12x per year
• Models can run in 3 weeks vs. months
• Full brand portfolio now measured monthly vs. five brands previously
• Hershey projects 4-5% increase in revenue attributable to media due to faster insights
Why It Matters: Marketing has often struggled to prove its value because the data arrived too late to defend decisions already made. Monthly MMM could give marketers a powerful argument for their budgets.
Your Action: The Hershey example only works because Tracer fixed its data pipeline first. Before evaluating any MMM vendor, answer these questions:
1. How stale is your media data? Pull last month's campaign report and check the timestamp on the most recent data row. If it's more than 72 hours old, you have a pipeline problem.
2. How many humans touch your data before it's usable? Every manual export, format conversion, or email handoff is a delay and an error point. If your answer is "more than two," your modeling frequency will be capped.
3. Are your retail/trade data and media data in the same place? MMM only works when you can connect media spend to sales outcomes. If your media data lives in one platform and your retail data lives in another (or in someone's spreadsheet), you're not ready for always-on modeling.
If you can't answer all three satisfactorily, the right vendor conversation to have right now is with a data infrastructure company, not an MMM platform.
Other Notable Headlines📌
Epic hedge: Google invests $40B in Anthropic - Google's funding offers fresh capital and 5 gigawatts of cloud computing capacity. Anthropic's valuation could climb toward $900B in new funding talks (that's more than OpenAI's recent $852B valuation). Amazon is investing too 🔒.
Hightouch raises $150M at a $2.75B valuation to build agentic marketing platform - The marketing data platform says one customer used its agents to generate ad creative 80% faster and drive $50M+ in incremental annual revenue.
News organizations push back against web archive used for AI training 🔒 - CNN, NBC, and USA Today are among 20 publishers asking Common Crawl, a nonprofit repository widely used to train AI models, to remove their content and stop unauthorized use.
Salesforce is crowdsourcing its AI roadmap — with customers - Rather than building to fixed product timelines, the company is using real-time input from 18K customers to determine what to build next, with some customers meeting Salesforce teams as often as once a week.
Sparks fly during trial over OpenAI's fate - Elon Musk, who co-founded OpenAI as a nonprofit in 2015, is suing to block the company's conversion to a for-profit corporation, alleging co-founders Sam Altman and Greg Brockman betrayed their original mission.
They don’t like each other…
That’s It For This Week 👋
The U of Digital AI Edge Newsletter is intended for AI Edge Pro subscribers, but occasional forwarding is okay!
To subscribe visit Uof.Digital/Newsletters or AI Edge Pro, or contact us directly for group subscriptions.
And remember, U of Digital helps teams drive better outcomes through structured education on critical topics in marketing / advertising like AI, ad tech, mar tech, privacy / identity, CTV, commerce media, and more. Interested in learning more about how we can supercharge your team? Let’s talk!
Thanks for reading!
