AI news, data, and consequences
for marketing leaders
Section
Where the money goes: AI line items in marketing budgets, vendor pricing moves, agency economics, and the spend decisions CFOs are starting to question.
Ramp's card data shows businesses capping AI spend even as the best model on the market underperformed on adoption. Ramp's own product says an eighth of that spend is recoverable.
Forrester's new model scores 200-plus markets for AI disruption. Its own research says three in four agencies pay for the AI themselves and never bill the client.
Dentsu's updated plan cuts up to 160 more international entities and names an AI-native operating model as the route to industry-leading profit per employee.
The stock-image giant's Q2 came with a dead merger, withdrawn guidance, and a named mechanism: AI answers are cutting the search traffic that fed iStock's new customer pipeline.
The design tool most small marketing teams run on told shareholders its AI features cost too much to roll out. It rebuilt them to run up to 30 times cheaper.
Both numbers come from the same survey of 602 marketers. The gap between them is where a lot of budget is currently sitting.
Perplexity called Time's markdown ads deceptive and would not define the term. Mobian's own site says every one of them opens with a sponsored label.
Half of large organizations narrowed, delayed or paused agent rollouts when costs outran value. The split between the ones getting returns and the ones not is whether they could see the bill.
Buyer Direct has been in beta since November, booking publisher inventory straight out of Google's ad server. The industry spent 2026 building AI agents to do the same job.
Both big independent ad verification firms are now privately owned. That lands right as software starts placing the media nobody is watching.
Bending Spoons is paying $1.285 billion for the tool running thousands of marketing teams' calendars, trackers, and workflows. Its record with Evernote and WeTransfer says what usually comes next.
OpenAI told advertisers by email that automatic advanced matching would switch on for every existing ad pixel on August 17. That date has passed. Unless someone opted out, it's running now.
Elad Gil says the next fight is over who gets a share of the AI budget. Uber, Accenture and Amazon already started rationing, and marketing has no cost-per-outcome number ready.
Nine weeks after launching hard daily caps, OpenAI switched live campaigns to seven-day averages. Google publishes how far it can overspend in a day. OpenAI doesn't.
Every marketing plan built since 2023 rests on one assumption: the model gets cheaper every year. The people who buy the computers are now arguing the opposite.
The first head-to-head data says machines pay less per impression and win more of the bids they chase. The industry's own forecast says almost no budget goes through them next year.
Time stripped its pages down to plain text so AI systems could read them, then started selling sponsored blocks inside. Ally Bank and the Project Management Institute are among the first buyers.
Claude Opus 5's headline gains are all in reasoning and agent work. On writing, the frontier models sit within about 50 points of each other. The upgrade lands somewhere other than the blog draft.
Jess Graham calls it the discovery tax: when an AI agent does the choosing, a brand wins the sale but loses the pricing power of being chosen. SparkToro's data shows you can't optimize the pick.
SE Ranking ran 50,032 commercial keywords through AI Mode. Ads showed on 29.45% of them, and the advertiser's own domain made the cited sources 11.53% of the time.
A revised Terms of Service took effect July 1 with no acceptance step, making automation the baseline condition. Advertisers still own everything it produces.
Moonshot's Kimi K3 put frontier-grade AI into open weights anyone can run. For marketers the story isn't China. It's the metered vendor bill, and the customer data you couldn't safely send out.
Ozone's Q2 benchmarking is the first clean measure of zero-click search on the sell side: fewer pages loading, fewer ads to buy, and rising prices for what's left.
Publicis raised guidance on a business where AI-powered marketing services are 87% of net revenue. Its own clients describe AI as a small slice of the actual work.
OpenAI told investors ChatGPT ads would hit $100 billion by 2030. eMarketer caps the entire chatbot ad market near $5 billion, and says the real AI ad money sits beside the answer, not inside it.
OpenAI's GPT-5.6 drops frontier AI to as low as $1 per million words in. It resets what a small marketing team can automate, but the per-token sticker is the wrong number to budget from.
HubSpot now bills $0.50 only when its AI resolves a ticket, $1 only for a qualified lead. Zendesk, Intercom, and Sierra already do the same. Your martech budget is now a variable cost.
A Spencer Stuart survey finds 37% of marketers at the largest firms were told to cut costs 20%+ in two years, with AI as the excuse. Zero percent say they've fully transformed with it.
Profound's Aim agent watches your brand across AI assistants around the clock, then drafts and routes the work. Always-on is the pitch. It's also a meter: agentic tasks can cost 30x a chatbot's.
Gartner's 2026 spend survey shows marketing money moving into AI faster than the process, data, and talent needed to earn it back. CFOs have noticed.
Gartner shows 39% of CMOs cutting agency budgets while paid media hits a five-year high. AI is moving production in-house. The retainer is the first line an AI-era CMO reaches for.
Global ad spend cleared $1 trillion and paid media hit a five-year high of 31.4%, but marketing budgets stayed flat. Every AI and media dollar is coming out of agencies, brand, and people.
The 2025-2026 data is consistent: marketing money is flowing into AI faster than the return shows up. Gartner, McKinsey, and BCG describe the same gap, and the CFO has noticed.
The net effect of AI on your marketing: the stories that matter, twice a week, in five minutes.