Sponsored by
Happy Thursday! Agentic commerce is leaving the demo room. AI shoppers are touching checkout, Google is moving ads into answers, and compute is becoming the toll road behind the whole thing.
The interesting part is not automation itself. It is who owns the risk, the margin, and the customer path when machines start doing the clicking.
In today’s MarketMaze:
🤖 AI agents test checkout risk
🚀 SpaceX sells the bottleneck
🔍 Google puts ads in answers
🛒 TikTok Shop leans on affiliates
🤖 AI marketing keeps humans busy
LET'S ENTER THE MAZE!
The Maze: AI shopping agents are already reaching retailer checkout flows. The harder part is not the purchase; it is proving who authorized it when something breaks.
A UK merchant survey found 58% believe AI-initiated transactions have reached their platforms, while 72% are preparing or planning for autonomous buying.
Liability is split: only 41% felt very confident in current frameworks, and a disputed GBP2,000 agent purchase produced no clear consensus on who should pay.
Standards work from Mastercard and FIDO may help, but merchants still need fraud systems, transaction labels, and service rules that can identify delegated intent.
Why it matters: Agentic commerce will scale only when trust is operational. Retailers need to know not just that a payment cleared, but whether a machine had permission to make it.
The Maze: Anthropic is buying what every AI company is short of: usable compute. SpaceX is turning that shortage into a monthly revenue line.
SpaceX's IPO filing shows Anthropic paying $1.25 billion a month through May 2029 for access to Colossus and Colossus 2 capacity.
Axios put the deal at $15 billion a year, which is material next to SpaceX annual revenue of around $18 billion.
The earlier partnership gave Anthropic access to more than 300 megawatts and 220,000-plus Nvidia GPUs, turning compute supply into Claude's growth valve.
Why it matters: AI infrastructure is becoming a marketplace, not just a moat. The company with spare GPUs can rent the bottleneck to the company with demand.
The Maze: Google is rebuilding paid discovery for AI Search. Ads are no longer just around the result; they can sit inside the answer, the product explainer, and the lead conversation.
Conversational Discovery ads generate creative from the user's query, while Highlighted Answers place sponsored recommendations inside AI Mode answer lists.
AI-powered Shopping ads add Gemini-written product explainers, turning feed data and Shopping Graph context into a new persuasion layer.
The broader GML recap shows the infrastructure shift: AI Max and Performance Max become the gateway into these AI Search ad experiences.
Why it matters: Search ads are becoming model-mediated. Brands still buy attention, but Google increasingly shapes the words, context, placement, and path to checkout.
The Maze: TikTok Shop's European rollout is not just a marketplace story. The UK has the scale, but Spain, France, and the UK show the real engine: creator distribution with checkout attached.
The UK Top 50 generated EUR 52.58M in one 30-day window, more than Germany, France, Italy, and Spain combined.
Affiliate sales made up 87.0% of Spain's Top-50 GMV, 82.0% in the UK, and 81.7% in France, so the salesforce is mostly creators rather than store teams.
Italy breaks the pattern with 34.3% self-operated GMV, showing that niche sellers and founder-led accounts can still win when the local category mix rewards owned audiences.
Why it matters: Brands entering TikTok Shop need a country-by-country operating model, not a generic channel launch. Rent creators in Spain, France, and the UK; build more owned-account muscle in Italy; run both engines in Germany.
The Maze: McKinsey estimates that agentic AI could power 70% of creation work, 70% of execution and optimization, 60% of ecommerce and web experiences, 60% of enablers, and 50% of strategy. Translation: the first big shock is not AI replacing marketers, but AI compressing the execution layer underneath them.
Creation and execution sit highest at 70%, because content production, media optimization, localization, agency management, and campaign testing are repeatable enough for agents to run faster than human-only workflows.
Ecommerce and web experiences sit at 60%, which matters because AI-mediated discovery and agentic commerce push marketing closer to product content, answer readiness, web experimentation, and checkout trust.
Enablers also sit at 60%, so data foundations, measurement, tech-stack integration, and governance become the control layer, not the admin layer, as platforms like Adobe build agent orchestration into marketing systems.
Why it matters: Retail and ecommerce marketers have spent years adding tools. Agentic AI rewards the teams that rebuild the loop from insight to content to media to ecommerce experience to measurement, with humans setting the rules and agents doing the heavy execution.
🏬 Everything else in Ecommerce & Big Tech
💳 Klarna released an AI shopping search app inside ChatGPT, pushing product discovery into the assistant layer where shoppers already ask.
🎥 ASOS brought shoppable video content to ChatGPT in the UK, turning retail media assets into conversational merchandising fuel.
🤖 Meta tightened its headcount message around AI investment, showing how platform operators are trading labor capacity for compute and model spend.
🛒 Google Merchant Center added AI performance insights and conversational attributes, giving sellers more machine-readable diagnostics for visibility.
☁️ Alibaba rebuilt its AI stack with a new chip, flagship model, and cloud upgrades, sharpening China’s commerce-infrastructure race.
🚚 FedEx approved the spinoff of its trucking business, separating freight economics from parcel delivery as logistics networks get more specialized.
🇪🇺 Otto opened its marketplace to European sellers, giving more brands a route into German demand without building a local channel from scratch.
SHARE THE MAZE
Your network thinks you’re as smart as the content you share. Share smarter stuff and help us grow. Win–win. Here’s what you get when friends join the Maze:
Here is your unique referral link to share with friends:
and link to the hub to check your progress.
THAT’S IT FOR TODAY!
You’re the reason our team spends hundreds of hours every week researching and writing this email. Please let us know what you thought of today’s email to help us create better emails for you.
What do you think of this issue?
If you enjoyed it please share it with a friend, or share it on LinkedIn and tag me (Artur Stańczuk), I’d love to engage and amplify!
If this was forwarded by a friend you can subscribe below for $0 👇
See you next time in the maze!
MarketMaze team
