AI is now the middleman between brands and their customers — and marketers either account for that or get left out. Two stories this week show how forward-thinking brands are building for the agentic discovery layer, while brands without clean data and authority signals are about to get filtered out by the algorithms doing the shopping.
Microsoft Copilot Just Became a Checkout Counter for 500,000 Merchants — Without Anyone Leaving the Chat
Microsoft has expanded Copilot Checkout to more than 500,000 merchants and brought it into the Copilot mobile app for U.S. users. Shoppers can now discover, compare, and buy products — including from Urban Outfitters, Anthropologie, Ashley Furniture, Etsy sellers, and Shopify merchants — without ever leaving the AI conversation. PayPal, Stripe, and Shopify handle payments. The retailer stays the merchant of record. Microsoft controls the interface.
The performance data Microsoft is citing is significant: shopping journeys involving Copilot are 33% shorter than traditional search paths, see a 53% increase in purchases within 30 minutes, and when shopping intent is present, are 194% more likely to result in a purchase than journeys without Copilot. The expansion also added WooCommerce compatibility, Universal Commerce Protocol support, and loyalty program integration — signaling this is no longer an experiment. It's infrastructure.
The strategic implication for brands is stark: whoever shows up in AI-assistant shopping interfaces with the best-structured product data, the clearest catalog metadata, and the most authoritative presence wins the sale — before the customer ever opens a browser tab. Adobe reported that AI-driven retail traffic surged 693% during the 2025 holiday season. If that trend continues through 2026, the Copilot checkout isn't a channel addition. It's a channel shift.
Copilot Checkout reaching 500,000 merchants isn't a product update — it's a market structure change. The checkout is moving upstream into the conversation, and brands without clean product feeds and structured content in AI-readable formats are going to get filtered out before a human even sees them. This is the same urgency I've been saying about GEO and AI SEO — but now it's not just about being cited. It's about being purchasable at the point of discovery.
This is a textbook channel disintermediation case. Microsoft is doing to e-commerce what Amazon did to search — inserting itself between the brand and the customer at the moment of intent. The brands that win in an agentic commerce world are the ones that have invested in first-party data infrastructure and product content quality. Those are boring, operational investments. But they're suddenly the most important marketing assets a brand has.
Meta Reports Tuesday. Analysts Are Expecting AI-Driven Ads to Hit $54B. The Real Question Is What Happens to Margins.
Meta Platforms reports Q1 2026 earnings on April 29 after market close. Analyst consensus sits at $55.5 billion in revenue (up 31% YoY) and $6.65–$6.73 EPS. Advertising revenue specifically is expected to hit $54.36 billion — a 31.3% year-over-year jump that would make this quarter a landmark proof point for AI-powered ad monetization. BofA's Justin Post has his numbers even higher: $56 billion in revenue and $7.44 EPS.
The central tension going into this print: Meta's AI ad infrastructure (Andromeda, Advantage+, the new AI creative tools) is clearly driving performance improvements. Ad impressions were up 18% in Q4, average price per ad up 6%. The AI story is working. But expenses are surging — 40% in Q4 — and 2026 capex guidance of $115–$135 billion is an extraordinary bet on AI infrastructure that needs to keep delivering returns. Morningstar notes that 2026 may be the first year Meta's ad sales exceed Alphabet's on a net basis.
For marketers, this earnings call is worth watching because Zuckerberg will lay out how AI agents — specifically the Andromeda ad retrieval system and the expanding AI ad creative suite — are performing at scale. The brands currently running AI-first ad structures on Meta are the ones with the most to learn from what gets disclosed Tuesday.
Every marketer running Meta ads should tune into this earnings call or read the transcript. When Zuckerberg talks about Andromeda and AI creative performance, he's telling you how Meta's algorithm is evolving — and what types of content are winning. The big insight from last quarter: diverse, distinct creative concepts outperform multiple minor variations. That's a brief-writing insight, not just a tech update. Watch for any signals on AI-generated creative adoption and what that means for brand safety thresholds.
Instagram's native affiliate data shift is bigger than it looks. When brands can see Reels-to-purchase attribution inside Meta without routing through LTK or ShopMy, the data ownership question changes entirely. Here's what that means operationally for brands running creator programs right now.
Instagram Gave Brands Their Attribution Data Back. Here's What That Actually Changes.
Adam Mosseri's announcement that creators can now tag affiliate products directly inside Reels — with conversion data flowing natively inside Meta — is being underreported relative to its strategic significance. For the past several years, every brand running creator partnerships through LTK or ShopMy handed over something they rarely acknowledged: the attribution layer. Those platforms solved the "link in bio" problem, but the clicks, conversions, and commission data all lived inside infrastructure someone else owned.
Instagram just solved it natively. Brands with verified product catalogs in Meta's Commerce system can now see which Reels drove clicks and which drove purchases — directly inside Meta Business Suite, without a third-party intermediary. That doesn't make LTK or ShopMy obsolete. Creator discovery, relationship management, and campaign coordination are still real capabilities worth paying for. But it does mean the attribution problem — the main justification for those platform fees — is materially reduced for brands already in Meta's ecosystem.
The operational implication: if you haven't enabled Instagram Shopping and affiliate commerce for your product catalog, do it this week. New Engen's data shows CPMs for paid creator amplification rose 51% from early April to Mother's Day week in 2025. The cheapest window to boost creator content is right now, before May demand spikes drive CPMs up again. The brands with their affiliate infrastructure live going into Mother's Day will have a measurable conversion advantage.
The brands that have been outsourcing their creator attribution to LTK and ShopMy just had their leverage calculation change. Meta taking over the attribution layer natively isn't just a feature — it's a power move. It pulls creator commerce back into Meta's walled garden and gives brands a reason to stay there rather than build multi-platform creator infrastructure. Smart move from Meta. The question for brands: do you have your product catalog in Meta's Commerce system yet? Because if not, you're still routing attribution through someone else's platform when you don't have to be.
Three AI announcements this week that matter for marketers: Google's $40 billion bet on Anthropic, OpenAI's GPT-5.5 launch, and what both mean for where the AI tools your clients will use are actually heading.
Google Is Betting $40 Billion on Anthropic. Here's Why Every Marketer Should Care.
Google confirmed on April 25 it will invest up to $40 billion in Anthropic — $10 billion upfront at Anthropic's $380 billion valuation, with up to $30 billion more tied to performance milestones. This follows Amazon's separate agreement to invest up to $25 billion in Anthropic weeks earlier. Combined, two of the world's largest cloud providers are now betting a combined $65 billion on the company whose Claude models are increasingly embedded in enterprise marketing stacks. Anthropic's annualized revenue has crossed $30 billion, driven largely by surging demand for Claude Code and Claude Enterprise.
For marketers, the strategic implication is this: the AI tools powering Adobe CX Enterprise, Microsoft's agent frameworks, and a growing list of martech platforms are being shaped by Anthropic's model development — and that development just got significantly better-resourced. Google securing a large stake also all but eliminates Apple's previously rumored acquisition path, meaning Anthropic is going to remain an independent AI lab with massive compute capacity rather than being absorbed into any single product ecosystem.
Meanwhile, OpenAI launched GPT-5.5 on April 23, rolling out to ChatGPT and Codex users. The model plans its own steps, catches its own mistakes, and executes document-heavy research with minimal prompting. The AI tool stack for marketing teams is evolving fast — and the two labs receiving the largest infrastructure bets (OpenAI at $852B valuation, Anthropic at $380B) are the ones building models that will power the tools your clients use every day.
$40 billion is not a venture bet. That's a strategic infrastructure acquisition without buying the company. Google is securing compute commitments, API access, and competitive moat against Apple simultaneously. For marketers, the story is simpler: the AI tools you're building your workflows around are about to get significantly more powerful, more expensive to develop, and more deeply embedded in the cloud infrastructure your clients already pay for. Start understanding which AI models power which tools in your stack. That knowledge will matter in client conversations by end of year.
The Google-Anthropic deal is a vertical integration play dressed as an investment. By owning infrastructure, model capacity, and a major customer relationship simultaneously, Google is trying to avoid the commoditization trap in AI services. For students studying competitive strategy: this is the platform economics playbook — control the layer below the application, and you capture value from everyone building on top. Amazon and Google are both running this play. Microsoft is running the same play through OpenAI.
Not Nvidia. Not Meta. This week: the company quietly building the networking backbone that every AI data center runs on — and reporting earnings May 5.
Arista Networks, Inc. (ANET)
Arista Networks is the networking backbone of the AI buildout — and it's barely talked about outside of infrastructure circles. Every large-scale AI data center needs high-throughput, low-latency networking to move data between GPUs. That's Arista's core product. CEO Jayshree Ullal's phrase from February earnings — "AI networking achieving production scale" — signals that what was once experimental is now mission-critical for hyperscalers. Q4 2025 came in at $2.49 billion in revenue, beating estimates of $2.38 billion, with 28.9% YoY growth. The company then guided 2026 to 25% growth, targeting $3.25 billion in AI centers revenue specifically.
On April 25, ANET touched a new 52-week high of $179.80 intraday before closing at $176.91. Earnings are May 5. Wall Street consensus sits at a Strong Buy with an average target of $180.42. KeyBanc's analyst called ANET the superior AI infrastructure pick over SMCI specifically because of its durable margin profile and free cash flow consistency — areas where Super Micro has struggled. With Google's $40 billion Anthropic investment confirmed this week, the AI infrastructure capex cycle shows no signs of decelerating, and Arista is well-positioned as a direct beneficiary.
Arista is the infrastructure story hiding in plain sight. Every data center running LLMs needs what ANET builds. As AI capex from Google, Microsoft, Amazon, and Meta continues to scale, the networking layer becomes more critical — not less. This isn't a speculative AI play. It's a picks-and-shovels position in a gold rush that just got $40 billion larger this week. Earnings May 5 are a near-term catalyst either way.
The cultural moments and media worth your attention this weekend.
🏥 The Pitt
Still the most relentless hour of television running right now. The real-time format creates a kind of tension no other show is attempting. If you haven't started, this weekend is your window.
🎬 Mandalorian & Grogu — Marketing Watch
Burger King's King Jr. Meals with character toys drop April 28 — the first wave of the May 4 campaign launch. Worth watching how the phased rollout drives social conversation before the May 4 full launch. The campaign architecture here is genuinely smart.
📱 YouTube — Claude Cowork
Behind-the-scenes look at building real marketing intelligence with AI in the loop. The experiment is live and getting interesting. Follow along.