Tariffs are shifting. De minimis rules are under pressure. And new enforcement could change how retailers operate overnight. If you sell across borders—through ecommerce, marketplaces, or direct—these changes don’t just impact compliance. They impact your pricing, margins, and conversion rates.

Register for our Trade and Tariff webinar series for regular updates.

This week gave us three different ways to read a company's real AI strategy, and none of them came from a press release. They came from the money.

Start with Shopify. It added $3B to its buyback, pushing the authorization to roughly $5B. This from the company that won't stop reminding you it was built for the AI era. Compare that to Google, Microsoft, and Amazon, who are spending everything they can get on data centers, chips, and models. Shopify grew 34% last quarter and decided the smartest use of the cash was handing it back to shareholders. A buyback is the cleanest way to protect a stock price. It is not how you win a platform shift. The financial hats won this round, and a CFO with a Wall Street background knows exactly which levers move the stock. A dividend would at least be honest about what kind of company this has become.

Then Apple. At WWDC we finally got the new Siri, the one that reads your screen, your email, your photos, and acts across apps. The engine underneath it is Google Gemini. Apple reportedly pays Google around $1B a year for that, which is almost funny next to the $20B Google pays Apple every year to stay the default search engine on iPhone. So Apple takes $20 billion from Google with one hand and pays $1B back with the other to run its flagship AI feature. This is the Apple playbook. Rent someone else's technology, ship it, then insource it once the path is clear. They did it with chips. They'll do it here. The open question is what happens to your privacy story when two companies this size share a stack.

Then Anthropic. It put out a memo sketching three futures for AI right as it lines up an IPO. The plateau, where models stop getting meaningfully better. The partnership, where AI keeps improving but still needs human judgment. And the runaway, where AI improves itself faster than anyone can steer. Anthropic also claims Claude already writes 80% of its own code, which makes the third scenario less of a thought experiment than the company probably wants you to sit with. According to the discussion we are 18 months to 2 years before we know which future we're actually in.

The messaging from all three points the same direction, toward AI, and toward the future. The spending tells a more honest story.

Connecticut just became the second state to ban surveillance pricing, and the more interesting question is what retailers are actually losing.

The answer, for now: not much. Gov. Lamont signed a bill barring grocers from using personally identifiable data to set individual prices. It takes effect July 1, 2027. Maryland's version, signed in April, goes live this October. New York's One Fair Price Act passed the legislature and sits on Hochul's desk. Three states moving the same direction inside two months.

But notice the timing. Personalized grocery pricing at the individual-shopper level barely exists today. Lawmakers are banning a capability that's mostly theoretical, which is why the political cost is near zero and the votes are easy.

The tell is Colorado. Polis vetoed his state's version last week, calling it too broad. He's right to worry. Where does a personalized discount end and an illegal "individualized price" begin? Loyalty programs already price people differently based on data. That distinction is the whole ballgame, and no bill has drawn it cleanly.

So grocers don't lose pricing power they're using. They lose optionality on pricing power they were planning to build. Watch the electronic shelf label bills next. That's where it gets expensive.