Good morning. This is The Pattern for Monday, April 06, 2026.
Geopolitics just became a supply chain problem that marketing departments are going to inherit. Business of Fashion reports that sportswear brands are facing tough decisions around the Iran war. Oil prices are surging. That means production costs are climbing. Shipping is more expensive. The question every brand is wrestling with right now is whether to absorb those increases or pass them to consumers.
Nike, Adidas, Asics. They all have the same calculation in front of them. And here is what matters. This is not a finance department issue that marketing can ignore. If you raise prices by 15 or 20 percent, you are repositioning your brand overnight. You are testing whether your logo has enough equity to justify the new price point. You are asking customers to value your product more at the exact moment they are feeling economic pressure everywhere else.
Some brands will pass this test. Others will not. And the ones that fail will spend years trying to recover.
Signal one. OpenAI and Anthropic just admitted something the AI industry did not want to say out loud. Documents reported by the Wall Street Journal show that inference costs, the cost of actually running these models, exceed half of their revenue. Not training costs. Running costs. This is the profitability problem nobody wanted to acknowledge. Every time someone uses ChatGPT or Claude, it costs real money.
And those costs are not declining fast enough to make the unit economics work at scale. If you are a brand investing in AI infrastructure right now, model the cost of running it, not just building it. The hype cycle sold everyone on training models. The reality is that serving them is where the money disappears.
Signal two. Dezeen has a lookbook of tattoo parlours that position themselves as ceremonial spaces, not service businesses. This matters because it shows what works in physical retail right now. Tattoo artists are not competing on speed or convenience. They are treating every session as a ritual. The space is designed to feel sacred. The transaction becomes an experience people remember. Compare that to most retail, which optimises for throughput and efficiency.
If your brand has physical locations, the question is whether you are designing around ceremony or around getting people in and out faster. The brands that survive the next wave of retail consolidation will be the ones that made their stores worth visiting for reasons beyond the product.
Signal three. Microsoft is force-updating Windows 11 with no way to fully opt out. The company says an intelligent machine learning system handles the rollout. Translation: they stopped asking permission. This is the shift from consent to compliance. Big Tech has decided that user choice slows down adoption too much. Expect consumer backlash campaigns. The brands that differentiate in the next 12 months will be the ones that offer control as a feature, not a bug.
Signal four. James Franco's brand Paly just scored a collaboration with the Rolling Stones, and it happened because Selfridges made the introduction. Glossy reports this is Paly's first official collaboration. What matters here is that retailers are becoming brand incubators and matchmakers, not just distribution channels. Selfridges is not selling shelf space. They are selling access to their network. If you are launching a brand, pitch retailers on what partnerships they can enable, not just how many units you will move.
Signal five. China's box office just crowned a Nintendo film while Hollywood struggles domestically. The Super Mario Galaxy Movie launched to number one in China. IP that works globally now gets built in markets Hollywood used to call secondary. Entertainment companies need to greenlight projects based on Chinese audience data, not US focus groups. The centre of gravity is shifting faster than the industry is adjusting.
Here is the pattern. External shocks are exposing which business models actually work under pressure. The Iran war is testing whether sportswear brands have real pricing power. OpenAI is admitting that inference costs make profitability harder than anyone expected. Microsoft is abandoning user choice because voluntary adoption is too slow. These are not three separate stories. They are three versions of the same realisation.
The era of growth through optimisation is over. What worked when oil was cheap, compute was subsidised, and users were patient is now breaking. The brands and platforms that survive the next 18 months will be the ones that redesigned their economics before the crisis forced them to.
Yesterday we predicted at least two major AI companies will announce media or creative agency acquisitions before June. Worth watching.
That's The Pattern for today. Before it's obvious. See you tomorrow.