Turn insights into solutions

The DTC premise was arithmetic: remove the retailer, keep the margin. In practice, the retailer's margin was replaced by customer acquisition costs that never stopped climbing. Across the DTC sector, acquisition costs have risen 222% over eight years, driven by ad auction inflation and, from 2021, Apple's App Tracking Transparency making the targeted advertising that enabled DTC brands to scale efficiently far more expensive. The middleman turned out to be useful. He had a shop with foot traffic.

But the ad squeeze only exposed a deeper problem: many of these brands were never selling products at real prices. They were selling venture-subsidised prices. A Casper mattress, a Blue Apron meal kit, a Dollar Shave Club razor — each was priced below true cost to drive growth numbers that would justify the next funding round. Consumers experienced what felt like a better deal. It was really a transfer from VC balance sheets to their doorsteps. The DTC segment accounts for roughly $213 billion, but the growth that built it was bankrolled by a venture capital market that went from $60 billion annually in 2012 to $643 billion in 2021. When that money dried up, prices normalised, and there was no loyalty to protect — because the loyalty in many cases had been to the discount, not the brand.

Besides abundant venture capital, the model depended on a temporary window of cheap digital reach and sparse competition for online attention. Once those conditions normalised — partly because DTC brands themselves flooded the ad market — the economics inverted. Nike went furthest with the thesis, targeting 60% direct sales by 2025. It reversed course after a 10% drop in digital sales and a $28 billion hit to its market value, rebuilding wholesale relationships it had spent years dismantling. Glossier, the brand built to bypass retail, generated roughly $100 million in its first year at Sephora — more than its direct channel had managed in comparable periods. Warby Parker, the original DTC poster child, now operates 230 physical stores and calls word-of-mouth its primary acquisition channel. Each case is a version of the same admission: distribution, not directness, drives scale.

US DTC ecommerce's share of total retail ecommerce plateaued at around 19% in 2025 and is forecast to remain flat. The channel still grows in absolute terms, but it no longer gains share — because every legacy retailer now runs the same playbook. The gap DTC brands once exploited — incumbents with poor digital presence — closed years ago. What remains is a common operating structure of mixed channels, shared fulfilment costs, and identical ad auctions. ...The distinction between "digital-native" and "traditional" has quietly dissolved. One former DTC darling, Allbirds, has abandoned retail altogether, sold its brand, and rebranded as NewBird AI — a company that will buy GPUs and lease access to smaller businesses. A brand built on eliminating the middleman is now, quite literally, becoming one.

OpenAI was founded on an explicit premise: that AI could be the most dangerous technology in human history, and that therefore the people building it needed to be held to an unusually high standard of integrity. The structure was designed to enforce this — a nonprofit board with authority to fire the CEO if he couldn't be trusted. In November 2023, it tried. The board had spent months documenting concerns: misrepresented safety protocols, patterns of deception, internal memos compiled by chief scientist Ilya Sutskever. They fired Altman. Within five days, 95% of employees had signed a letter demanding his return, Microsoft had moved to hire him, and the board that fired him had been replaced. The independent investigation commissioned as a condition of his return was never written down. Altman came back. Besides a story about one person's character, it is a story about what happened the first time the governance structure designed to constrain a character was actually tested. It failed miserably due to the self-interest of everyone involved. Employees had equity. Microsoft had $13 billion. Investors had a $90 billion valuation.

What makes this pattern larger than OpenAI is that it is not unique to Altman. Asked directly who should govern AI, his answer is consistent and revealing: aviation safety worked because society wanted it, regulation followed public demand, the same will happen with AI. It is evasive by placing responsibility everywhere except inside the company deploying the technology. Mustafa Suleyman, who co-founded DeepMind and knows the risks better than most, wrote an entire book diagnosing the same governance problem and proposed "containment" as the response — a concept that names the challenge without identifying who bears the primary obligation to meet it. Different vocabulary, same evasion. The people best positioned to govern AI are, structurally, the ones most invested in not being constrained by governance.

The governance gap is not waiting to be filled while society catches up. The deployment is already happening. The speed of deployment is itself a governance position: it forecloses options, creates dependencies, and shifts the baseline of what is politically possible to reverse. By the time the regulatory frameworks Altman describes arrive, they will be regulating a world the companies have already built. Every serious governance framework in history — aviation, pharmaceuticals, nuclear — was built on the premise that sincerity from the people building the technology is not a substitute for accountability structures they do not control. AI is the first technology of comparable consequence where the answer to that question has been: let's see.

The FDA warning letter was sent before the Times story ran. This is a detail the profile mentioned towards the end. Matthew Gallagher did use AI to build Medvi — ChatGPT and Claude wrote the code, Midjourney and Runway generated the ads. AI handled customer service. What he compressed to near-zero was not just the operational cost of a health company but also the friction cost of building one that looks trustworthy without being trustworthy. The before-and-after patient photos were deepfakes. The doctor profiles were fabricated. The website disclaimer — small print at the bottom — reads: "Individuals appearing in advertisements may be actors or AI portraying doctors and are not licensed medical professionals." One paragraph later, the site promises "doctor-led plans and coaching."

What Medvi was actually selling is the more troubling part. Compounded GLP-1s were legal during the US drug shortage period; Medvi, like hundreds of similar storefronts, kept selling them after the FDA declared the shortage over in April 2025. Its oral tirzepatide tablets — one of its headline products — are, according to a class action filed in March 2026, biochemically incapable of working: tirzepatide is a large peptide molecule that digestive enzymes destroy before it reaches the bloodstream. The only FDA-approved oral GLP-1 required a specialised absorption enhancer developed over years of research. A compounding pharmacy cannot replicate that. The customers paying $150-300 a month were largely buying something with near-zero efficacy, sold to them by an AI-generated doctor who does not exist. Sam Altman told the Times he would "like to meet the guy." He should probably meet the 250,000 customers first.

GLP-1 receptor agonists — the class of drugs behind Ozempic, Wegovy, and Mounjaro — have produced the first sustained decline in US obesity rates in decades. North America represents 77% of a $64 billion global market. In Europe and most of the world, access is almost exclusively private and unaffordable for most.

The coming fork is the patent expiry. Semaglutide patents expire in 2026 in Brazil, Canada, China, India, and Turkey — covering roughly 40% of the world's population. The Lancet published analysis in March 2026 estimating that generic injectable semaglutide could be produced for $28 per person-year and distributed across 160 countries covering 84% of the global obesity burden by end of 2026. The incumbents' response is to build patent thickets around delivery devices and new formulations — Novo Nordisk and Eli Lilly are both investing heavily in oral versions, combination therapies, and device patents that could extend exclusivity regardless of what happens to the core molecule.

The obesity burden is highest where income is lowest — Mississippi, West Virginia, and Louisiana all have obesity rates above 40% and some of the worst affordability ratios in the US. The same pattern holds globally. If the biosimilar wave reaches patients rather than being blocked by device patents and regulatory barriers, this becomes a genuine public health intervention at scale. If it doesn't, it remains an expensive consumer product for wealthy countries with a parallel grey market of dubious compounded versions for everyone else. Both outcomes are currently possible. The industry's track record on access suggests we should not assume the better one.

The gel in the shower, the cream on the hands, the fragrance that stays on skin for hours: they are repeated sensory experiences that accumulate into something closer to a ritual. Most brands focus on the packaging: an elegant bottle and a pleasant texture as finishing touches for a functional product. A few brands understood earlier that the sensory experience — first encountered in the store, confirmed every morning in the bathroom — was not decoration, but the key to strengthen relationships.

Rituals built a €1.7 billion business from a single Amsterdam store in 2000 by making this argument at an accessible price point: everyday routines could be transformed into moments of genuine pleasure. The brand remains independent and founder-led.

Aesop operates on the same philosophical premise at a higher price point and with a more deliberately theatrical retail experience — each store designed by a local architect, each visit structured around a central sink where a member of staff will wash your hands with the product.

Le Labo takes the ritual further: every fragrance is hand-blended at the moment of purchase, the label carries your name and the date, and thirteen scents exist only in the cities that inspired them. The product is made for you, once, in front of you.

Both Aesop and Le Labo are now owned by conglomerates — L'Oréal acquired Aesop for $2.5 billion in 2023, Estée Lauder acquired Le Labo for approximately $60 million in 2014, a price that looks extraordinary in retrospect. In both cases the acquiring group made the same pledge: creative autonomy, no interference with the model.

The success of these brands did not go unnoticed. Typology, founded in Paris in 2019, absorbed the aesthetic language and philosophical positioning of Aesop — the apothecary style amber bottles, the ingredient transparency, the restrained typography, the less-is-more formulation logic — and built a brand at a more accessible price point. It is not exactly a copy, but a considered strategic interpretation of the same cultural idea: consumers want to know what is in their products, want fewer ingredients, want the object on their bathroom shelf to signal intelligence and care.

Below them, there are dupes. Lidl sells an amber-bottled hand wash at about €3 that is explicitly compared to Aesop's Resurrection formula in scent and packaging. The aesthetic and the bottle shape can be copied, even the scent profile can be approximated. What cannot be copied is the daily experience of using something that is genuinely exceptional. The formulation performs differently as the scent behaves on skin over hours in a way that a fraction of the cost cannot replicate. The transparency of the ingredient list treats the buyer as someone who understands what they are putting on their body. And beneath all of it, the story the buyer tells themselves — that they chose this, understood why, and will notice the difference every morning.

Disney found out its billion-dollar partnership was dead less than an hour before the public did. The entertainment company had signed a three-year deal with OpenAI, licensed hundreds of characters for use in Sora, and was planning a Disney+ integration — all of which collapsed when OpenAI announced it would shut the product down entirely. No enterprise tier was preserved, no controlled access for partners. The consumer app closes on April 26; the API goes dark in September. Sora is being fully discontinued.

The usual narrative frames this as a technology-readiness problem — generative video that was too expensive, too unstable, too legally exposed to scale. All true, but secondary. What killed Sora was an internal resource war. While a whole team inside OpenAI was focused on making Sora work, Anthropic's Claude Code was winning over the software engineers and enterprises that drive revenue. OpenAI's application chief Fidji Simo held an all-hands meeting telling staff the company was done with "side quests" and would optimise everything for productivity. Video generation was classified as one such distraction. The model didn't fail on its own merits — it was triaged out.

The product decisions accelerated the collapse. OpenAI didn't position Sora as a professional tool or an API for studios. Sora 2 integrated social media features, and multiple outlets noted it was overtly similar to TikTok. They tried to build a consumer social platform on top of one of the most compute-intensive models in existence — a category error that combined the worst economics of both formats. The model used copyrighted material by default unless rights holders actively opted out, which made any large-scale content partnership structurally precarious.

What makes the story more than an OpenAI anecdote is what happened next — or rather, what didn't. The market didn't freeze. Kling AI, Runway, and Vidu all saw user gains within the first week of the shutdown announcement. Runway raised $315 million at a $5.3 billion valuation weeks earlier. Kling shipped three major updates between January and March 2026 alone, now generates clips up to two or three minutes at a fraction of Sora's cost, and is growing fast in markets where volume matters more than prestige. Google is integrating Veo directly into Workspace and YouTube Studio — treating video generation as a feature inside existing products, not a standalone destination. The market has segmented into clear tiers: Runway for professional quality, Kling for cost efficiency, Veo for ecosystem integration, and open-source alternatives like Seedance for local deployment.

The pattern is telling. The competitors that survived aren't trying to be the next consumer sensation. They're building API layers and production integrations — Runway sells to ad agencies and film studios through a developer platform, Kling plugs into content pipelines where volume matters more than prestige, Veo sits inside Google Workspace and YouTube Studio. None of them launched a social app. They all sell generative video as a feature inside tools people already use, not as a standalone product competing for attention. Sora proved generative video could look like cinema. What it never figured out was that the market wanted a production tool, not a spectacle.