Codes of Culture | Issue 99
Luxury, AI, and the New Infrastructure
Welcome back to Codes of Culture. I’m Ashumi Sanghvi.
Writing this issue from Geneva, where I spent a spectacular 2 days at Watches and Wonders, discovering and being immersed with the most exquisite watchmakers in the world.
Something that caught my eye this week was the AI influencers that SheerLuxe and SheerLuxe Labs launched. 4 bots- each one with a distinct look, personality and expertise. So real that if the content wasn’t AI labelled, it would have been hard to know the difference between real and AI. Curious to hear your thoughts.
In this week’s issue, luxury AI campaigns are failing, Nvidia gives fashion a stage in Paris, Google puts VTO in search, and can scent be turned into data?
If you are new here, or want to catch up on the best of Codes of Culture, we’d recommend you start here and remember to subscribe for full access to our news, insights, podcast and global events.
📖In this issue:
Luxury AI campaigns are failing the same test: audiences are reading intent before output.
NVIDIA gave fashion a stage at GTC Paris.
Google folded virtual try-on into search, pushing fashion discovery into infrastructure.
Allbirds sold its footwear business for $39 million and pivoted to AI computing.
Turning scent into data. The molecular archive now has cultural stakes.
1. LUXURY MARKETING AND AI
Luxury AI campaigns are failing the same test: audiences are reading intent before output.
What’s happening: Business of Fashion’s April 2026 analysis of AI in luxury advertising points to a recurring pattern across campaigns from Gucci, Moncler, Valentino, and Prada: the audience response is shaped less by technical quality than by what viewers think the technology was there to do. Gucci was criticised for opacity. Prada and Valentino drew scrutiny for uncanny imagery. Moncler landed differently by framing AI as a way to create scenes that could not otherwise exist.
TLDR:
Gucci, Prada, and Valentino were penalised when AI was read as a shortcut rather than a creative choice.
Moncler was explicit about intent and framed AI as an imaginative expansion.
Many luxury marketers are misreading the audience.
Heritage houses carry higher stakes when they use synthetic imagery.
Transparency matters, but motive matters more.
Why it matters: Luxury still depends on the perception of care, skill, and intent. AI does not automatically undermine that. Cost-saving logic does. The issue is not whether a brand used AI. It is whether the audience believes the tool extended the work or reduced the effort behind it.
2. FASHION TECH AND THE CREATIVE STACK
NVIDIA gave fashion a stage at GTC Paris.
What’s happening: NVIDIA’s AI Art Gallery at GTC Paris featured a dedicated fashion chapter curated in partnership with the Fashion Innovation Agency at London College of Fashion. The projects included an AI catwalk generating runway environments at speed, generative campaign and lookbook experiments developed with students, and early work on ALIGN FASHION, an AI agent designed to support circular decision-making in fashion design.
TLDR:
NVIDIA is positioning fashion as a serious domain for compute-led creativity.
The AI catwalk compresses visual development from weeks into minutes.
ALIGN FASHION applies AI to design decisions, not just image generation.
Students are already training on these AI tools.
The real constraints now sit with institutions, curricula, and creative leadership.
Why it matters: Fashion’s AI story is moving away from campaign novelty and toward production logic, education, and decision-making. That makes the institutional layer more important than the software itself. The schools and labs that shape how designers work with these systems will also influence the industry’s aesthetic language.
3. DISTRIBUTION AND AI COMMERCE
Google folded virtual try-on into search, pushing fashion discovery into infrastructure.
What’s happening: Google’s experimental try-on app Doppl will close on 30 April 2026, with the technology moving into Google’s core products. Virtual try-on will now appear directly on product listings and apparel image search results, powered by Gemini 2.5 Flash Image. DressX’s Agent platform adds avatar generation, AI styling, and checkout across more than 200 luxury brands and one million products.
TLDR:
Doppl is closing because virtual try-on is moving into search itself.
Discovery, visualisation, and shopping are starting to sit in the same environment.
DressX is building the digital fashion asset layer beneath that shift.
Virtual try-on is becoming part of the distribution layer rather than a standalone feature.
Brands treating it as a differentiator are misreading the direction of travel.
Why it matters: Once try-on sits inside search, discoverability starts with product data, image readiness, and inventory structure rather than campaign spend alone. That shifts power away from the front-end brand moment and toward the systems feeding the interface. In fashion commerce, infrastructure is becoming the new storefront.
4. CAPITAL AND PRODUCT CORRECTION
Allbirds sold its footwear business for $39 million and pivoted to AI computing.
What’s happening: Allbirds, once valued at $4 billion at its 2021 IPO, sold its footwear brand and assets to American Exchange Group for $39 million in late March 2026. The remaining listed entity will pivot to AI compute infrastructure under the name NewBird AI and has secured a $50 million convertible financing facility. The company plans to acquire GPUs and lease access to a market where supply remains tight.
TLDR:
A $4 billion brand sold its core business for $39 million. The listed shell is now targeting GPU leasing.
The $50 million compute raise exceeds the proceeds from selling the footwear business.
The market thesis is clear: compute is scarce, and scarcity is commanding the premium.
The company is also seeking to remove its B Corp commitments from its charter.
The stock spike showed how quickly markets will reward an AI infrastructure pivot.
Why it matters: This is a sharp readout of how capital is being repriced. A values-led consumer brand now attracts less excitement than a listed vehicle with a route into computing. That does not say everything about long-term durability, but it says a great deal about the current market. Infrastructure is where the premium sits.
5. CULTURE AND THE SENSES
Turning scent into data. The molecular archive now has cultural stakes.
What’s happening: Artist and entrepreneur Siddhartha Kunti translates the molecular structure of fragrance into digital artworks, using gas chromatography data to render what the nose detects but the eye cannot see. His first major series analysed more than 800 whiskies and generated over 65,000 lines of molecular data, debuting at the Saatchi Gallery in 2025 alongside nanotechnology scent cards developed with Sissel Tolaas. His latest work, commissioned for Art Dubai Digital 2026, centres on Oud.
TLDR:
Kunti combines molecular analysis with data visualisation to produce digital sculptures of specific aromas, from whisky to Oud.
The Saatchi Gallery's debut paired digital art with scent cards. The scientific method is part of the work’s credibility.
The Art Dubai commission treats Oud as cultural material, not decorative reference.
Digital art now faces a higher credibility threshold than it did during the NFT cycle. Method matters more.
The institutional layer is starting to catch up.
Why it matters: Scent has long shaped memory, ritual, and brand identity, but it has rarely been treated as something that can be archived with precision. That is what makes this work more than an aesthetic experiment. Oud is one of the most culturally specific materials in luxury. Once it is encoded digitally, the question becomes who defines the archive and who holds authority over the record.