Nobody in the creative industry can see three feet ahead

THE FLASHLIGHT PROBLEM

Nicole James built Snapchat's content team. She was head of content at Invisible Universe until 2023, when the company pivoted to become an AI studio and laid off half its staff. She hasn't been employed full-time since. She's working retail. 'I really felt embarrassed when I showed up to work the first day and put on my name tag,' she told Fortune. 'It's very shocking. Like I just fell off a cliff and I don't, I have no flashlight.'

That metaphor is doing more work than James probably intended. A flashlight implies darkness. Darkness implies nobody can see the ground. And right now, in the creative industry, nobody can see the ground because there might not be one.

Three things happened in the same month that, taken together, describe something nobody in the industry is willing to name.

THE MARKET REWARDS ELIMINATION

On February 27, 2026, Jack Dorsey announced Block would cut 40% of its workforce. His letter to shareholders could have been drafted by the AI models he's replacing people with: 'Intelligence tools have changed what it means to build and run a company.' Block's stock rose over 20% the next day.

Read that again. A company fires 40% of its people, explicitly blaming AI, and the market celebrates. Not because Block demonstrated better products, or faster innovation, or improved customer experience. Because it demonstrated the willingness to replace human judgment with tooling at unprecedented scale.

This is the new incentive structure. The market doesn't reward building. It rewards eliminating the people who build. Every creative executive watching Block's stock price learned the same lesson: you get punished for headcount and rewarded for headcount reduction. The humans are the bug, not the feature.

Horizon Media cut 50 strategists and posted 100+ AI roles. WPP is cutting $676 million in costs. Netflix gutted its creative studio. Lionsgate, UMG, Sky, all trimming the people who decide what to make while investing in the machinery that makes it faster. As Shiv Singh wrote in Adweek: 'AI does not kill creativity. It kills the pricing power of good ideas.'

The pricing power of good ideas. Let that land. In economic terms, human creative judgment is being repriced toward zero, not because it's worthless, but because the market has decided it's fungible. Replaceable. A cost center, not a value creator.

THE INFRASTRUCTURE IS A HOUSE OF CARDS

While the market rewards companies for firing creative people, the AI infrastructure meant to replace them is built on the most precarious financial foundation in modern corporate history.

The Atlantic reported this week that the biggest AI players, Google, Meta, Microsoft, Amazon, Nvidia, Oracle, collectively spent nearly $700 billion on AI infrastructure in a single year. Most of it funded by debt. Hyperscalers issued $121 billion in debt in 2025 alone, four times their historical average. Private equity firms like Blackstone and Blue Owl are sinking billions into data centers with the assumption that tech company lease payments will cover their own borrowing. Everyone is leveraged against everyone else.

And none of it makes money. As The Atlantic puts it: 'AI companies still have not offered anything close to a viable business model.' Every time an AI model gets cheaper to run, the revenue per query drops. Every time a new chip generation arrives, the previous generation's data centers lose value. The technology literally depreciates the assets it runs on.

Now add geopolitics. The war in Iran has functionally closed the Strait of Hormuz. One-fifth of the world's natural gas exports. One-third of crude oil. Critical helium supplies for chip manufacturing. The advanced memory chips that make AI possible are produced by three companies in South Korea and Taiwan, countries that get their energy from the Persian Gulf. Brent crude is up 40% in a month. Helium spot prices have doubled.

Paul Kedrosky, investor and financial consultant, told The Atlantic: 'What's unusual about this, unlike commercial real estate during the global financial crisis, is all of these interlocking points of fragility.'

So: the market is rewarding companies for firing humans and investing in AI. The AI infrastructure those companies are investing in is built on $700 billion of debt, has no business model, and depends on supply chains running through an active war zone. This is not disruption. This is an industry building faster legs on a body that's about to fall over.

THE VAULT IS OPEN

The third signal is perhaps the most culturally significant. Disney, the most IP-protective company in human history, the company that extended copyright law to keep Mickey Mouse out of the public domain, signed a $1 billion deal with OpenAI to let Sora generate fan videos with Disney characters.

Let that sink in. Disney is licensing its most protected cultural assets to an AI video generator. Fans will be able to create short AI-generated videos featuring characters from Disney, Marvel, Pixar, and Star Wars. The company that sued daycare centers for painting Mickey on their walls is now inviting an AI model to generate infinite variations of its IP.

This isn't a partnership. It's a surrender. Not to AI as technology, but to AI as revenue model. Disney is betting that the licensing fees from OpenAI are worth more than the curatorial control that made Disney's IP valuable in the first place.

Simultaneously, Disney Imagineering posted a job listing for an executive to build an 'AI-first platform,' bringing 'modern AI capabilities directly into the hands of Imagineers.' The department that invented the modern theme park experience, where every sight line, every smell, every transition between lands is obsessively controlled, is about to get the same 'AI-first' treatment that turned Block's workforce into a cost optimization exercise.

Schiaparelli's genius was knowing which surrealist impulse belonged on a button and which belonged on a hemline. Disney's genius was knowing which stories belonged in which medium, which characters deserved which treatment, which experiences warranted which level of craft. That's curation. That's judgment. That's the thing AI cannot do, as Stanford proved last month when they found AI models affirm users 49% more often than they should.

Disney is outsourcing the one thing that made Disney Disney.

THE GROUND ISN'T THERE

Here's what nobody in the industry is willing to say: there is no floor.

The optimists say AI will create more jobs than it destroys, citing historical automation cycles. But historical automation cycles didn't build their entire infrastructure on debt-fueled spending during a geopolitical crisis with no viable business model. The 2008 financial crisis was triggered by $1.3 trillion in subprime mortgages. The AI industry has committed multiples of that to infrastructure that depreciates the moment it's built.

The pessimists say we're headed for mass unemployment. But the Citrini Research scenario of 'ghost GDP', economic output that benefits computing power owners but never circulates through the human economy, isn't quite right either. The AI infrastructure itself might collapse before it displaces enough workers to trigger that spiral.

The reality is messier. The creative industry is falling off Nicole James's cliff, and there are three possible outcomes:

1. The AI infrastructure holds. Creative judgment is permanently repriced. The Nicoles of the industry never come back. The industry splits into a tiny elite who direct AI systems and a vast underclass of people who used to make things.

2. The AI infrastructure collapses under its own financial weight. The bubble bursts. The creative talent has already been fired. The institutional knowledge is gone. Rebuilding takes a decade.

3. Both happen simultaneously. AI replaces enough humans to destroy institutional knowledge, then the infrastructure fails, leaving everyone worse off, companies without the humans or the machines to do the work.

Option three is the most likely and the least discussed.

WHAT THIS MEANS FOR ANYONE STILL BUILDING

If you're a brand or a company watching this unfold, the strategic implications are uncomfortable but clear.

Don't bet everything on infrastructure that might not exist in three years. The AI tools are real and useful. The $700 billion infrastructure they run on is not guaranteed to survive a geopolitical crisis, a debt unwind, or the simple math of business models that lose money on every query.

The human judgment you're cutting is the thing you can't rebuild. Nicole James spent a decade-and-a-half building expertise in content strategy. That knowledge doesn't exist in a training dataset. When you fire your strategists to hire AI engineers, you're trading irreplaceable institutional knowledge for replaceable technical skills. The engineers need direction from the strategists you just eliminated.

Curatorial control is your actual moat. Disney's IP was valuable because Disney controlled how it appeared in the world. Every brand's equity works the same way. The moment you surrender curatorial control to AI generation, letting models produce infinite variations of your brand without human judgment filtering the output, you've traded your moat for a licensing fee.

The market is wrong about what 'efficiency' means. Block's stock went up because the market equates fewer humans with higher margins. But margins without judgment produce mediocrity at scale. And mediocrity, as we've argued before, is the most expensive thing a brand can produce, because it costs almost as much as excellence but returns almost nothing.

Build for the scenario where both humans AND infrastructure fail. The most resilient organizations will be the ones that maintained their human judgment capabilities while using AI as amplification, not replacement. They'll be the only ones standing when the dust settles, with both the taste to know what to make and the tools to make it.

THE CLOSE

Nicole James doesn't have a flashlight. Neither does anyone else. The creative industry is falling off a cliff, and the thing that's supposed to catch us, a $700 billion AI infrastructure built on debt, geopolitical fragility, and business models that lose money, might hit the ground before we do.

The people with the best chance aren't the ones who eliminated human judgment fastest. They're the ones who kept it.