The Landlord Always Wins
What YouTube, Shopify, and the App Store tell us about who will actually win the AI economy.
Tenants Do.
There is an iron law at work in every platform economy, and almost nobody talks about it clearly.
I want to name it.
Call it the Tenant Rule: for any platform to survive, the people building on top of it must collectively generate more economic value than the platform captures.
By "value" here, I mean gross throughput — GMV, billings, sales — not platform profit. This is not a theory. It is pure logic.
The landlord would not maintain the building if the sum of rents didn't cover the cost of the capital deployed to build it. And the tenants would not stay if they weren't generating more from the space than what they pay to occupy it.
Both sides must be true simultaneously.
Which means the builders — in aggregate — must always be generating more gross value than the platform captures. Always.
When you look at the numbers, the pattern is hard to ignore.
In 2025, Alphabet reported that YouTube revenue across ads and subscriptions exceeded $60 billion. Netflix reported $45.183 billion in total revenue for the same year. YouTube crossed Netflix. Quietly.
Independent market research estimates the global creator economy — across platforms and revenue streams — at roughly $250 billion in 2025. The total creator ecosystem is multiple times larger than any single platform's revenue line. The key point isn't that YouTube "caused" all of that value, but that once a platform becomes infrastructure, its own revenue is usually just a slice of a much larger downstream economy.
And YouTube's own economic impact report, produced with Oxford Economics, found that its creator ecosystem contributed over $55 billion to U.S. GDP in 2024 — supporting more than 490,000 full-time equivalent jobs in editing, production, brand partnerships, analytics, and merchandise that didn't exist as industries when the platform launched. The platform built the rails. The builders built the economy around the rails.
It's not just YouTube:
→ Apple App Store: In 2024, Apple-supported research estimates the App Store ecosystem facilitated approximately $1.3 trillion in billings and sales. Apple collected no commission on more than 90% of that total. The builders' economy is vastly larger than the platform's revenue line.
→ Shopify: In 2025, Shopify merchants processed $378.4 billion in GMV while Shopify booked $11.556 billion in revenue. The merchants' economy is approximately 33× the platform's own revenue.
→ Amazon Marketplace: Amazon doesn't publish marketplace GMV directly. What it does disclose is that in 2025 it reported $172.162 billion in third-party seller services and $68.635 billion in advertising services — both tied directly to seller ecosystem activity. The seller ecosystem generating those fees is, by definition, substantially larger than the fees themselves.
→ Substack: Writers have earned roughly an order of magnitude more in aggregate payouts than Substack has kept in fees. A single newsletter platform created dozens of million-dollar-a-year individual media businesses.
Across every major platform ecosystem with public data, the pattern is consistent: the builders' aggregate economy is substantially larger than the platform's own revenue. The Tenant Rule holds.
We've watched this movie twice. We missed it both times.
In 2005, YouTube launched.
Most of us thought it was a place to watch funny videos. We watched our friends start uploading content and thought it was a phase — something silly people did. We had no framework for understanding that the site was not a product.
It was infrastructure.
By the time we understood, MrBeast had built a media conglomerate valued at $5 billion. Kids' channels had been acquired for $3 billion. And around all of them, that 490,000-job ecosystem had quietly materialized — editors, agencies, merchandise operations, analytics tools, talent managers — entire industries built on top of a platform we'd dismissed as a video website.
Then the iPhone arrived in 2007. Same story.
We debated whether apps would replace websites. We asked whether mobile would hurt PC sales. Almost nobody predicted ride-sharing as a default transport layer, food delivery as a global industry, or mobile-first banking as a viable business model. The conversations were about the technology. The wealth was being created in the ecosystem.
We are at that moment again.
Except this time, we have the benefit of having seen it happen twice.
AI is not a chatbot. It's infrastructure.
The most important mental reframe available right now:
AI is not a product you use. It is infrastructure on which an entirely new economy will be built.
The conversations we're currently having about AI sound exactly like 2005 and 2007: ❌ "Will it replace copywriters?" ❌ "Will it automate analysts?" ❌ "Is it accurate enough to trust?" Those are the wrong questions — not because they're unimportant, but because they're the questions people ask when they're staring at the technology instead of the ecosystem it will generate.
The right questions are:
✅ What new job archetypes will exist that we don't have language for yet? ✅ What businesses become viable for the first time when a solo founder can operate what looks like a 20-person firm? ✅ What markets open when anyone can orchestrate AI agents as easily as a creator today uploads a video?
Every platform goes through three phases Phase 1 — The App Phase We debate features, benchmarks, accuracy. We compare it to what came before. We ask if it's "good enough." (YouTube: 2005–2008. AI: right now.) Phase 2 — The Infrastructure Phase We realize it's a platform, not a product. New categories of work become possible and cheap. Early movers build foundational positions. (YouTube: 2009–2014) Phase 3 — The Ecosystem Phase The real value shows up around the platform: new careers, companies, and communities that were impossible before. The tenant economy dwarfs the landlord. (YouTube: 2015–now) YouTube is in Phase 3. The smartphone ecosystem is deep in Phase 3. AI is barely out of Phase 1.
That gap — between where we are in the conversation and where the real value will eventually sit — is the opportunity.
Why did we miss it last time?
This is the question worth sitting with.
We had the internet. We had smartphones. We had YouTube. We watched creators build empires in real time. We had every piece of information we needed.
And most of us still stood on the sidewalk.
It was not ignorance. The tools were accessible. The economics were demonstrably in the builders' favour.
Most people who missed the YouTube wave were not uninformed.
They were embarrassed.
Making videos felt silly. Starting a channel felt narcissistic. Building a personal brand felt like something other people did. That psychological barrier — the friction of feeling foolish — is the real reason most people didn't participate in the last platform economy. And that same barrier is active right now with AI.
Building with AI feels technical. Automating your work feels premature. Starting something feels like getting ahead of yourself.
It's the same costume on the same instinct.
The embarrassment is the signal, not the warning.
What the AI ecosystem will generate that doesn't exist yet Based on the pattern from YouTube and the app economy:
New job archetypes with no names yet. "YouTuber" would have sounded absurd in 2004. "Influencer" would have sounded like parody. In five years, roles like agent orchestrator, AI workflow designer, and synthetic studio lead will be normal professional titles — and people who moved early will be the senior practitioners commanding the highest rates, exactly as the first serious YouTubers became the highest-paid talent on the platform. The one-person company that looks like a firm. A complete AI operating stack in 2026 costs between $3,000 and $12,000 per year — a 95–98% reduction in operating costs compared to building a human team to do equivalent work. In many domains, a solo founder can already run what looks, from the outside, like a 20-person professional services business. This isn't theoretical. It's happening now at the margins. In five years it will be the default model for a certain class of entrepreneur.
The second-order economy. The majority of jobs created by AI won't be "AI jobs" in any obvious sense. Just as YouTube created brand deal lawyers, thumbnail designers, and channel management agencies — entire professions that sound absurd until they're everywhere — AI will create its own surrounding trades. Trust verifiers.
Human-in-the-loop editors. Agent workflow consultants. Ethics auditors for automated systems. The ecosystem economy, not the model economy, is where the volume of opportunity will sit.
The keys are on the table.
OpenAI, Anthropic, Google — they are the new Landlords. They are collectively pouring hundreds of billions of dollars per year into infrastructure. And critically: they are competing with each other for your tenancy. That competition is a structural gift to builders — it means the unit cost of intelligence, of inference, of the fundamental raw material of AI-powered businesses, has been driven sharply down and will continue to fall.
The Stanford AI Index 2025 found that the cost of running AI inference has dropped by more than 280× in recent years. The landlords are subsidizing your rent in order to win your occupancy.
The tenant economy around AI will, by the iron logic of platform economics, generate more aggregate gross value than the platforms themselves.
This is not speculation. It has been true in every major platform economy we have data on. It will be true here.
The only question is whether you will be in the ecosystem when it enters Phase 3 — or whether you will be standing on the sidewalk, remembering the moment you watched it happen to someone else.
We had this choice in 2005. We had it in 2007. Most of us chose the sidewalk — not because we lacked information, but because building felt premature, strange, or foolish. The feeling of foolishness is the entry point. It always has been.
The people who walked through it built the creator economy.
The people who walk through it now will build the AI economy.
And if history holds — and it always holds — the tenants will, once again, collectively generate more than the landlord ever could alone.
What are you building?
Sources 1. Alphabet FY2025 earnings release (YouTube >$60B):
https://s206.q4cdn.com/479360582/files/doc_financials/2025/q4/2025q4-alpha bet-earnings-release.pdf 2. Netflix FY2025 annual report:
https://www.sec.gov/Archives/edgar/data/1065280/000106528026000034/nflx- 20251231.htm 3. Creator economy market size (~$250B):
https://www.grandviewresearch.com/industry-analysis/creator-economy-market- report 4. YouTube/Oxford Economics U.S. impact report ($55B GDP / 490K jobs):
https://services.google.com/fh/files/misc/us_full_report.pdf 5. Apple App Store ecosystem report (~$1.3T, >90% commission-free):
https://www.apple.com/newsroom/pdfs/2024-Apple-Global-Ecosystem-Report-J une2025.pdf 6. Shopify FY2025 financial results:
https://s27.q4cdn.com/572064924/files/doc_financials/2025/q4/Shopify_Invest or_Press_Release_Q4-25_FINAL.pdf 7. Amazon FY2025 annual report (3P seller services + advertising):
https://www.sec.gov/Archives/edgar/data/1018724/000101872426000004/amz n-20251231.htm 8. Stanford AI Index 2025 (inference cost down 280×+):
https://hai.stanford.edu/ai-index/2025-ai-index-report #ArtificialIntelligence #CreatorEconomy #Entrepreneurship #FutureOfWork #PlatformEconomics