Don't end up saying "Shoot, I messed up. I didn't see this coming."
Published October 10, 2025 • Based on Founder Reality Episode 28
Also available on: Apple Podcasts • Spotify • YouTube
This week, three things happened that look completely unrelated:
- FICO changed their licensing model and two public companies lost 10-20% of their market value in a single day
- ChatGPT launched shopping inside their interface with Shopify
- European AI founders are fleeing to San Francisco at record rates
But they're all the same story: Infrastructure owners are systematically eliminating middlemen to reclaim their margins.
And if you're building a company that depends on being the channel between infrastructure and customers, you're about to get squeezed out.
Let me show you exactly what's happening and what you need to do about it.
Story 1: How FICO Destroyed $5 Billion in Market Cap
FICO just launched something called the "Mortgage Direct License Program." The name sounds boring. The market reaction was brutal.
Equifax stock dropped 8.4% in a single trading session. TransUnion crashed 10.1%.
Here's what FICO actually did:
For decades, they licensed their credit scoring algorithms to credit bureaus (Equifax, TransUnion, Experian). Those bureaus would compile data, calculate FICO scores, and resell them to lenders.
The pricing was insane:
- FICO charges bureaus ~$5 per score
- Bureaus turn around and charge lenders $10 per score
- That's a 100% markup just for being the middleman
Those companies made billions being gatekeepers between FICO and mortgage lenders, personal loan companies, credit card issuers.
FICO finally had enough. They said: "We're going direct to lenders. We're cutting you out."
New price: $4.95 per score. A 50% reduction from what bureaus were charging.
Why This Matters More Than You Think
When I was building SimpleDirect Financing, we dealt with this exact ecosystem. We didn't run credits ourselves - our lending partners did. They paid those credit bureaus $10 per check.
I always wondered: Why does FICO let these middlemen take half their potential revenue?
Now I know the answer: They were building leverage. Waiting for the right moment.
The US mortgage industry is $12 trillion. FICO owns the IP, the scoring model, the algorithms. The entire industry relies on their specific model.
The credit bureaus don't actually own anything valuable. Sure, they have brand recognition. People say "check my TransUnion score" or "my Equifax report."
But at the end of the day, it's FICO presenting the algorithms and data. The bureaus are just gatekeepers with smart markups.
They're like Coca-Cola selling through distributors - except FICO just decided to bypass the distributors and sell direct.
Why I Had to Shut Down SimpleDirect Financing
A few episodes ago, I announced we're sunsetting SimpleDirect Financing by the end of 2025.
People asked: "George, why give up a product you spent four years building?"
The answer is simple: We didn't own the end-to-end experience.
We were the middleman. We sent data to lenders who made the final lending decisions. Sometimes it was instant, sometimes it took 1-2 days.
We had merchants asking us what's going on. Homeowners asking us what's going on. We were stuck connecting information in silos, dealing with problems that weren't our fault.
As volume increased, I realized: We created a silo business. We're never going to make customers truly happy. We're essentially just a volume-sending machine to lending partners.
If you don't own the IP, if you're just reselling someone else's API, you're always at risk.
Think about Credit Karma. They don't work with FICO directly - they work with Equifax and TransUnion. Those agencies can increase prices, make it harder to access data, introduce requirements you're not ready for.
You're always at the mercy of other people's decisions.
Story 2: ChatGPT and Shopify Just Killed Google Ads
A few days ago, OpenAI and Shopify announced instant checkout inside ChatGPT.
Now when you ask ChatGPT for product recommendations - "I need cool sunglasses for my trip to Europe" - it can instantly recommend products from Shopify merchants.
Here's the magic: You can check out right there in the chat. You don't leave the conversation, don't go to the merchant's website, don't create accounts, don't re-enter credit card info.
Shopify stock jumped 6% on the news. They're connecting 1 million+ merchants to ChatGPT's 700 million weekly active users.
They're calling it "agent commerce" - AI agents acting as personal shoppers, autonomously handling the entire purchase flow.
The Amazon vs Shopify Problem Just Got Solved
I've tried e-commerce before. The benefit of Amazon is traffic - it's one of the most visited websites in the world. Even if you're selling a small product on page six, you still get sales.
The problem with Shopify has always been: You own the tech, but you have to handle marketing yourself.
You don't give Amazon 20-30% of revenue, but you're responsible for driving all your own traffic. For most merchants, that's really hard.
Shopify just solved this. They're giving merchants access to 700 million weekly active ChatGPT users without Google Ads, Facebook Ads, or Amazon fees.
Who loses? Google Ads. Facebook Ads. Amazon marketplace fees. All the middlemen taking cuts from merchant revenue.
But Here's the Real Story: AI Needs to Monetize
Every major AI model offers a free tier - ChatGPT, Claude, Gemini, Grok. Training these models costs hundreds of millions to billions of dollars. Running them costs money on every single message.
I always ask: How do these companies subsidize this? The answer has been: "VC money. Saudi Arabia, UAE, Qatar pouring billions in."
But that can't last forever.
Elon Musk's xAI just announced they're adding advertisements in Grok. ChatGPT is testing commerce. They're all desperately trying to monetize.
I pay Claude $20/month and use it heavily for coding. I'm a daily ChatGPT user too - probably a dozen times per day for advanced voice mode, research, learning.
I'm only paying $20/month for ChatGPT. Honestly, the value I get is worth way more. But they're not raising prices yet because of market competition.
A founder friend I talked to recently is building a company that works with AI platforms and advertisers - helping inject ads directly into LLM responses. They're early, frontier stage, working with smaller AI companies desperate to monetize.
But ChatGPT, Claude, Gemini, xAI? They're figuring this out too. And unlike my friend's startup, they have direct distribution advantage.
They can cut out the middleman and work directly with advertisers.
Story 3: European Founders Are Fleeing to America
Wall Street Journal published an article last week: US investors now dominate European AI funding rounds.
American investors account for 71% of AI deals in Europe by value this year. Up from 57% last year.
The kicker: European founders aren't even staying in Europe. They're doing "Delaware flips" - living in Europe but setting up US holding companies just to access American capital.
One founder in the article said it perfectly: "In London, we'd get half our valuation. In Silicon Valley, investors understand market potential and let companies grow aggressively."
Another raised $500K in SF within a week of landing. He'd been trying for months in London.
Why Infrastructure Wins
This is the same pattern again: European VCs are the middlemen getting cut out.
US investors - Sequoia, a16z, Y Combinator - have built infrastructure not just for US founders, but for global founders. They write bigger checks. They move faster. Sometimes they commit capital within days of meeting.
European VCs are slower, more conservative, write smaller checks. Just like credit bureaus in story one and Google Ads in story two, they're being bypassed.
But it's not just about money. It's about speed.
In AI, you need speed to move past competitors. European funds take months. American funds take weeks.
I've Been on Both Sides of This
I'm from Canada, so I understand this problem intimately. I run companies in both Canada and the US.
Canadian VCs don't lead rounds. They only write support checks. They tell you: "Go to the US to find a lead investor, then we'll follow."
I always think: Why would I need you? If I can find a US lead investor, I'll just find supporting investors there who'll treat me better.
Three years ago in 2022, I attempted to raise a VC round. My lead investor was from San Francisco. My secondary lead was from New York.
Those guys had money. They understood. They were risk-takers. And that wasn't even their day job - they run their own companies and write angel checks on the side. Quarter million each. Pretty generous.
Meanwhile, Canadian VCs kept asking: "Do you have American investors? Do you have a lead?" Because I knew from experience - their LPs tell them they can't write checks unless a lead investor is found first.
Canadian VCs, like European VCs, are becoming middlemen getting cut out.
For successful founders, there's no reason to stay in those countries if you want to raise venture capital, especially in AI where speed is everything.
The Pattern Is Clear
In every layer of the stack - data, commerce, capital - infrastructure owners are killing middlemen.
Story 1 (FICO): Companies that own IP eliminate the markup. Bureaus don't own the score, so they got cut out.
Story 2 (ChatGPT/Shopify): Commerce infrastructure owner eliminates website friction. You no longer need a website to sell. LLMs become the new storefront.
Story 3 (US vs EU capital): Funding infrastructure eliminates slow, inefficient middlemen. Founders flee to the best infrastructure.
What This Means for Your Startup
The middle ground is dead.
You cannot be successful just being a marked-up reseller. You cannot be a slow local fund. You cannot be just a website trying to capture traffic.
LLMs, AI, infrastructure owners, IP owners - they're all going to squeeze you out.
Your Two Options
Option 1: Be the Infrastructure Owner
Own the IP. Own the algorithm. Own the core value creation.
Be FICO, don't be the credit bureau.
This doesn't take years or decades. You can start now by owning the end-to-end experience. Don't leave any core part of your customer experience entirely to another API or provider.
That's slow suicide. It'll get your business stuck eventually.
Option 2: Be So Lean You Don't Need Middlemen
Bootstrap as much as you can. Build global teams - India, Middle East, South America, North America. Optimize for efficiency.
This way you don't rely on gatekeepers, because gatekeepers are getting eliminated.
What You Cannot Be
You cannot build a business that depends on being the channel between infrastructure and customers.
I personally had to make this decision very recently about whether to go into a business like that. I chose not to.
Even though it's going well right now, I see myself getting blocked again by infrastructure partners eventually. They have the incentive to squeeze you out.
Maybe not now when they're early. But eventually they will.
I don't think about it as "if." I think about it as "when."
The Three Questions You Need to Answer
Question 1: Do you own the IP and algorithm, or are you reselling someone else's?
This happens more than you think. It's okay if your answer reveals vulnerability, but if you're white-labeling or charging a markup, you need to think about mitigation.
AI is reducing the risk and cost for IP providers to go direct.
Question 2: Are you a destination, or can you be eliminated by a better interface?
If you're a website selling products and ChatGPT launches a plugin for your category, you're going to lose volume. People use ChatGPT as their Google now.
If you're not using the dominant platforms (like Shopify in this example), your traffic will drop because buyers will purchase directly from integrated merchants.
If you're just a website that can be replaced by a ChatGPT plugin, you need to own the data layer and customer relationship in a defensible way.
Question 3: Are you structured for speed or structured for slow?
Delaware entity, global team? Or stuck in one geography?
If you're only accessible to local slow capital - even if you're in the US but not in Silicon Valley or New York - you're at a disadvantage.
European startups successfully raised capital by getting close to American investors. Is that something you can do?
If you're bootstrapping, can you lower costs so significantly that it doesn't matter if something goes wrong tomorrow?
My Hard-Earned Lesson
At SimpleDirect and ANC, I learned the hardship of not owning end-to-end experience the hard way.
It's extremely, extremely hard. You don't understand it at the moment, but 3-4-5 years down the road, you realize: "Shoot, I messed up. I didn't see this coming."
That's exactly what happened to me with SimpleDirect Financing.
It looked so easy when I started. I thought: "The biggest companies in the world do this, so it must work."
But the core truth is: They don't actually figure it out either. They just have enough leverage to survive being the middleman.
You probably don't.
The 2025 Playbook
Own the infrastructure. Optimize for geography. Eliminate friction.
The era of cutting out the middleman is happening right now.
Make sure you own:
- The infrastructure
- The IP
- The customer relationship
- The algorithm
Don't be the middleman.
Are you a middleman in your business model? Take the three-question test and email me at george@founderreality.com - I'm curious how many founders realize they're vulnerable.
Share this with a founder who needs to hear it. The middleman extinction is accelerating.
New episodes Monday/Wednesday/Friday at 9am EST. Real founder lessons about surviving infrastructure consolidation.
Daily thoughts: @TheGeorgePu on Twitter/X
Full episodes: founderreality.com
