S2E18: ChatGPT is going to sell you therapy | Ethan Ding (TextQL)

The Enterprise AI Reckoning Has Arrived

The AI spending party is over. Ethan reveals that public companies went on unprecedented buying sprees in 2024, with procurement teams purchasing up to 500 different AI tools in a single year. Now comes the hangover - these same companies have instituted total bans on new AI vendors and are mandating 50% cuts before they'll even take another meeting.

"We bought 500 pieces of AI software in the past year. We have a total ban on new vendors whatsoever. We have to cut at least 250 of it before we even have conversations ever again."

The dirty secret? Nobody's actually using these tools. Ethan estimates 50% of enterprise AI initiatives have already failed, but companies won't admit it publicly. Teams churned the tools internally, but the invoices keep coming because admitting failure isn't an option when boards demanded "buy one of each" strategies.

"Nobody on our team used it. So that's like 50% of our AI initiatives down the drain. You never want to admit that your AI initiatives have failed."

Information Blindness Creates Billion-Dollar Moats

Here's the shocking truth about AI adoption: most users have no idea what they're actually using. Ethan drops a bombshell - while everyone knows ChatGPT, less than 10% of users understand that OpenAI powers it. This information blindness creates massive opportunities for vertical AI products.

"I think people underestimate how many companies or how many people there are, who if they use Harvey for lawyers, you might never find out what ChatGPT is. Less than 10% of them know what OpenAI is."

The implication is profound: if you're first to a niche with an AI solution, you might own that market for 4-5 years. Users develop "infinite loyalty" to their first AI tool because they never discover alternatives exist. It's like Nokia still having devoted users despite the iPhone - once you capture a market segment, information penetration is so weak that switching barely happens.

Data Science: The Infinite Arms Race

Unlike fixed workloads like accounting, data science has infinite demand because it's fundamentally competitive. Ethan uses a brilliant example: when Blackstone analyzes housing prices weekly by city, Cerberus counters by going daily by zip code. Then Blackstone responds with hourly analysis by square footage.

"If Blackstone analyzes housing prices per city per week, then Cerberus will want to analyze it per zip code per day. Then Blackstone's gonna want to do it per single family unit size square footage per hour."

This creates exponential demand growth - give trading firms 10x faster analysis, and they'll make 100x more trades because they can now pursue opportunities previously too small to bother with. TextQL's entire business model depends on this dynamic: as they reduce costs, volume explodes exponentially. It's why they're usage-based while competitors offering flat pricing are getting crushed by token costs.

Don't Innovate UI, Dominate Distribution

Every AI startup makes the same mistake: trying to innovate on user interface. Ethan's blunt assessment? Every single UX innovation TextQL attempted was "a horrible idea." The winning formula is surprisingly simple: copy ChatGPT's interface exactly (chat on left, workspace on right), then put all innovation into branding and distribution.

"Almost every single innovation we have ever tried to do with this company on UX has been a horrible idea. We always go back to the base. Your branding is entirely unrelated to your product."

The painful truth for engineers: the product doesn't matter, positioning does. Say you're "the AI agent for laundromats," give it a hard hat, and hammer that message repeatedly. The opportunity isn't in better AI - it's in reaching the people who don't use AI yet and saying "I built this for you." Marketing matters 10x more than the product in today's AI landscape.

OpenAI's $350 Billion Ad Platform Play

OpenAI's aggressive pricing on GPT-5 isn't about winning the API war - it's about building the world's most powerful commerce platform. Ethan paints a dystopian but likely future: you tell ChatGPT you're sad, and it recommends therapist Frederick Carlson, books the appointment, charges your credit card, and informs you it's out-of-network for $500.

"ChatGPT says, 'Well, you should consider talking to a licensed therapist, Mr. Frederick Carlson.' It's like, 'You want me to book a meeting for you right now? I've used your credit card to pay for this therapy.'"

Every early OpenAI and Anthropic demo featured "order me a pizza" as the use case. When ChatGPT becomes the layer between you and commerce, DoorDash and Uber Eats will pay massive fees to be the "preferred carrier." With potentially a billion users, OpenAI is sacrificing API profits to build something far more valuable: the transaction layer for AI-mediated commerce. As Ethan notes, ads are a $350 billion profit business, and "people just like being sold to."

The Trillion-Dollar Coin Flip Strategy

While everyone else fights over $10 billion exits, Ethan has a different philosophy: he's only interested in trillion-dollar opportunities. His target? AWS. His strategy? Be willing to flip coins at 51% odds repeatedly, as long as the expected value is massive.

"I'm basically willing to flip. I'm kind of more like SBF. I'm pretty happy to flip the coin with 51% chance over and over again, as long as I have high EV. I'm only interested in trillion-dollar market opportunities."

TextQL follows the Bezos doctrine: "your margin is my opportunity." They'll trade 1% of profit margin for 2% growth every time, because volume creates the ability to hire the best engineers, optimize infrastructure, and ultimately offer better prices than any competitor. It's the same playbook AWS used to become a trillion-dollar business - sacrifice margins early, dominate on volume, then own the entire market.

"I don't want to build a $10 billion business. That seems incredibly boring to me. I just want to go after AWS."

The AI industry is experiencing a massive correction. Enterprises are drowning in unused tools, OpenAI is building an ad empire disguised as a chatbot, and the real winners will be those who understand that in AI, distribution beats innovation, volume beats margins, and the first mover in a niche might own it forever. As Ethan says - "It's not the model, it's the marketing."

Companies Mentioned

• OpenAI
• Anthropic (Claude)
• Google (Gemini)
• Amazon AWS
• Microsoft
• Meta
• Netflix
• Spotify
• Cursor
• Windsurf (acquired/sold)
• Replit
• Lovable
• Bolt
• Claude Code
• Harvey
• TextQL
• Databricks
• Snowflake
• Cognizant
• Blackstone
• Cerberus
• Goldman Sachs
• PWC
• Nokia
• DoorDash
• Uber Eats
• Walmart

See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.