OpenAI has already succeeded in a part of this race that most companies never complete. About 900 million people use ChatGPT each week. For many, it represents AI.
As its popularity increases, so do its expenses. This is the unusual engine driving the most successful product in software history.
OpenAI does not have a clear route to profit until 2030, while Anthropic expects to be profitable this quarter.
So the real question is not whether OpenAI is winning but whether it can afford its users.
Model breakdown #003 OpenAI Snapshot
01. The Tech
OpenAI began with a chatbot. It is now creating a hub where people can get work done.
ChatGPT is being integrated into the tools people already use, like spreadsheets, health apps, and bank accounts. It now has memory that carries over from one conversation to the next. Codex manages coding and is branching out into other jobs. Outside developers can now create apps that function within ChatGPT.
The change is happening faster than many realize. Codex, its coding tool, has reached over 5 million weekly users, which is a sixfold increase since February. Knowledge workers, not just engineers, make up the fastest-growing group of users.
The strategy is straightforward. If OpenAI controls the entry point for consumers, developers, and companies, it doesn't need the single best model. It simply needs to be where you start.
02. The Business Model
This is where the numbers get uncomfortable.
OpenAI's gross margin is about 33%, while healthy software usually has margins of 70% or more. The reason is straightforward: every answer, image, and agent task relies on expensive compute, leads to rising costs with increased usage. Inference is expected to grow from $8.4 billion in 2025 to $14.1 billion this year.
To improve the situation, OpenAI is expanding its revenue sources. Enterprise now makes up over 40% of revenue and is catching up to consumer sales. The company has also started selling ads. It raised $122 billion, the largest funding round ever, and it has committed over $500 billion to computing power, most of which it rents from competitors.
Even within the company, the pressure is evident. The Wall Street Journal reports that OpenAI recently fell short of its user and revenue goals. Its CFO has questioned whether the company can continue funding the necessary data centers.
OpenAI has the demand that nearly every company dreams of. What it lacks, for now, is a business that generates revenue from meeting that demand.
03. The Real MOAT
OpenAI's protection is not its AI models. Competitors can now match it within months. Infactm Anthropic is making better models. One of its co-founders, Andrej Karpathy, left for Anthropic in May.
The true advantage is harder to replicate: it was the first to enter the market. It is the go-to app for hundreds of millions and the standard tool in a million companies.
However, brand loyalty in AI is not as strong as it seems. OpenAI's position is real, but it is not dependable for power users.
Its user base tends to be younger and more casual. About 73% of chats are personal, not work-related, and nearly half of the users are under 25. This loyalty won’t last unless OpenAI continues to invest to maintain its lead, which brings us back to the issue of funding.
MY TAKE
Here's what I think is really happening. Beneath the discussions about AGI and the buzz around the IPO, OpenAI has realized it can't succeed based solely on model quality. It is now reaching out to every audience it can end consumers, developers, enterprises, and now advertisers. This is a smart strategy because an audience of 900 million people is something almost no competitor can create from scratch.
However, this approach does not address the finances. The $852 billion valuation overlooks this issue, which is why some analysts believe it's overvalued. OpenAI is hoping that scaling will eventually make the service affordable enough to generate profit, but that's still just a hope, not a solid plan. Moving forward, it will likely result in one of two scenarios: the service becomes too integrated to remove, or it turns into the largest tech company that never made a profit.
04. ONE THING TO WATCH
Whether OpenAI's gross margin improves as it grows. If costs keep rising as fast as revenue, no number of users will make this profitable.
That's the teardown.
Reply and tell me your feedback and which AI startup you want torn apart next. I read every one.
- Abhishek
P.S. If you found this useful, forward it to a founder who's tired of the hype.