The AI Prices You're Paying Today Aren't Built to Last

OpenAI filed for its IPO on Friday. The company filed confidentially with the SEC, targeting a valuation between $852 billion and $1 trillion and a public listing as soon as September.

The number that matters most in that filing isn't the valuation. It's this: OpenAI loses $1.22 for every dollar it earns.

In Q1 2026, with revenue running at roughly $25 billion annually and 50 million subscribers, the company is still burning more than it brings in. It projects a $14 billion operating loss for the year and plans to spend $600 billion over the next five years on compute infrastructure. It doesn't expect to become profitable until around 2030.

This isn't unusual for a pre-IPO tech company. Amazon lost money for years. Netflix burned cash for nearly a decade before tightening margins once it had scale. The pattern is familiar: spend aggressively to build the infrastructure, capture the market, then raise prices from a position of dominance.

What's less familiar is how directly this affects the tools your business is probably already using.

The $20/month you pay for ChatGPT Plus is not covering its share of the infrastructure costs. Neither is the $30 or $100 plan. The gap is being closed by investors who are collectively betting that once AI becomes indispensable, OpenAI can raise prices from a position of strength. That's a reasonable bet. It's also one that comes due eventually.

Going public is where the clock starts. Once shareholders own a piece of the company, quarterly earnings calls replace private investor patience. Analysts will push on the timeline to profitability. The company faces a clear choice: find a path to margins by expanding revenue, cutting costs, or both. Given that its core product is getting more capable and more expensive to run at the same time, the most obvious lever is pricing.

The practical implication is straightforward. The AI tools you're using today are probably cheaper than they will be in two years. Not necessarily by a lot. But the era where the cost of training and running these models is being absorbed by private capital has a natural end date. OpenAI made it visible this week by filing the paperwork.

The response to this isn't to panic about costs or switch everything preemptively. It's to spend a few honest minutes answering one question: if the AI tools your business uses doubled in price tomorrow, which ones would you keep?

Not which ones you'd keep out of habit. Which ones actually earn their cost. Which ones free up real time, reduce real errors, or let you do something you genuinely couldn't do before. That is a different list for most businesses than the one they're currently paying for.

The tools that survive price increases are the ones where the value is specific enough to be obvious. AI drafting that cuts proposal time from four hours to forty minutes. Customer response workflows that let one person handle what used to take three. Automated reporting that shows up every Monday without anyone building it. These survive because the ROI math is clear to whoever writes the check.

The tools that don't survive are the ones you pay for because they seemed like the right move, or because everyone else seemed to be using them, or because they did something impressive in a demo. Those look different at twice the price.

One more detail from the filing worth noting: OpenAI projects it will generate $2.5 billion in advertising revenue in 2026, scaling to $11 billion by 2027. If you've been using AI tools with the assumption that your interactions are solely for improving the model, the ad revenue projection is worth sitting with.

When the S-1 becomes public ahead of the listing, you'll be able to read the actual risk factors, the revenue breakdown by segment, and the customer concentration data. That level of transparency hasn't existed before. Private AI companies don't have to disclose any of it.

What IPO filings tend to do is make the economics of a business impossible to look away from. OpenAI has been one of the most consequential companies of the last decade and also one of the most opaque. That's changing.

The math in that filing is the clearest signal yet that the accessible-pricing phase of AI has an expiration date. The reasonable move, while it's still cheap, is to figure out which tools are actually earning their keep.