The Most Anticipated IPO in History Is Also the Most Honest Argument for Why Moats Beat Revenue
OpenAI filed its S-1. The numbers are staggering. The losses are bigger.
OpenAI confidentially filed its S-1 with the SEC on May 22, targeting a Q4 2026 public listing at a valuation between $852 billion and $1 trillion, with Goldman Sachs and Morgan Stanley leading the deal. If it prices at the top of that range, it’ll be the largest tech IPO in history — bigger than Meta, Alibaba, and Uber combined.
Here’s the number everyone keeps skipping over.
OpenAI is losing $1.22 for every $1 of revenue. The projected operating loss for 2026 is $14 billion. Revenue went from $2 billion annually in 2023 to $25 billion annualized by February 2026 — 12.5x growth in roughly three years. That’s extraordinary. And yet the company is burning faster than it’s earning.
The financial press is treating this as a growth story. It’s actually a moat story.
Mental Model: The Flywheel
A flywheel is a system where each turn makes the next turn easier. The energy compounds. What makes OpenAI worth a trillion dollars isn’t the current revenue or the current losses — it’s whether the flywheel is real.
Enterprise now represents roughly 40% of revenue and is on track to match consumer by year-end. ChatGPT’s ad business crossed $100 million in annualized revenue in under six weeks after launching May 5. That’s the flywheel showing up in the numbers. More users means more data means better models means more enterprise contracts means more compute investment means better models.
The bet isn’t that OpenAI is profitable today. The bet is that the flywheel is spinning fast enough that profitability becomes structurally inevitable — and that by the time it does, nobody can catch them.
That’s a legitimate investment thesis. It’s also the riskiest kind to hold in public markets, where quarterly earnings matter.
The Contrarian Take
Everyone’s debating the valuation. I think the more interesting question is what happens to every company built on top ofOpenAI’s infrastructure once those financials are public.
The public S-1 will reveal audited revenue, margins after compute costs, and OpenAI’s contractual obligations — the line items that will be front-and-center in analyst models. Right now, OpenAI’s pricing power over its API customers is opaque. Once the S-1 lands, every enterprise buyer will know exactly how much margin OpenAI needs to extract from them to close the gap on that $14 billion loss.
That’s when the conversation about build vs. buy shifts for a lot of companies. And that’s when teams with proprietary infrastructure — models they own, data pipelines they control — start looking a lot more valuable than the ones who outsourced their core intelligence to a platform that’s about to go public and answer to shareholders.
The September listing window may slip into Q1 2027 if the SEC review surfaces accounting questions. Watch that clock. The delay risk is real and the market’s patience for unprofitable trillion-dollar companies isn’t infinite.
The flywheel is spinning. The question is whether it’s spinning fast enough.