OpenAI Misses Key Revenue, User Targets in High-Stakes Sprint Toward IPO, What It Actually Means
There's a particular kind of irony that stings more than most. It's the kind where the company everyone assumed was printing money, the company whose name became a verb the way Google did two decades ago, turns out to be… stumbling.
OpenAI, the AI darling that defined the generative-AI era, just missed its own targets. Not Wall Street's targets. Its own. And it did it in the worst possible moment: months before what could be the largest public offering in tech history.
On April 27, 2026, the Wall Street Journal dropped an investigation that sent a shiver through the AI investing world. OpenAI, racing toward an IPO that could value it at $1 trillion, had failed to reach three critical milestones: its 1 billion weekly active user goal for ChatGPT, its 2025 annual revenue target, and multiple monthly sales targets in early 2026.
Let's unpack what actually happened, and more importantly, what it means for regular people watching this unfold from the outside. Because this story isn't really about spreadsheets or quarterly projections. It's about what happens when a company bets the farm and the farm starts looking smaller than expected.
What Exactly Went Wrong, The Numbers Laid Bare
Before we go deeper, let's get the uncomfortable math on the table. Three things fell short.
The Billion-User Goal That Never Came
OpenAI set an internal target: 1 billion weekly active users for ChatGPT by the end of 2025. That milestone was supposed to be announced, a triumphant marketing moment timed to fuel the IPO narrative. It never happened.
By December 2025, weekly active users hovered around 900 million. Close. But not close enough. And "close" doesn't count when you're trying to justify a trillion-dollar valuation on a product that still loses money on every heavy user who doesn't pay.
The user growth curve also tells a quieter, grimmer story: monthly user growth dropped from 42% earlier in the year to just 13% by September. Mobile downloads declined over 8% in October. Daily active sessions fell 21%. Growth hasn't stopped — but its momentum is fading at exactly the wrong time.
Revenue Fell Short, and Kept Falling Short
OpenAI's 2025 realized revenue came in around $13.1 billion. That's impressive in isolation, up from roughly $6 billion the year before. But it fell short of the internal annual target.
Worse, the trend continued into 2026. Multiple monthly revenue targets were missed in the early months of the year. For a company racing toward a public listing, sustained underperformance raises the kind of questions no glossy investor deck can answer: Is the growth trajectory real, or was it a pandemic-style spike that's now normalizing?
The Churn Problem Nobody Talks About
Here's the number that should keep executives up at night: only about 5% of ChatGPT's weekly users are paying subscribers. That means 95% of users are using the product for free, and every query those free users submit costs OpenAI real money in compute, electricity, and inference time.
A Deutsche Bank analysis estimated OpenAI spends roughly $2.25 for every $1 of revenue it generates. Sam Altman himself has admitted that even the $200-per-month Pro tier is unprofitable. This isn't a SaaS company. Traditional software has near-zero marginal costs after development. But AI inference, every "ask ChatGPT to rewrite my email" or "generate a dinner recipe", costs real compute cycles that don't get cheaper with scale the way SaaS does.
ChatGPT subscription churn also emerged as a persistent headache. Google's Gemini gained significant traction throughout 2025, pulling users away. And in the enterprise coding and API market, an even sharper competitor was sharpening its claws.
The Real Villain, OpenAI's $600 Billion Compute Gambit
None of these missed targets would matter as much if OpenAI weren't sitting on top of a spending mountain so large it defies intuition.
When "Buy More GPUs" Becomes a Liability
Sam Altman spent years advocating for one thing above all else: secure as much compute as possible. His logic was sound, in the AI race, whoever controls the compute controls the frontier. So he pushed the company into a series of massive data center procurement deals.
The result: OpenAI committed to approximately $600 billion in compute spending through 2030. That's a scaled-back version of an even more aggressive $1.4 trillion figure Altman initially floated. But $600 billion is still… $600 billion. It's more than the GDP of most countries.
When ChatGPT was growing explosively, this "buy-everything" strategy looked visionary. But now that growth is decelerating, it looks like a liability, a set of contractual legs the company may not be able to outrun.
Oracle alone inked a $300 billion five-year computing deal with OpenAI. NVIDIA pledged billions. Amazon and OpenAI recently expanded their cloud partnership by an additional $100 billion. These aren't optional expenses, they're signed agreements. When revenue slows, those commitments don't shrink. They just get heavier.
CFO Sarah Friar's Quiet Alarm Bell
This is where the story gets human. Sarah Friar, OpenAI's Chief Financial Officer since mid-2024, has been sounding a warning inside the company. Her message was blunt: if revenue growth doesn't accelerate, OpenAI may struggle to fund its future compute obligations.
Friar brought a particular lens to the job. Before OpenAI, she led Nextdoor through its post-pandemic valuation collapse. She's seen what happens when growth narratives meet market reality. So when she told board members and senior executives that the company isn't yet ready to meet the stringent disclosure and internal-control standards required of a public company, it carried weight.
She didn't say the IPO is dead. She said the timeline needs more runway. And that created friction with Altman, who wants to sprint toward a Q4 2026 listing, ideally ahead of Anthropic's own planned IPO, to claim the narrative crown of "first AI super-platform stock."
Competition Is Eating OpenAI's Lunch, Quietly, Methodically, from the Side
For two years, the story was simple: OpenAI had the brand. Anthropic had the nerds. Google had the legacy. But that simplicity has crumbled.
Anthropic and the Enterprise Steal
Anthropic's Claude has been quietly dominating the coding and enterprise AI market. Recent data shows the market share gap between OpenAI and Anthropic has narrowed dramatically. According to Ramp's enterprise spending report from March 2026, OpenAI held 35.2% of enterprise AI adoption, while Anthropic surged to 30.6%, a gap of just 4.5 percentage points, down from nearly 13 points at the start of 2025.
In specific verticals, information services, finance, insurance, and professional services, Anthropic has already overtaken OpenAI. And 73% of first-time enterprise AI buyers are now choosing Claude over ChatGPT.
Why? Because Anthropic bet on API and enterprise integration from day one. Claude Code, Claude Cowork, and Claude Mythos have resonated with developers who want AI tools embedded directly into their workflow, not a separate chat window. Anthropic's annualized revenue reportedly surged from $90 billion at the end of 2025 to over $300 billion by March 2026.
Google Gemini, the Quiet Third Wheel
Google's Gemini also surged in late 2025, pulling market share away from ChatGPT in the consumer space. The one-two punch of Anthropic taking enterprise and Gemini eroding consumer has squeezed OpenAI from both ends.
This isn't a story of OpenAI "collapsing." It's a story of open-field running coming to an end, and the race getting genuinely competitive right when the company needs to look invincible.
Inside the Boardroom: The Altman-Friar Tension
The Wall Street Journal report surfaced something the public hadn't previously seen: open tension between the CEO and the CFO over spending discipline and IPO timing.
Altman wants to push forward aggressively, securing more compute, moving toward a Q4 2026 listing, and projecting confidence. Friar and other members of the leadership team have been tightening cost controls, reviewing data center commitments, and urging caution on the timeline.
In a joint statement following the report, Altman and Friar dismissed the idea of internal conflict as "ridiculous," insisting: "We are totally aligned on buying as much compute as we can and working hard on it together every day".
And perhaps they are aligned now. But the pattern is recognizable to anyone who's watched companies navigate pre-IPO turbulence: a growth-oriented CEO and a risk-conscious CFO pulling in different directions. The friction isn't necessarily toxic, but it signals that the path to a public listing is more complex than the official narrative suggests.
The board has reportedly begun more rigorous review of Altman's compute procurement strategy, questioning whether continued spending at this scale is prudent when growth is decelerating.
The IPO Timeline, Still On, but at What Cost?
OpenAI has not officially withdrawn or delayed its IPO plans. The company raised $122 billion in March 2026 at an $852 billion valuation. It has expanded its finance team, hired a head of investor relations, and begun informal conversations with Wall Street banks. No S-1 filing has been made yet, but the target listing venue is Nasdaq, and the internal goal has been Q4 2026.
However, the company's own cash burn projections complicate the picture. OpenAI expects a $14 billion loss in 2026, with annual cash burn potentially rising to $57 billion by 2027. The company doesn't expect to reach profitability until 2029 or 2030, and cumulative losses through that period could exceed $200 billion.
That means public-market investors would be buying into a company that plans to lose enormous sums of money for several more years, with the hope that eventual market dominance justifies the journey.
The recent Microsoft renegotiation also recasts the landscape. Microsoft gave up its exclusive license to OpenAI's intellectual property, meaning OpenAI can now offer its models through Amazon Web Services and Google Cloud, creating new revenue pathways. But the restructured deal also caps the revenue share Microsoft receives, which frees OpenAI's margins while loosening its most important strategic anchor.
Separately, OpenAI has told investors it will reserve a portion of IPO shares specifically for retail investors, a move designed to broaden public ownership and build trust.
What This Means for You, Regardless of Whether You Invest
The OpenAI story matters beyond the stock ticker.
First, it's a test case for the entire AI economy. OpenAI, Anthropic, and SpaceX are all targeting IPOs in 2026, a concentration of mega-listings that some investors warn could overwhelm the market's absorption capacity. If one of them stumbles badly, the chill could spread across the entire AI investment ecosystem.
Second, it validates a suspicion many have harbored: that AI's unit economics aren't yet solved. When the most famous AI company in the world admits it loses money on its premium product and converts only 5% of users to paid subscriptions, the industry has a structural profitability problem, not just an OpenAI problem.
Third, the competitive dynamics suggest that the AI market may not be a winner-take-all play. Anthropic is proving that a focused enterprise strategy can win specific verticals even against a competitor with vastly more brand recognition and users. Google is showing that deployment at scale can win consumer mindshare. The AI economy may look more like cloud computing, a multi-player oligopoly, than a platform monopoly.
For anyone considering exposure to an eventual OpenAI IPO, the watchlist is clear: revenue growth acceleration, free-to-paid conversion rates, compute cost trends, and whether the board and leadership unify around a realistic timeline. These are the signals that will separate a defining tech event from a cautionary tale.
A Crossroads, Not a Collapse
It's tempting to read the headline, "OpenAI Misses Key Targets Ahead of IPO", and assume the sky is falling. That would be the wrong read.
This is a recalibration, not a collapse. OpenAI still has 900 million weekly active users. It still has the largest AI funding round in Silicon Valley history. It still has the most recognized brand in generative AI. GPT-5.5, released this month, tops multiple industry benchmark tests. Codex, its coding agent, is growing rapidly.
But the growth-at-all-costs era is ending. The public markets will demand proof that the economics work, not just that the models are impressive. Sarah Friar's caution and Sam Altman's ambition aren't opposites; they're two sides of a necessary tension that every company must resolve before stepping into the unforgiving light of quarterly earnings calls and analyst expectations.
OpenAI hasn't lost the race. But it's suddenly feeling the burn in its legs, and the finish line, for the first time, looks farther away than it did a quarter ago.