The Wall Street Journal reported on April 27, 2026 that OpenAI has missed its own internal targets for both new user acquisition and revenue, with CFO Sarah Friar privately telling colleagues the company may not be able to fund its computing commitments if growth does not accelerate. The numbers behind the story reveal a startup that is simultaneously the fastest-growing software company in history and one of the most financially fragile.

What the WSJ Report Says

According to the Wall Street Journal report, OpenAI fell short of several monthly sales targets in 2026 after rival Anthropic gained ground in the coding and enterprise markets (Wall Street Journal, 2026). ChatGPT also failed to reach the company’s internal target of 1 billion weekly active users by the end of 2025, finishing the year at approximately 900 million (Moneycontrol, 2026). Subscriber defection remains a problem as Google Gemini gained popularity throughout 2025 (Wall Street Journal, 2026).

CFO Sarah Friar, in conversations with other company leaders, “expressed concerns that if OpenAI doesn’t increase sales fast enough, it may not be able to afford its future computing needs,” according to the WSJ report (Moneycontrol, 2026). The disclosure is significant because OpenAI is preparing for what could be the largest IPO in history, with filings expected by late 2026 or 2027 at a valuation potentially approaching $1 trillion (Reuters, 2025).

The Revenue Numbers Look Spectacular Until You Examine Them

OpenAI’s revenue growth is genuinely historic. Annualized revenue climbed from approximately $1 billion at the end of 2023 to $5.5 billion by Q4 2024, $10 billion by mid-2025, $12 billion by year-end 2025, and roughly $19 billion by March 2026 (Second Talent, 2026; European Business Magazine, 2026). The company now generates approximately $2 billion per month in revenue.

The 2026 revenue target reportedly sits at approximately $29.4 billion (Second Talent, 2026), which would require revenue acceleration in the second half of the year. The WSJ report indicates that this acceleration is not happening at the expected pace, with monthly sales targets being missed.

The Conversion Problem: 900 Million Users, Only 5% Pay

The most structural weakness in OpenAI’s business model is its monetization rate. ChatGPT crossed 900 million weekly active users in early 2026, more than double the 400 million figure from the same window in 2025 (Second Talent, 2026). However, only approximately 5.5% of those users pay for any subscription tier (European Business Magazine, 2026). The remaining 94.5% access ChatGPT for free, while OpenAI bears the compute cost of every single query across that user base (European Business Magazine, 2026).

For comparison, Spotify converts approximately 46% of its monthly active users to paid subscriptions. Netflix converts close to 100% (it has no free tier). Even by the most generous benchmarks for freemium consumer software, ChatGPT’s conversion rate is a structural anomaly. The reason matters: years of free access have trained users to view AI chat as a free utility, which makes raising prices or paywalling features politically difficult and competitively risky as Gemini and Claude offer comparable free alternatives.

The Loss Trajectory: $44 Billion Before First Profit

OpenAI’s internal financial projections, first reported by The Information and confirmed by multiple subsequent outlets, show losses of approximately $14 billion in 2026 alone, nearly triple the loss recorded in 2025 (R&D World, 2026; Forbes, 2026). The company expects cumulative losses of approximately $44 billion between 2023 and 2028, with first profitability not arriving until 2029 (European Business Magazine, 2026; Yahoo Finance, 2026).

The cash burn is among the largest of any private company in history. OpenAI raised its 5-year projected cash burn to approximately $115 billion through 2029 (Sybrid, 2026). The company spends roughly $1.4 billion per year on computing resources alone, and its annual cash burn now reaches approximately $17 billion in 2026, primarily driven by the cost of GPU capacity and data center infrastructure (Forbes, 2026). To put this in perspective: OpenAI is on track to lose more money in 2026 than the entire annual revenue of S&P 500 companies like Costco’s e-commerce division or the entire annual GDP of Iceland.

The Competitive Pressure: Anthropic and Google Are Closing Fast

The WSJ report explicitly cites Anthropic’s gains in coding and enterprise markets as a factor in OpenAI’s missed targets (Wall Street Journal, 2026). Anthropic’s revenue has surged from $9 billion at the end of 2025 to over $30 billion run rate in early April 2026, and the company captured more than 73% of all spending among first-time enterprise AI buyers (TechCrunch, 2026). On a relative basis, Anthropic is now growing faster than OpenAI in the most lucrative customer segment.

The web traffic share data is even more concerning for OpenAI than the revenue picture. According to Similarweb’s Global AI Tracker, ChatGPT’s share of AI chatbot web traffic fell from 86.7% in January 2025 to 64.5% in January 2026, a 22.2 percentage point decline in twelve months (R&D World, 2026). Google Gemini captured the majority of that share, growing from 5.7% to 21.5% over the same period. ChatGPT remains the dominant consumer brand, but the trajectory shows competitive moats eroding faster than most market observers anticipated.

Our Analysis: Five Implications

1. The IPO timeline is now under real pressure. CNBC reported in early April that OpenAI was preparing for an IPO with retail investor allocation, and that its recent private placement led by JP Morgan, Morgan Stanley, and Goldman Sachs was three times oversubscribed (CNBC, 2026). That oversubscription was based on assumptions about revenue trajectory that the WSJ report now puts in question. If 2026 revenue lands meaningfully below the $29.4 billion target, the IPO valuation framework, which has been guided toward $1 trillion, becomes harder to justify. Public market investors will demand a credible path to profitability that the current $14 billion loss profile does not support without a sharp acceleration in the conversion rate or enterprise revenue.

2. The enterprise pivot is necessary but not sufficient. Chief Commercial Officer Denise Dresser told CNBC that enterprise currently accounts for 40% of revenue and is expected to match consumer revenue by year-end 2026 (CNBC, 2026). This is the right strategic direction. Enterprise customers pay multi-year contracts at higher per-seat economics than ChatGPT Plus subscribers. The problem is that enterprise sales cycles are 6 to 18 months long, and OpenAI is now competing in that segment against a rejuvenated Anthropic, a Microsoft Copilot ecosystem that bundles Office workloads, and Google’s Workspace integration. Winning enterprise share is harder and slower than winning consumer share.

3. Compute commitments are the hardest constraint to walk back. CFO Friar’s reported concern is specifically about funding “future computing needs.” OpenAI has committed to spending approximately $1.4 trillion on hardware and cloud infrastructure between 2025 and 2035 (NextBigFuture, 2025). These commitments are made years in advance to secure GPU capacity and data center access. Renegotiating them is possible but signals weakness to suppliers and partners. Pulling back on compute also means reducing the rate of model improvement, which could accelerate market share losses to Anthropic and Google. This is the structural trap: OpenAI must spend to compete, but it cannot spend at this pace without proportionally faster revenue growth.

4. The advertising experiment is a partial answer at best. OpenAI announced in early 2026 that it would test ads in ChatGPT’s free and Go tiers (Sybrid, 2026). This is the obvious lever to pull, but the math is unforgiving. Even if OpenAI achieved Meta-level advertising revenue per user (~$45 ARPU annually for Meta in 2025), the 850 million free users would generate only about $38 billion in ad revenue, and that assumes a saturation level that took Meta two decades to achieve. More realistic near-term ad revenue is likely $2 to $5 billion in 2027, useful but not transformative against a $17 billion annual cash burn.

5. The “pre-IPO disclosure” pattern is itself revealing. The fact that internal financial concerns from the CFO are leaking to the WSJ before an IPO filing is, by itself, a meaningful signal. Companies preparing for highly-anticipated IPOs typically maintain extreme messaging discipline. Information leaks from senior leadership to the financial press in the months before a filing usually indicate either internal disagreement about strategy, deliberate pre-positioning by leaders trying to manage future expectations, or organizational stress that is hard to contain. None of those scenarios is positive for the IPO narrative.

Conclusion: The Hardest Problem in AI Is Not Technology

OpenAI is not facing a product crisis. ChatGPT is still the most-used AI application in the world, GPT-5 remains competitive at the frontier of capability, and the company’s brand recognition is unmatched in consumer AI. The crisis is structural and financial. The product that built the world’s largest AI distribution platform is also the product that loses money on every free query. The user base that justifies a $1 trillion valuation is also the user base whose 94.5% non-paying majority makes that valuation impossible to support through normal SaaS economics.

The WSJ report does not change the underlying business model. It confirms what the financial projections have been telegraphing for months: OpenAI’s path to profitability depends on a combination of revenue acceleration, conversion rate improvement, advertising monetization, and infrastructure cost discipline that has never been achieved at this scale by any consumer software company. The IPO will likely happen on schedule because the financial commitments require the capital. The question is whether public market investors, looking at the same data the WSJ now reports, will pay $1 trillion for a story whose execution has just become measurably harder.

References