Tips, corrections, or questions? support@omniscient.media
Consequential AI, explained and evaluated, every weekday.
The Omniscient Bulletin: 5 to 7 items a day with the take, not the recap.
OpenAI and Anthropic have built novel governance structures to survive public markets. Neither has been tested. As both companies approach public listings, the question isn't whether their governance frameworks look good on paper - it's whether those structures can withstand the slow, compounding pressure of quarterly earnings scrutiny.
A company valued at $852 billion, fresh from the largest funding round in Silicon Valley history at $122 billion, and projecting losses of $14 billion this year has now been revealed to have missed multiple internal revenue targets in early 2026. OpenAI's finances were, by any measure, already precarious. The Wall Street Journal's report on Monday made them dramatically more complicated to defend.[1]
The most consequential detail in the WSJ report is not the missed targets themselves but what CFO Sarah Friar reportedly told other company leaders: if revenue does not accelerate, OpenAI may struggle to honor its future compute contracts.[1][2] That warning is the context for everything else. The underlying causes are specific: OpenAI ceded ground to Anthropic in the coding and enterprise segments - its highest-margin categories - while ChatGPT's user growth slowed visibly in the back half of 2025.[1][2] Subscriber defections added to the drag, and the company fell short of its internal target to reach one billion weekly active users by year-end, a milestone cited internally as a key indicator ahead of a planned public listing.[1][2]
Altman and Friar moved quickly to contain the story, issuing a joint statement reading: "This is ridiculous. We are totally aligned on buying as much compute as we can and working hard on it together every day."[2] The damage-control framing is notable: they did not deny the missed targets, only the suggestion of internal discord. That is a meaningful distinction as the company prepares an S-1.
Taken in isolation, a growth company missing monthly stretch targets is unremarkable. The AI sector runs on ambition-inflated projections by design. What makes this development consequential is not the miss itself but the financial architecture surrounding it - and specifically what it means for the single largest bet made on OpenAI's trajectory.
As we reported in early April, SoftBank wired its first $10 billion OpenAI tranche on April 1, drawn entirely from a $40 billion bridge loan arranged with JPMorgan Chase, Goldman Sachs, Mizuho, SMBC, and MUFG.[3][4] The loan is unsecured, carries no collateral, and matures on March 25, 2027. Two further $10 billion tranches follow on July 1 and October 1 of this year.[3] When the sequence completes, SoftBank's total investment in OpenAI will reach $64.6 billion, representing roughly 13% of the company at the $852 billion post-money valuation.[5]
The loan's entire repayment logic rests on one assumption: that OpenAI goes public before March 2027. The preferred shares SoftBank holds convert to liquid equity upon a listing. S&P Global, which revised SoftBank's credit outlook to negative in March, made the dependency explicit - naming an IPO as the primary condition for restoring SoftBank's credit quality.[6] Strip out that scenario, and SoftBank faces a bridge refinancing with $40 billion in unsecured debt, holding illiquid preferred stock in a company now known to be missing its own growth targets.
OpenAI has been targeting a Q4 2026 public listing, which would land within the loan's maturity window - though barely. CFO Sarah Friar confirmed the company is preparing to operate like a public company, telling CNBC in April that retail investors would receive a share allocation in the IPO and that "at our scale, raising equity forever doesn't make any sense."[7] The revenue miss narrative is acutely ill-timed for that schedule. IPO investors price growth momentum as much as current revenue; a pattern of missed targets, even if temporary, becomes a liability in S-1 disclosures and roadshow narratives. Underwriters will want a clean, accelerating growth story. What they have instead, at least for now, is a chart that bends in the wrong direction at a company with $14 billion in projected losses for the year it intends to list.
The competitive pressure from Anthropic is a specific concern. Enterprise and coding markets are OpenAI's highest-margin, stickiest revenue segments - the categories most likely to support the recurring-revenue multiples that justify a valuation above $800 billion.[1][2] If Anthropic has genuinely made inroads there, the revenue mix problem is structural, not a temporary blip that better marketing can correct before a roadshow.
OpenAI has committed to spending $600 billion on compute by 2030, a figure the company itself provided to investors in February after previously touting larger infrastructure commitment totals.[8] The revenue base needs to grow fast enough not just to cover losses, but to credibly signal it will cover those contracts. That is the bar the CFO is worried about clearing.
Masayoshi Son has made this bet before - not in scale, but in structure. The Vision Fund's $32 billion loss in fiscal 2022-23, the largest single-year loss the fund has recorded, came in part because a generation of illiquid private bets could not be exited when the market turned.[9] The current OpenAI position replicates that structural vulnerability: preferred shares that are illiquid until an IPO, concentrated in a single asset, backed by borrowed money with a hard maturity date. Son has also been spectacularly right - the Alibaba investment remains one of the most lucrative venture bets in history. The question is not whether the long-term AI thesis is sound. It is whether the transaction's engineering leaves SoftBank enough runway to survive a delayed IPO.
The answer, based on the deal's architecture, is: not much. SoftBank's press release on the bridge facility noted that borrowings "are expected to be repaid in stages by the maturity date through the utilization of existing assets and other financing measures" - language deliberately broad enough to encompass Arm share sales, a bridge refinancing, or an IPO.[4] After selling its entire $5.83 billion Nvidia position in late 2025, SoftBank's portfolio is now concentrated in two assets: Arm, its publicly listed semiconductor holding, and OpenAI, which now equals Arm's share of total investment assets at roughly 30% each.[6] That is not a portfolio with abundant liquidity levers to pull if the IPO window closes.
Today's news does not make the OpenAI IPO less likely - it makes it less flexible. An IPO delayed to 2027 or beyond would force SoftBank to refinance the bridge on terms that now include a revenue-miss narrative in every lender conversation. The timing dependency that was already the most fragile element of this deal has just become more fragile. For SoftBank investors watching a $40 billion unsecured clock tick toward March 2027, that is the headline beneath the headline.
Wall Street Journal: OpenAI Misses Key Revenue, User Targets in High-Stakes Sprint Toward IPO (April 28, 2026) Inline ↗
Reuters via WIXX: OpenAI falls short of revenue and user targets as it races toward IPO (April 27, 2026) Inline ↗
SoftBank Group Corp.: Execution of Follow-on Investment (First Tranche) in OpenAI (April 1, 2026) Inline ↗
SoftBank Group Corp.: Execution of Bridge Facility Agreement Primarily for the Follow-on Investments in OpenAI (March 27, 2026) Inline ↗
SoftBank Group Corp.: Follow-on Investments in OpenAI (February 27, 2026) Inline ↗
Silicon Republic: SoftBank credit outlook hit after betting $30bn more on OpenAI, citing S&P Global research (March 3, 2026) Inline ↗
CNBC: OpenAI will allocate IPO shares to retail investors as it preps for debut, CFO says (April 8, 2026) Inline ↗
CNBC: OpenAI resets spending expectations, targets around $600 billion by 2030 (February 20, 2026) Inline ↗
CNBC: SoftBank full year 2022 earnings: Vision Fund posts $32 billion loss (May 11, 2023)
