News highlights ― OpenAI revealed it applied "on its own initiative"
Let me first organize the facts. On June 8, 2026, OpenAI officially announced that it had "recently submitted a confidential S-1 to the SEC." Noteworthy is the tone of that announcement: in its statement, the company said, "We figured this would leak at some point, so we wanted to share the news ourselves. We haven't determined a timeline for going public. It may be a while — there are some advantages to staying private. But this does give us the option to move faster if we choose." In other words, this was not a triumphant declaration of "we are going public," but rather a highly suggestive self-disclosure to the effect of "we have completed the filing process and kept the option of an IPO in hand."
Cross-referencing press reports, the actual confidential filing occurred before the June announcement, consistent with the timing of U.S. media reports from around May 20 stating that the submission could come "as early as Friday the 22nd" (i.e., late May). Bloomberg, Reuters, CNBC, The Wall Street Journal, The Washington Post, the Financial Times (UK), Nikkei, and Bloomberg Japan all reported the June 8 announcement almost simultaneously. The Nikkei ran the headline: "OpenAI Announces IPO Filing — 'Timing Undecided,' Valuation May Reach ¥160 Trillion." Goldman Sachs and Morgan Stanley have been named as lead underwriters, with some reports indicating JPMorgan has also joined as a lead.
At this point, I want to note one important caveat that runs throughout this entire piece. The filing in question is a *confidential* S-1. Under U.S. rules established since 2017, companies seeking to list can submit their initial draft registration statement to the SEC confidentially, keeping detailed financials under wraps until just before the formal investor roadshow. Accordingly, the revenue figures and loss amounts circulating about OpenAI are not numbers disclosed from this filing, but are based on reporting and analyst estimates that predated the submission. This piece treats financial figures on that same basis.
What is OpenAI in the first place — From ChatGPT to a "Super App"
Before getting into the specifics, let us take a fresh look at the company OpenAI and its flagship service. OpenAI was founded in 2015 by Sam Altman, Elon Musk, and others as a nonprofit artificial intelligence research organization, and it became a global phenomenon overnight when it launched the conversational AI "ChatGPT" in November 2022. ChatGPT is a generative AI that responds in natural written language to text prompts entered by users, much like a conversation between people.
What exactly can it do? For example, if you ask it to "put together a three-night, four-day itinerary departing Tokyo for San Francisco for a business trip next week," it will draft a schedule; if you paste a lengthy English contract and say "summarize this in three lines in Japanese," it will distill the key points; and if you ask it to "fix the bug in this Python code," it will correct the program. Its uses have expanded rapidly beyond writing, translation, and summarization to include image generation, spreadsheet data analysis, meeting-minute drafting, and even "agents (autonomous AI)" that carry out multiple steps on the user's behalf. The scale of usage is extraordinary: ChatGPT's weekly active users reached approximately 900 million as of February 2026 (doubling from 400 million in the same month the previous year), and monthly active users of the app reportedly surpassed one billion in June 2026 (per Sensor Tower estimates, as reported by Reuters). Paid subscribers number around 50 million, paid enterprise users exceeded 11 million as of May 2026, and the number of prompts sent per day stands at approximately 2.5 billion.
Now, with an IPO on the horizon, OpenAI is making a significant shift in its product strategy. According to a report by the Financial Times on June 7, the company is reimagining ChatGPT—from a simple chatbot into a "super app" that brings together the coding assistant tool "Codex," autonomous agents, search-like tasks, image generation, and even external services such as Canva and Booking.com under a single entry point. The move is driven by a conviction that "the future of AI lies in agents that autonomously handle multi-step tasks such as travel booking and calendar management," and the number of weekly Codex users has reportedly surged sixfold since launch to over five million. For the company, the key question shaping its post-IPO valuation is not just consumer-facing buzz, but how to grow its enterprise business—which currently accounts for roughly 40% of revenue and is planned to be raised to 50% by year's end.
What is a "Private Application" ― Why the Finances Are Not Visible
Now we turn to the specifics. The key to correctly understanding this latest development lies in the nature of the procedure known as a "confidential S-1 filing." An S-1 is the registration statement (the prototype of a prospectus) that U.S. companies submit to the SEC when going public, detailing their business operations, risk factors, financial statements, capital structure, and more — essentially a document that serves as both a "company report card" and a "health checkup." Normally this is made public and accessible to anyone.
What OpenAI chose, however, was a "confidential" filing. This is an approach the SEC has permitted for all prospective IPO companies since 2017, whereby the contents are worked out exclusively between the company and the SEC during the initial draft stage, going through multiple revisions before the full text is finally made public a certain period before the roadshow begins (generally at least 15 days prior). The advantages are clear: the company can proceed with IPO preparations without exposing its inner workings (detailed financials and segment-by-segment figures) to competitors, and if market conditions turn unfavorable, it can quietly postpone. OpenAI's cautious stance of "timing is undetermined" can be read as a posture aimed at making maximum use of this system's flexibility.
Conversely, this means that at present neither investors nor the press have seen OpenAI's "official" financials. The revenue figures and deficit amounts discussed below are not drawn from the filing itself, but rather from prior reporting, statements from those close to the company, and reporting by specialist media outlets (such as The Information). The actual figures will only emerge the moment OpenAI switches its confidential S-1 to a "public" version. In that sense, this announcement is not the "finish line" — it is merely "the starting gun" for the long procedural process of disclosure.
The Peculiarity of Corporate Governance — The "PBC" Structure Where a Nonprofit Controls a For-Profit
One aspect that cannot be avoided when discussing OpenAI's IPO is its highly unusual corporate governance. OpenAI completed a restructuring on October 28, 2025, reorganizing its for-profit division into "OpenAI Group PBC," a Public Benefit Corporation (PBC), and its non-profit division into the "OpenAI Foundation." A PBC is a for-profit company legally required to pursue not only shareholder interests but also a public benefit purpose stated in its charter — in OpenAI's case, "ensuring that artificial general intelligence (AGI) benefits all of humanity." It is this OpenAI Group PBC that will be listed.
At the time the restructuring was completed, the ownership structure was publicly disclosed as follows: the OpenAI Foundation (non-profit) holds approximately 26%, Microsoft holds approximately 27% (worth roughly $135 billion, or approximately ¥21.6 trillion at the time of valuation), and the remaining approximately 47% is held by employees and other investors. What is critically important here is that even though its stake is a minority interest, all directors of OpenAI Group PBC are appointed and removed by the OpenAI Foundation (non-profit). In other words, economic ownership and control through voting have been intentionally separated. Microsoft, despite being the largest single shareholder, is not in a position to control the board of directors.
This structure raises unique considerations for general shareholders after the listing. Investors in the public market will be buying shares in a company bound by a mission that is not solely aimed at profit maximization, while the power to appoint and remove directors remains in the hands of a non-profit foundation. The design of voting rights (such as whether dual-class shares exist) and the extent to which the non-profit foundation's judgments about "public benefit" may affect future management flexibility and shareholder returns will likely be the most closely watched governance issues in the public S-1 filing. This is also directly connected to the argument that Elon Musk has long criticized — that "assets gathered as a non-profit were converted to for-profit use" — and this point is closely intertwined with the litigation discussed later.
Fundraising Trajectory ― From $300 Billion to $852 Billion in One Year
OpenAI's valuation has exploded in the private market without waiting for a public listing. The most recent milestone was a funding round totaling $122 billion completed on March 31, 2026, which brought the post-money valuation to $852 billion. A year earlier, at the end of March 2025, a $40 billion round led by SoftBank Group valued the company at $300 billion — meaning the valuation roughly tripled in just one year.
The lineup of investors in the March 2026 round is heavily skewed toward strategic corporations, defying conventional venture capital norms. According to reports, Amazon contributed approximately $50 billion, while Nvidia and SoftBank each invested around $30 billion, with Microsoft also continuing its participation. The round further included a who's who of prominent VCs, institutional investors, and sovereign wealth funds from the US, Europe, and Asia — among them Andreessen Horowitz (a16z), Sequoia Capital, Coatue, Altimeter, Dragoneer, Fidelity, BlackRock-affiliated funds, Temasek, Thrive Capital, and the University of California endowment. In addition, OpenAI opened its doors to individual investors through banks for the first time, reportedly raising more than $3 billion from retail participants. The fact that OpenAI has already experienced what amounts to a quasi-public offering while remaining private is a point that cannot be overlooked when assessing its upcoming IPO.
The SoftBank-led round from a year earlier is equally telling. Of that $40 billion, SoftBank contributed $30 billion and a coalition including Microsoft, Coatue, Altimeter, and Thrive covered the remaining $10 billion. The bulk of SoftBank's portion was contingent on OpenAI completing its conversion to a for-profit entity within the year — a condition that was ultimately met, with SoftBank paying in the full amount by the end of 2025. In other words, the restructuring into a Public Benefit Corporation described in the previous chapter was also a prerequisite for closing that massive funding round. OpenAI is also advancing plans alongside SoftBank and Oracle to build "Stargate," a data center network with a total investment of up to $500 billion, and much of the capital raised is being directed toward compute infrastructure. It is precisely this flow of funds — where investors, customers, and construction partners overlap in multiple layers — that has become fertile ground for the "circular transaction" criticism discussed later.
How Silicon Valley VCs and Analysts See It
So how are Silicon Valley investors and analysts reading this filing? The most repeated take is that "there is inherent value in being first to reach the public markets." Dan Ives of Wedbush Securities told CNBC in May: "In the middle of this arms race, being the first to reach the public markets is critically important. You get to set the valuation benchmark, and there's a first-mover advantage in being the first to tell your story to investors." The underlying view is that the race for AI supremacy has become inseparable from the competition to raise capital in public markets.
What accelerated this "first-mover wins" dynamic was the elimination of the IPO's biggest legal risk. On May 18, 2026, an advisory jury of nine at the U.S. District Court for the Northern District of California deliberated for less than two hours and rejected all claims in the lawsuit Elon Musk had brought against CEO Sam Altman and others; Judge Yvonne Gonzalez Rogers then dismissed the case on statute-of-limitations grounds. In February 2024, Musk had sued Altman, President Greg Brockman, and OpenAI, alleging that they had "privatized and converted to profit an organization founded for charitable purposes." According to reports, had Musk's side prevailed, OpenAI and Microsoft could have faced disgorgement of profits on the order of up to $150 billion (approximately ¥24 trillion). Ives called the outcome "a huge relief for OpenAI, removing a roughly $134 billion (approximately ¥21 trillion) overhang that had weighed heavily on the path to an IPO" (estimates of the potential damages vary across reports). Musk has signaled his intent to appeal, so the embers have not been fully extinguished, but the uncertainty the market feared most has receded significantly.
That said, internal ranks are not entirely unified. CFO Sarah Friar told The Washington Post optimistically: "We want to be ready. Having access to the public markets is a good thing — the public markets are much larger than the private markets." At the same time, some reports indicate that Friar is cautious, believing "the organization is not sufficiently prepared to go public this year," and that there is a gap in urgency between her and Altman over the timing of a listing. The anecdote of prominent investor Brad Gerstner (Altimeter, also an OpenAI backer) questioning Altman about revenue targets on his podcast "BG2" is telling: when Gerstner floated "$100 billion (approximately ¥16 trillion) by 2028–2029," Altman shot back, "Why not '27 instead of 2029?" — a moment that captures both the boldness of management and the measured distance investors maintain as they watch on.
The True Picture of Finances and the "Circular Transaction" Criticism
From a VC perspective, the most contentious aspect is the current financial structure and capital circulation. Aggregating media reports and analyst estimates, OpenAI's annualized revenue run rate surpassed $20 billion (approximately ¥3.2 trillion) in 2025 and reportedly reached around $25 billion (approximately ¥4 trillion) by the end of February 2026. While user growth and revenue growth have been rapid, profitability remains distant — as the Washington Post put it, the company is "losing more than it earns." Multiple reports indicate that OpenAI is on track to post a loss of approximately $14 billion (approximately ¥2.2 trillion) in 2026 (roughly triple the 2025 figure), with no path to profitability expected until around 2030, while internal plans call for $100 billion (approximately ¥16 trillion) in revenue by 2029. Some analyses point out that at current levels the company is "losing $1.22 for every $1 of revenue earned" (equivalent to an operating margin of negative 122%). It should be noted that all of these figures are based on conventional reporting rather than any non-public S-1 filing.
What alarms many in Silicon Valley is the phenomenon known as "circular financing." The structure works as follows: Nvidia invests in OpenAI (the reported partnership framework is up to $100 billion, or approximately ¥16 trillion); OpenAI pays enormous sums to cloud providers such as Oracle (the Oracle deal has been reported at a scale of $300 billion, or approximately ¥48 trillion); and those cloud providers in turn purchase massive quantities of Nvidia GPUs. AMD has also reportedly signed a supply agreement with OpenAI valued at around $200 billion (approximately ¥32 trillion) and has granted stock warrants to customers. Because capital circulates among investors, customers, and suppliers, the effect is that "demand appears to be real demand, and revenues appear robust." Various analyses from 2026 estimate that the total value of these mutually interdependent arrangements exceeds $800 billion (approximately ¥128 trillion).
Critics draw parallels to the final stages of the dot-com bubble, describing it as "companies buying each other's services to manufacture the appearance of growth." The concern is that if genuine external demand weakens, the cycle could rapidly reverse, leaving behind nothing but excess capacity and massive losses. On the other hand, economic commentator Noah Smith and others defend the model, arguing that "capital-intensive industries — railroads, telecommunications, and past computing investments — have always been launched with suppliers providing financial support to customers. If the underlying demand is real, this is not necessarily a bubble." OpenAI's IPO will be the first time this trillion-dollar question — "Is the circulation a down payment on real demand, or a self-reinforcing illusion?" — is subjected to public market pricing. It is also worth noting that, regarding SoftBank, one of the largest investors, CNBC reported on June 4 that the company's massive bet on OpenAI and its growing debt are heightening liquidity concerns, and market attention is beginning to turn toward the risks on the side of the major capital providers as well.
AI Major IPO Rush — A Three-Way Battle Between Anthropic, SpaceX, and Databricks
OpenAI's filing must be read not as an isolated event, but as part of a wave of "frontier AI mega-IPOs" concentrated in 2026. Emblematic of this is the fact that rival Anthropic (developer of Claude) moved first. On June 1, 2026, Anthropic filed confidentially for an IPO at a valuation of $965 billion (approximately ¥154 trillion), following a $65 billion (approximately ¥10.4 trillion) Series H. This surpassed OpenAI's $852 billion (approximately ¥136 trillion) valuation, and coverage reported that "Anthropic overtook OpenAI in valuation for the first time."
From a VC perspective, this reversal is telling. Anthropic's annualized revenue surged from approximately $9 billion (approximately ¥1.4 trillion) at end of 2025, reaching an estimated $30 billion (approximately ¥4.8 trillion) annualized in April 2026 per the company's own figures, and approximately $45 billion (approximately ¥7.2 trillion) in May per research firm Sacra's estimates, with projections pointing toward surpassing $50 billion (approximately ¥8 trillion) soon (figures vary depending on methodology). Its revenue mix — heavily weighted toward high-margin enterprise and developer sales anchored by the coding assistant Claude Code — was highly valued for its profitability and sustainability. While OpenAI dominates in consumer reach with ChatGPT's 900 million to 1 billion users, Anthropic has drawn level or pulled ahead on valuation and recent revenue momentum — a contrast that sharpens the question of whether public markets will place a higher premium on "scale" or "profitability."
Other giants are also on stage. Elon Musk's SpaceX submitted a draft to the SEC in April 2026, filed a public S-1 on May 20, and is reportedly targeting what would be the largest IPO in history at a valuation of approximately $1.75 trillion (approximately ¥280 trillion) and a raise of $75 billion (approximately ¥12 trillion), with a roadshow beginning around June 4 and targeting a pricing on June 11 and a Nasdaq listing (ticker: SPCX) on June 12 (its primary revenue driver being satellite communications service Starlink). Data and AI infrastructure company Databricks is expected to file its S-1 in the second half of 2026 at a valuation of $134 billion (approximately ¥21.4 trillion), growing annualized revenue of $5.4 billion (approximately ¥864 billion) at 65%, and stands out as the only profitable company among AI-related IPO candidates. In contrast, payments giant Stripe, despite its valuation rising to $159 billion (approximately ¥25.4 trillion) in a tender offer in February 2026, keeps its distance from a listing, stating: "We can grow sufficiently on our own capital and we are already profitable. An IPO is a solution in search of a problem." The 2026 IPO pipeline is shaping up to be the most AI-skewed in history, with AI-related companies accounting for the majority of deals and over 90% of the total value. OpenAI, at its center, stands as the ultimate litmus test for the market's capacity and appetite.
Will the market have enough funds? — The structural handicap of being "the last of three massive IPOs"
The simple yet weighty question that must be asked here is whether the public markets can truly absorb capital of this magnitude. In the early summer of 2026, a veritable triple wave of frontier AI and advanced technology mega-IPOs is crashing into the market. Leading the charge is Anthropic (valuation: $965 billion), followed by SpaceX (valuation: approximately $1.75 trillion, raising $75 billion), with OpenAI (valuation: $852 billion) bringing up the rear. Adding the three valuations together yields more than $3.5 trillion, and the fresh cash to be drawn from the market alone — led by SpaceX's $75 billion — could reach an aggregate on the order of $100 billion. That compresses into a matter of months the new supply that an ordinary bull market would take years to digest.
The bearish logic is straightforward. The capital available to flow into public markets is finite, and institutional portfolios already carry technology and AI allocations at historically elevated levels. If Anthropic and SpaceX — arriving first — exhaust investors' wallets and risk appetite ahead of schedule, OpenAI, the last to arrive, will find itself competing for "whatever is left." The value of "arriving first" that analyst Ives extolled is, read in reverse, nothing other than the disadvantage of arriving last. OpenAI itself has maintained a cautious posture — "timing of the IPO is undecided," "staying private is easier for now" — and CFO Friar has let slip that "we are not set up for an IPO this year." One can read all of this as a sober calculation that already prices in the handicap of their place in line.
The concerns do not end with sequencing. First, none of the three companies will be included in major equity indices immediately after listing, meaning they can count on virtually none of the automatic buying that comes from passive, index-tracking funds. Offerings of historic scale must be digested entirely by active investors who must actively assess each company's worth, creating a structurally loose supply-demand dynamic. Second, OpenAI is pursuing the largest raise of the three while carrying an annualized deficit of roughly $14 billion and a reported revenue structure in which it loses $1.22 for every dollar of sales. While profitable companies like Databricks or self-sustaining ones like Stripe feel no urgency to go public, the company that needs capital most and bleeds the most red ink is coming last and asking for the most — this ordering itself invites market wariness. Third, should the circular-transaction dynamics described earlier go into reverse, AI-related equities broadly could experience cascading price declines, and if any one of the earlier IPOs stumbles post-listing, the valuation benchmark for the last-in-line OpenAI could be cut down accordingly.
The footing of capital providers is also far from solid. CNBC's reporting on SoftBank's liquidity concerns suggests that even those underwriting the giant bets are nearing their limits. In an environment of stubbornly high interest rates and tightening macro conditions — where capital retreats from risk assets — handling the "final mega-IPO" becomes exceptionally difficult. The fact that AI accounts for more than 90% of the 2026 IPO pipeline by value is itself a vulnerability: the fragility of an all-in bet with no room for diversification. There are counterarguments, of course — OpenAI has already raised more than $3 billion from individual investors in private rounds, counts sovereign wealth funds and corporate investors around the world among its shareholders, and if the three listing timelines spread apart, some breathing room for capital absorption emerges. But even netting all of that out, two facts remain immovable: the structural handicap that the position of "last of three mega-IPOs" imposes on OpenAI, and the question of whether markets truly have enough capital — which will be the ultimate test of this offering. The investors who buy latest, at the highest price, are the ones who may end up footing the bill for demand borrowed in advance — and it is precisely those seats that OpenAI's future shareholders are now preparing to take.
Future Developments ― When, What, and in What Order Things Will Move
Finally, let me organize what will happen from here, and when. On the procedural side, the SEC will return review comments on the confidential S-1 it has received, and OpenAI will enter a cycle of successive revisions. If the company decides to proceed with a listing, it must file a "public S-1" at least 15 days before the start of its roadshow — and it is only at that moment that the actual financial statements, segment information, risk factors, and voting rights structure will be exposed to full public scrutiny. This is the single most pivotal juncture at which the recurring themes in this article — revenue, losses, and governance — shift from "reported" to "disclosed."
On timing, multiple reports indicate that OpenAI is targeting a listing as early as September 2026 (autumn), and no later than the fourth quarter, with an IPO target market capitalization of $730 billion to $850 billion (approximately ¥117 trillion to ¥136 trillion) — in line with recent private-round valuations — while some analysts see a possibility of exceeding $1 trillion (approximately ¥160 trillion). However, OpenAI itself has stated that "timing is undecided" and that "remaining private is sometimes easier for the time being," meaning the autumn outlook is not a confirmed schedule. How the market prices the upcoming IPOs of Anthropic and SpaceX will serve as an ideal set of "live test shots" (comps — comparable benchmarks) for OpenAI, and depending on those results, the decision to push forward or wait will likely hinge on what that data reveals.
The specific "measurement points" investors should watch from this summer through autumn are, first, the filing of the public S-1 and its contents (in particular, the absolute scale of losses and the path to profitability, disclosure of circular transactions, and governance provisions reflecting control by the nonprofit foundation); second, whether the push toward becoming a "super app" and the goal of raising the enterprise revenue ratio to 50% are proceeding as planned; third, the sustainability of the capital circulation and compute investment linking Nvidia, Oracle, AMD, and SoftBank; fourth, the degree of market tolerance signaled by the post-IPO share prices of Anthropic and SpaceX; and fifth, the interest rate and macro environment. What Silicon Valley VCs see in this filing is not merely the IPO of a single company. The entire AI economy — built atop a massive capital circulation that critics have called "pre-borrowing against real demand" — is being subjected for the first time to the ruthless pricing mechanism of public markets. As the prologue to that reckoning, OpenAI's IPO filing deserves to be recorded as a landmark event.