OpenAI and the path to IPO: What's happening exactly

According to reports from May 20, 2026, OpenAI is preparing a confidential S-1 filing with the SEC. It is working with banks Goldman Sachs and Morgan Stanley, which will lead the entire process. A confidential filing means the documents will not be immediately public — the SEC will first review them internally, send them back with comments, and only after the changes are incorporated will the prospectus be released publicly.

The timeline according to available information looks like this: May–June 2026 — confidential S-1 filing and SEC review begins; summer 2026 — incorporation of comments, preparation of roadshow for institutional investors; October–November 2026 — expected window for the IPO itself and the first day of trading. OpenAI is targeting a valuation above 1 trillion dollars with proceeds from the primary offering of around 60 billion dollars. For comparison: the last private funding round in March 2026 valued the company at 852 billion dollars.

A key verdict: Musk loses court case against Altman

A major obstacle to the IPO was removed on May 17, 2026, when a federal court in Oakland dismissed Elon Musk's lawsuit against OpenAI. The jury unanimously decided that Musk waited too long to file the lawsuit — a verdict that Judge Yvonne Gonzalez Rogers immediately accepted. Musk claimed that OpenAI violated its original nonprofit status and that Sam Altman and Greg Brockman acted fraudulently. But the court sided with OpenAI. This eliminated a significant legal risk factor that would have complicated the IPO.

In an ironic twist of fate, Musk's own company SpaceX (which merged with xAI this year) filed its IPO prospectus just a day earlier and is preparing the largest public offering in history with a valuation of 1.25 trillion dollars. The two former co-founders — Musk and Altman — are thus moving their rivalry from the courtroom straight to Wall Street.

Financial reality: 25 billion in revenue, but still in the red

OpenAI today generates approximately 2 billion dollars monthly, which represents an annualized run rate (ARR) of around 25 billion dollars. ChatGPT has over 900 million weekly active users and more than 50 million paying subscribers (ChatGPT Plus, Pro, and Team). Enterprise customers account for over 40% of revenue and their share continues to grow.

Despite massive revenues, however, OpenAI is still losing money. Estimates point to an annual loss of around 14 billion dollars for 2026. The main reasons are astronomical costs for computing infrastructure (GPU servers, data centers), salaries of top researchers (reportedly up to 10 million dollars as a hiring bonus), and training increasingly larger models like GPT-5.5.

CFO Sarah Friar has publicly admitted that the company needs to adopt "the discipline of a publicly traded company" — in other words, start controlling costs. Investors in the IPO prospectus will see detailed numbers for the very first time: gross margin after deducting computing costs, commitments to Microsoft (which holds approximately a 27% stake), and the company's governance structure.

The competition isn't sleeping: Anthropic and SpaceX are also going public

OpenAI isn't the only AI company heading to public markets this year. Anthropic — the creator of Claude and ChatGPT's main rival — is planning its own IPO for the end of 2026 with a target valuation of around 900 billion dollars. Unlike OpenAI, Anthropic is already reporting its first operating profit: for Q2 2026 it expects 559 million dollars in net profit on quarterly revenue of 10.9 billion. Moreover, Claude has surpassed OpenAI in many areas in the enterprise segment and developer tools (Claude Code).

For investors, this marks a pivotal moment: for the very first time, a public market for top-tier AI company stocks will emerge. Until now, only large institutional players could invest in OpenAI and Anthropic through private rounds. After the IPO, anyone with a brokerage account will be able to buy shares — including Czech investors through platforms like XTB, eToro, Degiro, or Interactive Brokers.

What this means for ChatGPT users and the AI market

For everyday ChatGPT users, nothing changes in the short term. The service remains available in both the free version and subscription (Plus at around $20/month, Pro at $200/month). In the long term, however, public market pressure for profitability could mean higher subscription prices, more advertising, or limitations on the free tier. OpenAI has already hinted that it plans for AI advertising revenue in the future — it estimates the AI advertising market will reach 100 billion dollars by 2030.

For European and Czech users, it's important to note that ChatGPT is fully available in Czech, and the quality of both translation and Czech text generation improves with each new version. Moreover, in April 2026, OpenAI adjusted its partnership with Microsoft so that it can more flexibly offer services on other cloud platforms (AWS, Google Cloud) — this could improve availability and latency for European customers as well.

The IPO also means regular quarterly reports in which OpenAI will have to publish detailed financial results. For the first time in history, we will see the actual economics of running a large language model: how much training costs, what the margins on the API are, and what the customer churn rate is. This transparency will help the entire market better understand whether the "AI as a service" business model is sustainable in the long term.

Microsoft, regulation, and risks

Microsoft owns approximately 27% of OpenAI and remains the main infrastructure partner through the Azure platform. The April update to the partnership agreement gave OpenAI more freedom — the company no longer has to run everything exclusively on Azure and can offer services through AWS or Google Cloud as well. This is important for European customers who prefer cloud data centers in the EU.

From a regulatory standpoint, OpenAI is under the EU's microscope. The AI Act, which came into force in 2024 and whose full effectiveness takes effect in 2026, classifies large language models as "general-purpose AI" with strict requirements for transparency, risk management, and copyright protection. For OpenAI, this means additional compliance costs — but at the same time, less risk of sudden regulatory interventions, because the rules of the game are finally clear.

Among the main risks for investors are: a narrowing technological lead (Google Gemini and Anthropic Claude are rapidly catching up to GPT-5.5 benchmarks), extreme capital intensity (just training a new generation of models can cost billions of dollars), and uncertainty over whether OpenAI can retain its user base in the face of growing competition. April Sensor Tower data showed a year-over-year increase in ChatGPT uninstalls of 132%, while Claude saw an elevenfold growth in downloads.