OpenAI: $7bn in shares repurchased from employees

OpenAI reportedly completed a $7 billion secondary sale

OpenAI reportedly completed a large secondary transaction allowing its employees to sell around $7 billion worth of shares, according to a report published by TechCrunch. The significance of this transaction is not limited to its amount: it reportedly does not correspond to a primary fundraising round, that is, an injection of new capital into the company’s accounts. Rather, it would provide a liquidity path to employees and, potentially, former staff members holding shares.

In the world of private technology companies, this type of transaction has become a major financial instrument. Employees frequently receive part of their compensation in the form of shares or options. Yet as long as a company remains private, those securities cannot be freely sold on a public market. A secondary offering, often called a tender offer in the US ecosystem, allows certain holders to convert a portion of that theoretical stake into cash.

The distinction is important in OpenAI’s case. A primary fundraising round generally increases the resources available to fund infrastructure, models, recruitment or research. A secondary sale, on the other hand, arranges a meeting between sellers of existing securities and buyers. The company may oversee the process, set eligibility conditions or the volumes that may be sold, but the funds do not necessarily become new operating capital.

The article’s headline, “OpenAI: $7bn in shares repurchased from employees,” thus reflects a reality that requires clarification: the information reported by TechCrunch concerns a secondary transaction intended for employees. Without complete public details on the legal structure of the transaction, it is advisable not to conclude that all shares were directly repurchased by OpenAI on its own balance sheet. In this type of transaction, the securities may be acquired by existing investors or new investors authorized to join the company’s capital, according to rules set by the company.

The $7 billion figure nevertheless places the transaction in a rare category. It reflects the scale OpenAI has reached since the launch of ChatGPT in November 2022, and the strategic value attributed to its teams. The company, founded in 2015 as a nonprofit organization, has become one of the central players in generative artificial intelligence. Its evolution has turned questions of governance, funding and compensation into topics followed far beyond Silicon Valley.

TechCrunch presents the offer as completed. But the available information does not make it possible to infer a precise share value, the number of securities sold, the list of participants or the identity of all buyers. Nor does it confirm a new listing timetable. Secondary arrangements are generally much less transparent than an initial public offering: the parameters are negotiated in a private setting and complete documents are not necessarily published.

This discretion does not lessen the importance of the signal. Such a transaction addresses three very concrete constraints for a company like OpenAI: giving employees a prospect of financial gain, limiting internal pressure for a rapid IPO, and preserving room for maneuver against competitors capable of offering extremely high recruitment packages. In cutting-edge AI, where a few hundred researchers, engineers and executives possess expertise that is difficult to replace, equity has become an element of industrial competition.

A liquidity mechanism that differs from traditional financing

To understand the scope of the transaction reported by TechCrunch, it is necessary to distinguish between the major categories of financing and circulation of securities. In a primary round, a company issues new shares. Investors provide funds that flow into the company, while the relative stake of existing shareholders may be diluted. In a secondary sale, the shares already exist: they change hands between current holders and buyers.

In a private company, this difference has immediate consequences for employees. A person may hold options or shares representing significant paper value without having a simple mechanism for selling them. Transfer restrictions, rights of first refusal and the absence of a public market generally prevent any spontaneous sale. An offering organized by the company provides a framework: a participation window, a set price or pricing mechanism, a maximum volume per participant, and compliance rules.

This partial liquidity does not mean an employee fully exits the company’s capital. Companies often choose to limit the quantity of securities that can be sold. They thus preserve the long-term incentive associated with share ownership, while preventing the earliest employees from being permanently locked into illiquid assets. The balance is delicate: too little liquidity can fuel frustration, but total liquidity too early can reduce the retention effect associated with holding securities.

For OpenAI, the need for such a mechanism comes within a phase of exceptional growth in the sector. In just a few years, generative models have gone from research demonstrations to products used by individuals, companies, developers and public administrations. Training and operating these systems nevertheless require considerable resources: computing power, data, software engineering, model safety, global deployment and user support.

OpenAI has a distinctive institutional trajectory. The organization was founded in 2015 by several tech figures, including Sam Altman and Elon Musk, with an initial ambition of artificial intelligence research. In 2019, OpenAI created a capped-profit structure, while the nonprofit entity remained at the top of its governance. This original structure has since been regularly examined through the lens of the organization’s stated mission, the scale of its financial needs and the influence of its partners.

Microsoft has become a major partner of OpenAI. In 2019, the group announced a $1 billion investment as well as a technology partnership. In 2023, Microsoft then communicated on the extension of this relationship as part of a multiyear, multibillion-dollar investment, without officially detailing all financial parameters. The use of Azure infrastructure has been a decisive element of this relationship, in a sector where access to data centers and specialized processors determines the ability to train large-scale models.

The secondary sale reported by TechCrunch adds to this landscape, but it should not be confused with these industrial agreements or with an investment intended to fund operations. It primarily represents a transaction involving existing equity. For investors, participating in such a transaction may provide access to a sought-after company without waiting for it to open its capital to the public. For selling employees, it realizes part of the value created. For OpenAI, it may help stabilize its shareholder base and teams without being subject to the disclosure obligations of a listed company.

The model also has limitations. Prices obtained in private markets do not benefit from the daily transparency of a stock exchange. They depend on a limited number of buyers and sellers, the availability of securities, the rights attached to the different classes of shares, and negotiated conditions. An implied valuation resulting from a secondary transaction should therefore not automatically be treated as a universal market price. It indicates a level of interest and a reference point, but does not provide the permanent liquidity of a listed security.

The amount mentioned by TechCrunch nevertheless confirms that OpenAI can organize a very large-scale transaction while remaining private. This is precisely what makes the event significant for the analysis of its financial maturity. For a long time, an IPO was the main route allowing employees and early investors to realize their gains. Large private technology groups now have alternative mechanisms: structured secondary offerings, controlled sales and access reserved for a selection of institutional investors.

The war for talent gives strategic value to employee shares

In artificial intelligence, the most experienced talent has become a strategic resource just like chips, data centers and high-quality data. Teams capable of training foundation models, designing alignment methods, optimizing inference systems or turning research prototypes into global products are few in number. Specialized companies, major digital platforms and research laboratories are therefore competing for profiles whose mobility can quickly alter the competitive balance.

A $7 billion secondary offering, if it is indeed structured as reported by TechCrunch, can act as a retention tool. It reminds employees that their securities can have concrete value before a potential listing. It can also reduce the immediate appeal of a competing offer accompanied by a signing bonus or a more generous equity package. Employees are no longer forced to wait indefinitely for a hypothetical liquidity event in order to benefit from a share of their employer’s growth.

This issue is particularly sensitive in a company whose image is closely tied to its researchers and executives. Generative AI is not an industry where physical assets are enough to guarantee a lasting advantage. Models can be replicated, technical approaches circulate in scientific literature, infrastructure is available to several major groups, and skills move. The ability to retain teams that know the systems, training processes and deployment constraints therefore becomes essential.

OpenAI is not alone in facing this pressure. Anthropic, founded by former OpenAI members, has established itself as an important competitor in language models intended for companies and developers. Google is developing its Gemini family, while Meta has invested in AI research for years and has made Llama models a visible part of its open strategy. Amazon, Microsoft, xAI and numerous young companies also participate in a competition playing out across models, products, computing and recruitment.

These comparisons do not mean that every player follows the same equity model. Google, Microsoft, Meta and Amazon are listed groups whose shares can be granted and resold on a public market. A private company such as OpenAI must create its own liquidity windows. This is a major difference in the design of compensation: an employee at a listed group can follow the price of their stock daily, whereas one at an unlisted company depends on occasional transactions or a future market event.

The size of the reported sale suggests that this dimension is now being handled at a strategic level. It is not merely a matter of responding to the wealth-management needs of a few executives or early employees. A multibillion-dollar transaction may involve a much broader group of eligible holders, even if the exact distribution is not public. The mechanism may be particularly important for employees who joined before the explosion in ChatGPT’s popularity, whose options may have acquired significant value without being easily monetizable.

This rationale also has an internal governance dimension. A team whose members have already obtained some liquidity may be less vulnerable to compensation-related tensions. Conversely, the conditions for accessing the offer matter greatly: if they are perceived as unequal or opaque, they may become a social and managerial issue. Private companies therefore generally seek to reconcile financial confidentiality, regulatory compliance and fairness across employee categories.

The OpenAI case is being watched particularly closely because the company sits at the intersection of a research laboratory, a software product provider and strategic AI infrastructure. Its teams do not merely produce commercial features: they also work on model capabilities that can transform entire segments of the digital economy. In this context, compensating and retaining people who possess this expertise is not a peripheral issue. It is a condition for continuity of the company’s technology roadmap.

For the market, the signal is clear: competition over salaries is not measured only by base pay or bonuses. It also involves access to capital, its perceived value, its liquidity and the company’s credibility in creating a financial event for its teams. A major secondary offering can thus become a competitive advantage, without appearing in the same format as a recruitment campaign or product announcement.

A milestone in financial maturity, without confirmation of an IPO

The transaction inevitably fuels speculation about a future OpenAI IPO. Yet a secondary sale is neither the announcement nor proof of an imminent IPO. It may, on the contrary, reduce pressure for a rapid listing. By providing access to liquidity to its employees and certain shareholders, the company gives itself more time to choose its pace, stabilize its governance and pursue its investments before submitting to the permanent demands of public markets.

An IPO entails a profound transformation. A listed company must regularly publish detailed financial information, respond to the expectations of public investors, explain its risks, disclose its dependencies and face continuous assessment of its results. For a leading AI company, these obligations can prove complex: computing costs are high, products evolve quickly, commercial agreements can be sensitive and competition unfolds over very short technology cycles.

OpenAI must also contend with a governance architecture unlike that of a traditional startup. The relationship between the nonprofit organization and the commercial entity is at the heart of debates over its evolution. Any path toward public markets should therefore be analyzed in light of this distinctive feature, without assuming that the company will mechanically follow the path of a traditional software platform. TechCrunch’s information on the secondary sale provides neither a timetable nor terms concerning a possible listing.

Secondary sales serve precisely as a bridge between two worlds: that of the high-growth private company and that of the liquidity sought by security holders. They can be repeated at regular intervals, with different parameters, without leading to an IPO in the short term. They can also attract investors wishing to increase their exposure to the company before a potential public market. But this possibility should not be turned into certainty: nothing in the reported information makes it possible to set a date or state that a decision has been made.

The $7 billion amount is nevertheless indicative of financial execution capability generally associated with the most mature private companies. Setting up a transaction of this size requires organizing shareholder rights, compliance procedures, the circulation of sensitive information and the sometimes divergent interests of participants. It is a more complex operation than a simple individual sale of securities.

It also reflects the interest of investors able to mobilize substantial amounts to acquire shares in a private company. This interest does not guarantee future performance or OpenAI’s lasting commercial success. It nevertheless indicates that demand for exposure to the company’s equity remains strong enough to support a large-scale secondary transaction, according to the account published by TechCrunch.

In the AI sector, financial maturity does not mean that spending is declining. On the contrary, developing more powerful models and deploying them at very large scale may require continuous investment. Needs in infrastructure, energy, computing capacity and engineering make capital allocation particularly important. A company may therefore seek primary funding for its operations while simultaneously organizing secondary sales for its employees. The two mechanisms serve different purposes and can coexist.

For observers, the main lesson is therefore less the promise of an IPO than the evolution of OpenAI’s status. The company is no longer assessed solely on the quality of its research demonstrations or the immediate success of a conversational product. It is considered an organization whose capital, talent, infrastructure and governance are the subject of transactions and expectations on the scale of the largest private technology groups.

Implications for Europe and the French-speaking AI market

From the perspective of France and Europe, the transaction reported by TechCrunch illustrates a persistent gap between the funding mechanisms available to major US AI companies and those available to a large share of European startups. Europe has recognized laboratories, high-level engineers, software companies and players specializing in models, optimization or vertical applications. But secondary transactions worth several billion dollars remain exceptional in the local ecosystem.

This gap does not concern only the amounts invested. It also affects compensation tools. For growing French companies, granting shares or options is an important lever for attracting international technical talent. But the value of these instruments depends on the credibility of the liquidity path. An acquisition, a listing or an organized secondary sale can give concrete meaning to this deferred compensation. The OpenAI example shows just how much access to this type of mechanism can strengthen a company’s appeal.

The French market is directly concerned by OpenAI’s ability to retain its teams, because its technologies are integrated into the thinking of many companies, developers and public organizations. Generative models raise issues in France relating to competitiveness, sovereignty, data protection, copyright, training and the organization of work. Changes in the equity structure of such an influential provider are therefore not merely US financial news: they contribute to the stability, strategy and innovation capacity of a player whose tools are used internationally.

For European companies that are customers or partners of AI providers, an actor’s financial maturity can have a dual interpretation. On the one hand, the ability to offer liquidity to employees may help reduce the risk of mass departures and promote continuity among the teams developing products. On the other hand, the concentration of capital and skills around a handful of private US companies raises the question of technological dependency. The more these players accumulate talent, infrastructure and capital, the harder it becomes for regional competitors to offer alternatives at the same scale.

The European Union nevertheless has its own framework. The European regulation on artificial intelligence, the AI Act, adds graduated obligations according to uses and risks. For providers of general-purpose AI models, debates over transparency, copyright compliance, technical documentation and risk assessment are becoming increasingly important. These rules do not directly determine the structure of a secondary sale, but they influence the environment in which providers operate and the compliance costs associated with their expansion in Europe.

France is simultaneously seeking to strengthen its computing capabilities, public research and specialized companies. In this context, the compensation conditions of researchers and engineers are a concrete issue. Academic laboratories and European companies must compete with groups able to offer not only high salaries, but also equity stakes that may become liquid before an IPO. OpenAI’s transaction is a reminder that global competition for skills is not limited to research budgets: it is played out in the complete architecture of financial incentives.

A simplistic interpretation should nevertheless be avoided. A large secondary transaction does not, by itself, resolve difficulties related to retention, governance or system safety. Nor does it guarantee that a business model will be sustainably profitable. It is one tool among others, alongside the quality of research, team autonomy, computing capacity, commercial contracts and user trust.

The long-term perspective is therefore shaped by a paradox. OpenAI can use private markets to offer substantial liquidity to its employees while remaining outside public markets. This flexibility strengthens its recruiting power and gives it time to define its next financial step. But it also increases attention to its governance, capital needs and influence in a sector that is gradually shaping the digital tools used in Europe.

If secondary offerings become a regular mechanism among the leading AI laboratories, they could durably change how talent evaluates their careers. The alternative would no longer simply be between joining a risky startup or a stable listed company, but between several private companies capable of offering large-scale liquidity prospects. For France and Europe, the challenge will be to create an environment in which local companies can fund research, retain their teams and offer credible value-creation paths, without abandoning the requirements of transparency, competition and technological sovereignty.