Founded in 2015 to develop artificial intelligence for the benefit of humanity, OpenAI has become one of the leading forces behind a global technological transformation. From its earliest research to ChatGPT, from its partnership with Microsoft to its historic fundraising rounds, its trajectory reveals a profound evolution: that of a scientific laboratory becoming an industrial power, now confronting the demands of capital and the prospect of public markets.

OpenAI's history begins with an ambition that now seems almost paradoxical: to develop a technology capable of transforming the global economy without allowing its control to depend exclusively on the commercial interests of a handful of corporations. A decade later, the organization founded around that ambition stands at the center of an international industrial competition, mobilizes extraordinary amounts of investment, and is reportedly considering an initial public offering that could place it among the world's most valuable companies.

Between these two moments, artificial intelligence has undergone a profound transformation in public perception. Long associated with laboratories, universities, and specialized applications, it has become an everyday instrument for production, communication, programming, and analysis. OpenAI did not invent the discipline, but it played a decisive role in transforming scientific advances accumulated over several decades into technologies accessible to users around the world.

This achievement raises a question that extends well beyond the company itself: how can an organization created to promote broadly beneficial artificial intelligence preserve its mission when it becomes dependent on expensive infrastructure, powerful investors, and a technological competition whose demands continue to intensify? OpenAI

Before OpenAI, a Long Scientific History

Artificial intelligence did not emerge in Silicon Valley during the 2010s. Its foundations reach back to Alan Turing's work, early research on neural networks, and the 1956 Dartmouth conference, widely regarded as one of the discipline's founding moments.

For decades, progress alternated with periods of disappointment. Expert systems generated considerable expectations before encountering their limitations. Machine learning advanced through improvements in algorithms and the growing availability of data. Deep neural networks subsequently experienced a dramatic acceleration, supported by graphics processing units, new training methods, and increasingly large datasets.

AlexNet's victory in the 2012 ImageNet competition marked a major milestone. It demonstrated the ability of deep neural networks to substantially improve image recognition. Other laboratories contributed to this progress, notably Google DeepMind, whose AlphaGo system defeated champion Lee Sedol in 2016.

The Transformer architecture, introduced by Google researchers in 2017, opened another chapter in natural language processing. By improving the ability of models to capture relationships between elements in a sequence, it became one of the technical foundations of modern large language models.

OpenAI therefore emerged in the middle of a revolution already underway. Its contribution was not to discover the principles of modern artificial intelligence single-handedly, but to assemble researchers, computing resources, and a development strategy capable of accelerating their application at scale.

This distinction matters. Technological revolutions are rarely the work of a single organization. They develop through accumulated discoveries, sometimes reaching a turning point when one actor succeeds in transforming scientific potential into widespread use.

2015: An Organization Born of Concern

OpenAI was announced in December 2015 as a nonprofit research organization. Among the figures associated with its founding were Sam Altman, Elon Musk, Greg Brockman, Ilya Sutskever, Wojciech Zaremba, and John Schulman.

Its stated objective was to contribute to the development of artificial general intelligence, or AGI, capable of benefiting humanity as a whole.

AGI generally refers to artificial intelligence capable of performing a broad range of intellectual tasks with a versatility extending beyond specialized systems. Its precise definition, measurement criteria, and the conditions under which it could be considered achieved remain subjects of debate.

From the outset, OpenAI distinguished itself through its emphasis on safety and the distribution of technological benefits. Its founders were concerned that highly advanced artificial intelligence could concentrate considerable economic and strategic power in the hands of a small number of actors.

The decision to establish a nonprofit organization reflected that concern. It was intended to prioritize research and the public interest over immediate financial returns.

Yet a contradiction soon emerged. Developing the most advanced systems required scarce scientific talent, growing computing capacity, and financial resources that traditional research laboratories could not always mobilize.

In other words, the ambition to make artificial intelligence broadly beneficial increasingly required assembling resources whose concentration could itself become a source of power.

This tension would accompany OpenAI throughout its evolution.

2016–2018: The Laboratory Years

OpenAI's early years focused on fundamental research, reinforcement learning, and the development of environments in which intelligent systems could be trained and evaluated.

OpenAI Gym, launched in 2016, provided tools designed to facilitate reinforcement learning research. The organization also explored robotics, video games, and methods allowing artificial agents to learn through interaction with their environments.

These efforts did not yet produce anything resembling ChatGPT. They nevertheless helped establish the scientific and technical capabilities required for more ambitious developments.

In 2018, OpenAI introduced its first GPT model, based on the Transformer architecture. The underlying approach involved training a model on large quantities of text before adapting it to different language tasks.

This represented an important shift. Rather than designing a separate system for every problem, researchers sought to develop general-purpose models capable of transferring what they had learned to new applications.

The potential was considerable, although performance remained far below that of subsequent generations.

During the same period, OpenAI experienced its first internal transformations. Elon Musk left its board of directors in 2018. The disagreements that later emerged over the organization's strategy and mission would illustrate the difficulty of maintaining a shared vision as technological and financial stakes increased.

2019: The First Compromise with Capital

The year 2019 marked an institutional turning point.

OpenAI established a commercial structure designed to attract the capital required to develop increasingly powerful models. The original arrangement placed limits on investors' financial returns while preserving control under the nonprofit organization.

This hybrid architecture sought to reconcile two objectives: mobilizing private capital and preserving a public-interest mission.

Above all, it reflected a new economic reality. Frontier artificial intelligence no longer depended solely on scientific talent or algorithmic ingenuity. It required large-scale computing infrastructure, substantial energy capacity, and recurring investment.

That same year, Microsoft announced a $1 billion investment and entered into a strategic partnership with OpenAI.

The alliance fundamentally altered the organization's trajectory. Microsoft possessed financial, commercial, and computing capabilities that few companies could match. Its Azure platform provided infrastructure essential to training and deploying increasingly sophisticated models.

For Microsoft, the agreement offered access to technologies capable of transforming its software, cloud services, and competitive position against Google and Amazon.

For OpenAI, it provided privileged access to resources indispensable to its expansion.

But the partnership also introduced a strategic dependency. An organization founded partly to prevent excessive concentrations of technological power now relied on one of the world's largest corporations to finance and industrialize its research.

The paradox was not necessarily irreconcilable. It nevertheless demonstrated that scientific independence, industrial autonomy, and financial sovereignty are distinct concepts.

From GPT-2 to GPT-3: The Power of Scale

In February 2019, OpenAI introduced GPT-2, a model capable of generating text with greater coherence than earlier generations.

The organization initially adopted a gradual release strategy for the complete model, citing concerns about potentially malicious uses of automated text generation. This decision contributed to a debate that would become central to the industry: should the most advanced models be widely released, or should access be restricted in the name of safety?

In 2020, GPT-3 represented another major advance. With 175 billion parameters, it illustrated the possibilities created by increasing model size, training data, and computing power.

The model could write, summarize, translate, answer questions, and perform certain tasks based on natural-language instructions.

Its results remained imperfect. It could generate errors, fabricate information, or reproduce biases present in its training data. Yet a new possibility was becoming clear: a single system could perform numerous functions previously handled by specialized software.

OpenAI subsequently developed a distribution strategy based on application programming interfaces, or APIs. Rather than exclusively commercializing end-user applications, the company allowed other organizations to integrate its models into their own products.

This approach helped establish a distinctive business model: artificial intelligence as a service, with charges linked to the consumption of computing resources.

The technology was beginning to become a software infrastructure upon which other businesses could build.

2021–2022: Artificial Intelligence Leaves the Laboratory

The following period considerably expanded the range of applications.

With Codex, OpenAI explored code generation and programming assistance. Models became capable of interpreting natural-language instructions and producing computer code.

With DALL·E and subsequently DALL·E 2, the organization demonstrated that generative techniques could extend beyond text to create images from written descriptions.

These advances suggested that related technological approaches could transform several intellectual and creative activities.

Yet it was the launch of ChatGPT on November 30, 2022, that triggered the decisive public breakthrough.

The innovation did not lie exclusively in the capabilities of the underlying model. It also came from the conversational interface, its ease of use, and the opportunity for ordinary users to experiment directly with technology previously accessible mainly to researchers and developers.

Users no longer needed to master a programming language to interact with a sophisticated system. They simply had to formulate a question, instruction, or problem.

This apparent simplicity profoundly changed the relationship between humans and software.

For decades, users had learned to adapt themselves to computer interfaces. With conversational assistants, software began adapting much more directly to human language.

ChatGPT did not resolve the fundamental problems of reliability, reasoning, or verification. But it dramatically reduced barriers to adoption.

Generative artificial intelligence ceased to be primarily a research subject and became a global economic, cultural, and political phenomenon.

2023: Acceleration and Crisis

The year 2023 confirmed OpenAI's transformation into a major industrial actor.

Microsoft announced a new multiyear, multibillion-dollar investment. The integration of OpenAI technologies into Microsoft's products accelerated, while Google, Anthropic, Meta, and other companies intensified their own programs.

In March, OpenAI unveiled GPT-4, which substantially improved performance across numerous complex tasks and introduced new multimodal capabilities.

ChatGPT's commercial and public success simultaneously transformed corporate expectations. Generative artificial intelligence became a priority for executive management teams, investors, and governments.

But this expansion also exposed internal tensions.

In November 2023, OpenAI's board abruptly removed Sam Altman as chief executive, citing a loss of confidence in his communications with the board.

The decision triggered a major crisis. A substantial proportion of employees supported Altman, Microsoft intervened publicly amid the turmoil, and the organization's governance came under intense scrutiny.

Within days, Altman returned as chief executive, while the board's composition was reorganized.

The episode went beyond a personal dispute. It exposed the difficulties of a structure in which a body responsible for preserving a nonprofit mission was expected to oversee a commercial enterprise that had become strategically important to employees, partners, and investors.

The fundamental issue concerned effective power. Who could determine the direction of a company when its economic value, contractual obligations, and scientific ambitions had become deeply interconnected?

The 2023 crisis demonstrated that formal legal control mechanisms do not always guarantee an uncontested ability to make decisions.

2024: From Conversation to Multimodal Systems

In 2024, OpenAI continued developing models capable of processing multiple forms of information.

GPT-4o improved interactions combining text, images, and audio. Voice exchanges became more natural, while perception and generation capabilities expanded.

Sora, introduced that year, illustrated progress in video generation. The automated creation of moving images from text instructions broadened commercial possibilities but also raised questions about copyright, content authenticity, and disinformation.

At the same time, OpenAI introduced its o1 family of models, designed to devote additional computation to certain reasoning processes before generating an answer.

This development was significant. Progress no longer depended exclusively on increasing the resources devoted to training. It also depended on how computing power was used when solving a problem.

A new economic dimension emerged: performance could be improved by allocating more computation during inference, when models were actually being used.

The company now had to optimize two particularly expensive activities: developing new generations of models and serving growing user demand.

As artificial intelligence became more widely adopted, operating costs grew as strategically important as scientific performance.

2025: Artificial Intelligence Becomes Industrial Infrastructure

The year 2025 marked another stage in OpenAI's development.

The company continued improving its models, notably with GPT-5, while expanding its activities in programming, professional tools, and systems capable of carrying out multistep tasks.

OpenAI was no longer seeking merely to provide answers. It was progressively developing tools capable of participating in entire workflows.

This direction brought artificial intelligence closer to the foundational infrastructure of the digital economy.

Applications extended across software development, document research, data analysis, customer service, content creation, education, and certain scientific uses.

Yet broader capabilities did not automatically guarantee profitability. Corporate customers had to measure productivity gains, manage risks, and adapt their organizations. Model providers, meanwhile, had to finance infrastructure whose economic returns depended on the intensity and duration of its use.

Energy became an increasingly important consideration. Data centers required substantial electricity supplies, cooling systems, specialized components, and suitable transmission networks.

Competition in artificial intelligence was therefore also becoming a competition for access to capital, energy, semiconductors, and physical infrastructure.

Against this backdrop, OpenAI announced a $40 billion fundraising round led by SoftBank in 2025, at a reported valuation of $300 billion.

That same year, its governance underwent a decisive transformation.

In October, the organization's restructuring was completed. The nonprofit entity became the OpenAI Foundation, while the commercial business adopted the form of OpenAI Group PBC, a public benefit corporation under US law.

The foundation retained control rights over the commercial enterprise, while the new structure was intended to facilitate access to capital and simplify shareholders' economic rights.

This evolution institutionalized the compromise initiated in 2019. OpenAI did not formally abandon its founding mission, but it adapted its organization to the requirements of a company whose financing needs had reached an entirely different scale.

Preserving that mission now depended less on nonprofit status alone than on the effectiveness of governance mechanisms, the transparency of decisions, and the ability of oversight bodies to withstand economic pressure.

2026: Financing on an Unprecedented Scale

In 2026, OpenAI crossed another threshold.

In March, the company announced the closing of a $122 billion fundraising round, based on a post-money valuation of approximately $852 billion.

These figures illustrated the exceptional position OpenAI had acquired in financial expectations surrounding artificial intelligence.

They should not, however, be interpreted as a direct measure of profitability or fundamental economic value. A private valuation reflects transaction terms, the rights attached to securities, and investors' expectations. It is neither a continuously observable market capitalization nor a guarantee of future value.

The commercial development of ChatGPT and professional services strengthened revenue prospects. Yet business expansion was accompanied by substantial commitments to computing, infrastructure, and research.

OpenAI's financial strategy increasingly depended on a complex balance: financing the capacity required for expansion quickly enough while demonstrating that revenue growth would eventually absorb the associated costs.

Competition made this balance more difficult.

Anthropic developed its own models and strengthened its position among enterprise customers. Google benefited from vertical integration, infrastructure, and established commercial activities. Meta pursued a different strategy around open-weight models, while other American, European, and Asian companies sought to establish positions in particular market segments.

Competition was therefore no longer limited to the quality of model-generated answers. It also concerned inference costs, reliability, safety, commercial distribution, software integration, and customer retention.

In this industry, possessing the most capable model at a particular moment did not guarantee a lasting advantage.

October 2026: When Capital Begins to Negotiate

Reports published on October 31 by the Financial Times and relayed by Reuters indicated that OpenAI had reportedly reduced the target for a new fundraising round from approximately $30 billion to $20 billion.

The Wall Street Journal separately reported that the company had obtained approximately $11 billion in investor commitments. The valuation under discussion was reportedly around $1.25 trillion, compared with an earlier ambition of $1.4 trillion.

These reports concerned financial discussions whose final terms had not been established. Reported commitments should not be treated as funds fully received, and no definitive conclusion about the company's ownership structure could be drawn from those figures alone.

A $20 billion fundraising round would remain exceptional. Nevertheless, the reported reduction in the original target was significant: it suggested that OpenAI's ability to attract investors, however considerable, was not immune to financial trade-offs.

The company that helped popularize generative artificial intelligence now had to convince financial markets not only of the technological value of its models, but also of the economic sustainability of their development.

The distinction was fundamental.

Demand for artificial intelligence services could continue to increase even as investors became more selective about entry valuations, margin prospects, and future financing requirements.

Similarly, rapid revenue growth would not necessarily translate into immediate improvements in cash flow if the investments required to sustain that growth increased at a comparable or faster pace.

The issue was therefore not whether artificial intelligence had an economic future. It was how much of the value created would ultimately accrue to model developers, infrastructure providers, semiconductor manufacturers, software distributors, and end users.

At this stage, the sustainable profitability of each level of the value chain remained a separate question.

Microsoft, Nvidia, and New Structures of Dependency

OpenAI's evolution also provides a way to understand the changing balance of power within the technology industry.

Microsoft played an essential role in financing and industrializing its models. Nvidia occupied a major position in supplying computing accelerators. Data center operators, electricity producers, and computing equipment manufacturers participated in an industrial value chain whose complexity extended far beyond software development.

These interdependencies changed the meaning of technological power.

A company could control highly advanced models without controlling all the resources necessary to operate them. Conversely, an infrastructure provider could benefit from the industry's expansion regardless of which model ultimately dominated the market.

Competitive advantage became multidimensional. It depended on research, but also on securing supplies, negotiating costs, distributing products, and financing successive investment cycles.

OpenAI stood at the center of this transformation without controlling all its components.

Its trajectory demonstrated that scientific innovation and industrial power were related but distinct realities. The former opened new possibilities. The latter often determined the conditions under which those possibilities became economically accessible.

The Promise of Artificial General Intelligence

Since its founding, OpenAI has maintained the development of artificial general intelligence as its stated long-term objective.

That ambition continues to shape its public discourse and scientific direction, but it raises several difficulties.

The first concerns the definition of AGI itself. Model capabilities continue to advance in programming, reasoning, mathematics, and tool use. Yet no universally accepted definition establishes an indisputable threshold at which an artificial intelligence system becomes general.

The second concerns safety. More autonomous and versatile systems could generate considerable benefits, but they could also introduce new risks of failure, manipulation, or malicious use.

The third concerns the distribution of economic benefits.

If artificial intelligence does improve productivity across numerous industries, those gains will not necessarily be distributed evenly. Their allocation will depend on market structures, public policies, access to technology, and organizations' ability to integrate it.

Finally, the concentration of resources required to develop frontier models could restrict the number of actors capable of participating at the technological frontier.

OpenAI's founding paradox thus reappears with renewed intensity. A technology designed to expand human capabilities could simultaneously reinforce economic concentration if its development remained restricted to organizations possessing extraordinary financial and industrial resources.

OpenAI and Its Competitors: From Technological Leadership to the Battle of Ecosystems

ChatGPT's success gave OpenAI an advantage that competitors could not immediately reproduce: it established an interface, a brand, and a new way of using computers in the public imagination. For several years, the company's name became closely associated with generative artificial intelligence itself, much as certain companies had previously come to embody the rise of search engines, smartphones, or e-commerce.

Yet brand recognition does not provide lasting protection against competition. As models improve and their performance converges on certain tasks, competition shifts toward dimensions that are more difficult to master: computing costs, reliability, integration into professional environments, commercial distribution, and the ability to convert adoption into recurring revenue.

OpenAI occupies a distinctive position in this transformation. It benefits from strong visibility among users, a developer ecosystem, and a growing enterprise presence. Its strategy seeks to establish ChatGPT as a general-purpose interface connecting information retrieval, content creation, programming, analysis, and task execution.

This ambition extends beyond conversational assistants. It increasingly reaches into functions traditionally performed by search engines, productivity software, and specialized applications.

Each of those markets, however, already has established leaders.

Google possesses a particularly powerful combination of assets: a global search business, Android, Chrome, Workspace, Google Cloud, computing infrastructure, and its own Gemini models. This integration allows it to distribute artificial intelligence through products already used at enormous scale. Where OpenAI must build or negotiate certain channels of access to users, Google can integrate AI directly into existing environments.

Microsoft presents a different situation. A historic partner, investor, and major distributor of OpenAI technologies, the company is also developing its own capabilities and retains strategic interests that do not entirely coincide with those of its partner. Its control of Windows, Microsoft 365, GitHub, and Azure gives it exceptional proximity to professional workflows. This relationship illustrates a defining feature of the industry: an essential partner can simultaneously become a competitor in particular market segments.

Anthropic represents another form of rivalry. Through Claude, the company seeks to differentiate itself through model performance, programming capabilities, enterprise applications, and its approach to safety. Its presence intensifies competition for corporate customers, researchers, infrastructure, and capital. It also demonstrates that OpenAI's early lead in consumer adoption does not guarantee equivalent dominance across every category of application.

Meta follows a distinct approach through its Llama family and a strategy centered on open-weight models. This can reduce certain barriers to adoption and allow companies to deploy systems within their own environments. It also places indirect pressure on business models built around proprietary access to the most advanced technologies.

Other competitors include xAI, specialized model providers, and Chinese companies such as DeepSeek and Alibaba with Qwen, alongside laboratories developing increasingly credible alternatives. Their progress illustrates that competition is not confined to a handful of American companies and that differences in costs, regulation, and market access can support several industrial trajectories.

The central question therefore becomes less about which company temporarily possesses the most capable model than about which will control durable economic relationships with users.

A model can be replaced. An interface that has become habitual, integrated into professional processes, and connected to an organization's data is more difficult to displace. Yet that advantage depends on trust, service quality, switching costs, and the availability of sufficiently competitive alternatives.

OpenAI is therefore seeking to build an ecosystem around its models, while its competitors use their existing ecosystems to distribute their own models.

This difference captures much of its strategic challenge.

Google, Microsoft, Amazon, and Meta possess established businesses capable of financing their investments in artificial intelligence. Despite its commercial expansion, OpenAI remains much more directly exposed to the economics of AI itself. Its growth depends on monetizing AI services, whereas several competitors can also benefit from integration with advertising, software, commerce, or cloud activities.

Cost structures reinforce this asymmetry. Developing frontier models requires substantial resources, but operating them at scale also involves continuous expenditure. Growth in user numbers does not automatically improve margins if applications simultaneously become more complex and computationally demanding.

OpenAI must therefore accomplish a delicate transformation: preserve its technological appeal while establishing competitive advantages that do not depend exclusively on the temporary superiority of its models.

The challenge is particularly important because artificial intelligence may evolve into an industry organized across several layers. Some companies will control infrastructure and semiconductors, others foundation models, and still others interfaces and specialized applications.

Economic value will not necessarily be distributed equally across these layers.

The company that did the most to popularize a technology will not automatically be the one that captures the largest share of its profits.

This is precisely the uncertainty investors must now incorporate when assessing OpenAI.

The IPO: The Next Test of Valuation

OpenAI's evolution could enter another phase through a potential initial public offering.

Following the nonprofit structure established in 2015, the creation of a commercial business in 2019, and the restructuring completed in 2025, access to public equity markets would represent another transformation in the company's relationship with capital.

Media reports suggesting preparations for such a possibility do not, however, constitute a definitive announcement, a confirmed timetable, or a guarantee that an offering will take place. An operation of this kind would depend on market conditions, governance, regulatory requirements, and decisions taken by the company and its shareholders.

Its significance would nevertheless extend beyond financing.

Until now, OpenAI's valuations have primarily been established through private transactions, in which investors negotiate particular terms and commit capital based on long-term growth expectations. Such valuations can be extraordinarily high, but they do not result from continuous trading between a broad population of buyers and sellers in an organized market.

A public listing would change that.

Share prices would become observable on a daily basis. Financial results would face regular scrutiny. Investors could compare revenue growth, margins, capital expenditure, and cash flows more directly with those of other publicly traded technology companies.

The question would no longer be simply how much capital OpenAI could attract, but at what price the market would be willing to finance its future development.

This distinction becomes particularly important in the context of valuations discussed in 2026.

Following a transaction announced in March at a post-money valuation of approximately $852 billion, October media reports referred to discussions around a potential valuation of $1.25 trillion. The Financial Times and Reuters also reported a reduction in the target for a new fundraising round, from $30 billion to approximately $20 billion.

These figures should not be confused with an observed public-market capitalization. They reflect announced private transactions or reported discussions, whose economic terms, security rights, and degree of completion may differ.

A public valuation exceeding $1 trillion would nevertheless place OpenAI in a category historically reserved for a very limited number of global corporations.

To sustain such a valuation, investors would need to assess not only its technological potential, but also its ability to convert that potential into economic returns.

Several questions would become central.

What proportion of revenue would come from individual subscriptions, enterprise customers, and developer services? To what extent could higher usage volumes reduce unit costs? Would investments in data centers and new model generations generate sufficient returns? How much future revenue would be exposed to price reductions resulting from competition?

These operational questions would be accompanied by governance considerations.

The structure established through the 2025 reorganization preserves a particular role for the foundation responsible for safeguarding the organization's mission. A public offering would require prospective shareholders to understand the relationship between that mission, control rights, and the economic interests attached to their securities.

Public markets can accommodate unusual governance structures. They nevertheless require investors to understand the rights they acquire, the risks they assume, and the mechanisms governing strategic decisions.

A potential OpenAI IPO would therefore be more than an exceptional financial transaction. It could become a moment of clarification for the entire artificial intelligence industry.

Since ChatGPT's launch, financial markets have assigned substantial value to the prospect of a profound economic transformation. Investments in semiconductors, cloud infrastructure, data centers, and foundation models have largely rested on expectations of sustained demand growth.

A public listing of OpenAI would provide an additional benchmark for assessing the economics of a major foundation-model developer, distinct from those of the diversified corporations currently financing much of this technological revolution.

It would not, by itself, provide a definitive measure of artificial intelligence's value. Markets can overestimate or underestimate a company's prospects, particularly when technologies and business models are evolving rapidly.

But it would make expectations more visible, comparisons more frequent, and financial judgments more immediate.

In 2015, OpenAI was founded around a question concerning the consequences of increasingly powerful artificial intelligence and how its benefits should be shared.

A decade later, the organization has become one of the principal symbols of a global industry whose expansion requires extraordinary financial, energy, and industrial resources.

It helped transform research into a product, the product into a platform, and the platform into a strategic asset.

A potential public listing would mark another transition: the moment when the ambitions of artificial intelligence would face the continuous, public scrutiny of capital markets and their expectations of financial returns.

OpenAI has played a major role in changing how the world evaluates the possibilities of technology. Financial markets may soon be called upon to evaluate, in turn, what that transformation is actually worth.