As of mid-2026, Artificial Intelligence is no longer a futuristic promise but the invisible engine of the global economy. At the heart of this metamorphosis lies an entity whose name has become synonymous with disruption: OpenAI. Its journey, worthy of a technological epic, is one of immense ambition that eventually sacrificed its original ideals at the altar of computing power and commercial dominance.
The Genesis: A fortress against monopolies (2015–2018)
The adventure began in December 2015. The tech landscape was then dominated by giants like Google, which had just demonstrated its prowess with DeepMind. It was in this context that Sam Altman, Elon Musk, and several renowned researchers announced the creation of OpenAI. Originally, the structure was a non-profit organisation with an initial capital of $1 billion, sourced from private donors.
Their mission was rooted in a radical ethic: to develop Artificial General Intelligence (AGI) that was safe and, above all, whose benefits would be distributed equitably. The watchword was Open Source. OpenAI published its research, shared its code, and stood as a counter-power against the risk of intelligence being privatised by Big Tech. During these early years, the San Francisco lab shone with breakthroughs in reinforcement learning, most notably with OpenAI Five, which proved capable of defeating world champions in the video game Dota 2.
The strategic pivot: The price of scale (2019–2022)
The year 2019 marked a sharp and controversial turning point. OpenAI’s leadership realised that the quest for AGI required phenomenal computing power, costing billions of dollars in servers and electricity. Consequently, the non-profit created a “capped-profit” subsidiary. This change in status allowed the company to accept massive investments.
This is where Microsoft entered the fray with an initial $1 billion investment, securing exclusive access to OpenAI’s technologies for its Azure cloud platform. The era of total openness came to an end. In 2020, GPT-3 was launched, but unlike its predecessors, its code remained closed. OpenAI became an increasingly mysterious “black box,” yet an incredibly high-performing one. The lab was no longer just researching; it began producing tools that shook the world, culminating in the media explosion of ChatGPT at the end of 2022.
Acceleration and crises: A clash of cultures (2023–2024)
By 2023, OpenAI was no longer a startup; it was a global centre of gravity. The release of GPT-4 proved that the company was several laps ahead of the competition. However, this frantic growth created unbearable internal tensions. The world watched, stunned, during the “putsch” of November 2023: the board of directors dismissed Sam Altman, fearing he was moving too fast and sacrificing human safety for profit.
This five-day psychodrama ended with Altman’s triumphant return, backed by almost the entire workforce and Microsoft. This victory marked the final defeat of the early “idealists.” A period of frenetic innovation followed: Sora revolutionised video production, and multimodal AI capabilities (sight, hearing, speech) became the industry standard. OpenAI moved further away from its non-profit roots to become a commercial juggernaut, all while maintaining that this was the only path to AGI.
Related: The end of OpenAI’s Sora: Is AI video already has-been?
2025–2026: The agentic era and the AGI horizon
Today, in 2026, OpenAI has reached a new milestone: the era of Agentic AI. We are no longer in the age of simply asking questions to a chatbot. With the release of GPT-5 and its subsequent iterations, AI has become an agent capable of autonomous action across the web and operating systems. It books your flights, codes entire applications from a simple sketch, and manages corporate logistics without human intervention.
The company has also undergone a major structural shift this year. To prepare for its historic Initial Public Offering (IPO), it restructured into a public benefit corporation, completely removing the profit cap for its initial investors. Its valuation now exceeds $1 trillion.
Crucially, OpenAI claims to have reached what it calls “escape velocity” towards AGI. Their latest models demonstrate logical reasoning and complex mathematical problem-solving capabilities that equal, and in some cases surpass, the best human experts in almost every cognitive domain.
Related: OpenAI and GPT-5.3 Instant becomes less preachy and more efficient
A questionable legacy
From the idealistic association of 2015 to the hegemonic multinational of 2026, OpenAI has succeeded in its technical gamble: it has changed the world. But the debate remains fierce: has humanity truly benefited from the trade-off? While the tool has become universal, the original promise of Open Source has been buried in favour of centralised power.
The history of OpenAI is a fascinating transition from the utopia of shared knowledge to the reality of an industrial revolution driven by silicon. The next chapter, that of daily cohabitation with a superior level of intelligence, has only just begun.
