The journey of the founders, the transformation of an altruistic mission into a technological superpower, and the new players shaping the future of AI, driven by invisible infrastructure: data centers.
OpenAI was founded in December 2015 as an ambitious experiment: to develop artificial general intelligence (AGI) safely and for the benefit of humanity. Its founders—among them Sam Altman, Ilya Sutskever, Greg Brockman, Wojciech Zaremba, and Elon Musk—proposed a nonprofit model, distancing itself from the secrecy and corporate control of big tech. However, the journey from those foundational ideals to becoming a tech giant has been as transformative as it has been controversial.
Today, OpenAI leads the development of advanced models such as GPT-4o, DALL·E, and ChatGPT, and has been at the forefront of one of the most relevant strategic alliances in the sector with Microsoft. Yet, behind the media hype and accelerated growth, the story of OpenAI is also one of an industry marked by ethical decisions, internal tensions, and a geopolitical race for algorithmic supremacy.
The Founders: Evolution and Breakup
Of the eleven original founders, only three remain connected to the company, while others have embarked on equally ambitious new projects:
- Sam Altman remains the public face of OpenAI, although in November 2023 he was briefly ousted before being reinstated under pressure from the board and the team.
- Ilya Sutskever, a key figure in technical development, left the organization in 2024 to found Safe Superintelligence Inc., dedicated exclusively to developing safe AGI.
- Greg Brockman, co-founder and president, is currently on leave.
- Elon Musk departed in 2018 due to conflicts with Tesla and has become a staunch critic of OpenAI’s commercial shift.
- Other notable names like John Schulman, Durk Kingma, or Pamela Vagata have also left, some to projects like Anthropic or Google DeepMind.
From Non-Profit to Capital Superpower
In 2019, OpenAI adopted a “capped-profit” structure that allowed it to receive private investment without completely abandoning its original mission. This hybrid model attracted Microsoft, which has invested over $10 billion and integrated OpenAI’s models into key products like Bing and Azure.
The launch of ChatGPT in November 2022 skyrocketed the company’s notoriety, allowing it to reach over $6.6 billion in funding and one of the highest valuations in the sector. However, this growth has also intensified debates about governance, transparency, and the impact of artificial intelligence on society.
The Invisible Infrastructure: Data Centers
Behind every advancement in AI lies critical infrastructure that often goes unnoticed: data centers. Training models like GPT-4o requires colossal computing power, which can only be sustained with high-density distributed architecture, advanced cooling, and low-latency connectivity.
David Carrero, co-founder of Stackscale (Aire Group)—a European provider of cloud and bare-metal infrastructure—emphasizes:
“The rise of AI is testing the global capacity of data centers. It’s not just about having more servers, but about providing efficient, secure, and sustainable environments that can handle complex workloads for weeks or months.”
Carrero insists that Europe must develop its own sovereign computing capacity if it wants to compete on equal footing with giants like the U.S. or China:
“AI needs data, energy, and control. If we do not have sovereignty over the infrastructure, we will not have sovereignty over the decisions algorithms make either.”
New Players on the AI Board
The departure of key figures from OpenAI has led to an explosion of new initiatives that shape a diverse and burgeoning ecosystem:
- Anthropic: founded by Dario and Daniela Amodei, former OpenAI employees. Its Claude model is already competing with GPT-4.
- xAI: Elon Musk’s project aims to create an “ultra-curious” AI and is integrated with products from X (formerly Twitter).
- Cohere: founded by ex-Google Brain, focused on the enterprise use of language models.
- Mistral AI: based in Paris, promotes an open-source approach, with its Mixtral model gaining traction in technical communities.
- Google DeepMind: after merging with Google Brain, it competes with Gemini against GPT.
- Meta AI: bets on openness with models like LLaMA, taking a radically different approach from OpenAI.
- Inflection AI: founded by Mustafa Suleyman, a former DeepMind employee, aiming to create hyper-realistic conversational assistants.
Conclusion: Power, Purpose, and Hardware
The story of OpenAI is also the story of the dilemma between purpose and power, between ethics and scalability. Behind every innovation, there is an underlying layer of cables, racks, clusters, and data centers that makes the algorithmic revolution possible.
The debate about the future of AI is not limited to software. As David Carrero concludes:
“There is no artificial intelligence without data centers. And there is no digital sovereignty without our own infrastructure. The next big battle is not just over models, but over who controls the hardware that runs them.”
The key question is no longer just which model will dominate the market, but who will train it, with what data, from which servers… and with what values.