Altman announced OpenAI's "small goal": to invest 1 trillion in infrastructure within a year and achieve fully automated "AI researchers" by 2028

Through a new agreement with Microsoft, OpenAI has completed its controversial for-profit restructuring. CEO Sam Altman announced the completion of the restructuring during a live broadcast on Tuesday, the 28th, revealing that the company plans to significantly expand its infrastructure investment, with the ultimate goal of investing $1 trillion annually in infrastructure.

Altman disclosed that OpenAI has so far committed approximately $1.4 trillion for infrastructure development, equivalent to about 30 gigawatts (GW) of data center capacity. This total includes previously announced deals between OpenAI and partners such as AMD, Broadcom, NVIDIA, and Oracle.

Altman stated that this is just the starting point. In the long run, OpenAI hopes to establish a technology and financial system capable of adding capacity at a rate of 1 GW per week, with a cost of about $20 billion per GW. Altman admitted that this means OpenAI ultimately needs to achieve annual revenues in the hundreds of billions of dollars. He noted that OpenAI is on a steep growth curve toward this goal but still faces technical and financial obstacles.

Before Altman made these remarks, OpenAI had just finalized a significant agreement with Microsoft, facilitating the completion of its for-profit organizational restructuring. Under the new structure, OpenAI's for-profit division has now become the public benefit corporation OpenAI Group PBC, while the non-profit organization holding shares in this entity has been renamed the OpenAI Foundation, which holds approximately 26% of PBC's shares, currently valued at about $130 billion.

Infrastructure Investment Plan

In the live broadcast on Tuesday, Altman elaborated on OpenAI's infrastructure expansion plans. He stated that the committed $1.4 trillion in infrastructure investment will be deployed over the next few years, helping to clarify many of the previous announcements regarding partnerships with chip manufacturers, data centers, and financing.

However, Altman admitted that achieving the goal of adding 1 GW of capacity per week still presents challenges. "This is the target we want to reach, and we will do a lot of work in the coming months to see if we can achieve it," he said.

To support such massive capital expenditures, OpenAI needs to significantly increase its revenue scale. During the Q&A session, Altman indicated that enterprise customers will become an important source of revenue, but he also sees avenues for funding from the consumer side, not limited to monthly subscription fees. "Ultimately, we need to reach annual revenues in the hundreds of billions of dollars, and we are on a steep growth curve toward this goal," Altman added.

As for whether OpenAI plans to go public, Altman stated that an IPO is likely to happen eventually, but there are currently no specific plans or timelines. "Given our capital needs and the scale of the company, it can be said that this is our most likely path," he said.

AI Research Capability Timeline

In the same live broadcast, Altman also revealed OpenAI's aggressive timeline for AI research capabilities. He stated that based on internal progress tracking, OpenAI expects to achieve intern-level research assistants by September 2026 and fully automated "real AI researchers" by 2028OpenAI's Chief Scientist Jakub Pachocki described this AI researcher in a live broadcast as a "system capable of autonomously completing large research projects." Pachocki added, "We believe that deep learning systems may be less than ten years away from superintelligence." He defined superintelligence as a system that surpasses humans in a wide range of critical actions.

To achieve these goals, OpenAI is betting on two key strategies: continuous algorithm innovation and significantly expanding "compute during testing"—the time the model spends thinking about problems.

Pachocki stated that the current model can handle tasks with a time span of about five hours and has reached the level of top human performers in competitions such as the International Mathematical Olympiad. However, he believes this time span will quickly extend, partly by allowing the model to invest more computational resources to think through complex problems. For major scientific breakthroughs, he mentioned that it would be worthwhile to dedicate the entire data center's computing power to a single issue.

Restructuring Completed and New Agreement with Microsoft

OpenAI's for-profit restructuring has finally been completed after more than a year of negotiations. The company's for-profit division has now become the public benefit corporation OpenAI Group PBC, while the non-profit division has been renamed the OpenAI Foundation, which holds equity in the for-profit division currently valued at approximately $130 billion.

The non-profit foundation will begin operations with a $25 billion healthcare and disease-focused project and an "AI resilience" project. Once OpenAI's for-profit division reaches an undisclosed valuation milestone, the non-profit division will also receive "additional ownership."

This news comes some time after OpenAI received approval for the restructuring from the offices of the Attorneys General of California and Delaware. Without these regulatory approvals, OpenAI would not have been able to proceed with the restructuring. The restructuring also faced prolonged legal disputes with OpenAI co-founder and Tesla CEO Elon Musk, who has been attempting to block OpenAI's transformation through litigation.

As part of the restructuring, Microsoft has reached a new agreement with OpenAI. Under the new agreement, Microsoft's ownership stake appears to have decreased. Previously, Microsoft held a 32.5% convertible equity stake in OpenAI's for-profit organization. According to the new agreement, based on the diluted conversion basis including all owners, Microsoft will hold approximately 27% equity, which is valued at about $135 billion.

Commentators believe that the new agreement addresses one of the biggest sticking points in the relationship between OpenAI and Microsoft—Microsoft would lose rights to its technology after OpenAI achieves AGI (Artificial General Intelligence). This so-called "AGI clause" was previously somewhat vague, and the new agreement clarifies the situation. The new agreement between the two companies also covers OpenAI's upcoming consumer hardware plans and the next steps if AGI is ultimately realized