OpenAI CEO Sam Altman is arguing for a broader conversation about who gets the gains from advanced AI. In a new essay on his personal blog, he said OpenAI is open to a “compute budget” and other “strange-sounding” ideas meant to “enable everyone on Earth to use a lot of AI.”
The central issue is not whether AI can become more capable. Altman’s warning is that capability alone does not guarantee that the benefits will reach people evenly, especially as the technology begins to affect work, capital, safety, and personal autonomy.
The distribution problem
Altman’s essay starts from a familiar promise of technological progress: over long periods, many important measures can improve. He specifically pointed to “health outcomes, economic prosperity, etc.” as examples of metrics that tend to get better on average and over the long term.
But he separated that broad progress from equality. In his view, “increasing equality does not seem technologically determined,” which means better tools do not automatically produce a fairer society. That is why he says new ideas may be needed.
The clearest concern is the relationship between capital and labor. Altman wrote that “the balance of power between capital and labor could easily get messed up,” and that this may require “early intervention.”
That concern is not presented as a distant theoretical issue. The source article notes that AI is already affecting the labor market through job cuts and departmental downsizing. Experts have also warned that mass unemployment is a possible outcome if the rise of AI is not matched with the right government policies and reskilling and upskilling programs.
What a compute budget is trying to solve
Altman’s “compute budget” concept is offered as one possible answer to unequal access. The article does not lay out a full operational plan, and it makes clear that such ideas may be easier to propose than to execute.
Still, the direction is clear. If powerful AI requires access to compute, then access to compute becomes part of access to opportunity. A compute budget would aim to make sure that more people can use meaningful amounts of AI rather than leaving the most capable systems only to those with the most resources.
That matters because Altman also describes AI progress as expensive at the frontier. He observed that “you can spend arbitrary amounts of money and get continuous and predictable gains” in AI performance. The article links that point to reports that OpenAI is in talks to raise up to $40 billion in a funding round and has pledged to spend up to $500 billion with partners on an enormous data network.
At the same time, Altman argues that access may improve as systems become cheaper to use. He says the cost to use “a given level of AI” falls about 10x every 12 months. In plain terms, pushing the leading edge may remain costly, while users may still get access to stronger tools over time.
AGI, scale and supervision
Altman again says artificial general intelligence is near. He defines AGI as “[an AI] system that can tackle increasingly complex problems, at human level, in many fields.”
But he does not describe AGI as flawless or fully independent. He warns that it may “require lots of human supervision and direction.” He also wrote that “[AGI systems] will not have the biggest new ideas,” and that they will be strong in some areas while “surprisingly bad at others.”
The article says Altman sees the real value of AGI coming from scale. Similar to Anthropic CEO Dario Amodei, he imagines thousands or even millions of highly capable AI systems working across “every field of knowledge work.”
That framing matters because it shifts the question from a single model to a large deployment pattern. If AI systems become useful because they can be run in large numbers, then the economic and social effects could also arrive through scale: many tasks, many fields, and many institutions adopting systems at once.
Safety, openness and power
Altman also says OpenAI will likely make “some major decisions and limitations related to AGI safety that will be unpopular.” That statement sits alongside questions about OpenAI’s own structure and incentives.
The article notes that OpenAI once pledged to stop competing with and start assisting any “value-aligned,” “safety-conscious” project that came close to building AGI first. That pledge came when OpenAI intended to remain a nonprofit. The company is now in the process of converting its corporate structure into a more traditional, profit-driven organization.
The same section notes that OpenAI reportedly aims to reach $100 billi on in revenue by 2029, a figure the source compares to Target and Nestlé’s current annual sales. It also says Microsoft and OpenAI reportedly had a contractual definition of AGI involving AI systems that can generate $100 billion in profits, which would allow OpenAI to negotiate more favorable investment terms once met.
Altman added in a footnote that OpenAI does not expect to end its relationship with Microsoft by using the term AGI. He said OpenAI “fully expect[s] to be partnered with Microsoft for the long term.”
The bigger trade-off
Altman’s stated goal is for OpenAI to “trend more towards individual empowerment” while preventing “AI being used by authoritarian governments to control their population through mass surveillance and loss of autonomy.”
He also acknowledged that OpenAI may need to give people more control than it has historically, including by open-sourcing more. The article notes that Altman recently said he thinks OpenAI has been on the wrong side of history on open-sourcing its technologies, even though OpenAI has released some technology openly in the past while generally favoring a proprietary, closed-source approach.
The tension is direct: more openness can support individual empowerment, while safety concerns can justify limits. Altman’s essay does not remove that conflict. It names it as a trade-off that OpenAI and the broader AI field will have to navigate as AI spreads through the economy and society.
His broader claim is that AI will become embedded everywhere. As he put it, “AI will seep into all areas of the economy and society; we will expect everything to be smart.” The question raised by the essay is whether that future is shaped only by capability and capital, or whether access, labor, safety and autonomy are addressed early enough to change how the benefits are shared.