Zuckerberg’s AI bet: Keep models open, or risk letting China take the lead; 'personal superintelligence' to be made accessible to all

Meta CEO Mark Zuckerberg is making a new argument in the race for artificial intelligence: the US needs to keep AI models open and widely accessible if it wants to maintain its lead over China.

In a 6,500-word essay titled “The Future is for Everyone”, Zuckerberg laid out his vision for the next phase of AI, arguing that increasingly powerful systems should not be controlled by a small number of technology companies or governments. Instead, he wants AI models to be made available to developers and users, including through open-source or open-weight models.

He starts of by saying, "Still, it is surprising that the discourse from many developing AI is so filled with doom. I do not understand why anyone who believes that AI will eliminate most jobs and much of humanity’s relevance would rush to build that future. The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic."

The argument comes as Meta doubles down on its own AI ambitions. The company has released Muse Glimmer, an open-source model designed to run on personal computers, alongside its broader push towards what Zuckerberg calls “personal superintelligence”,  which are AI systems that can work continuously on behalf of individuals.

For Zuckerberg, the issue is not simply about giving consumers cheaper access to AI. It is also about where power sits as the technology becomes more capable. “Everyone will have free or affordable access to these tools," he writes. “For everyone to be part of the future, everyone must have the ability to use superintelligence to improve their lives and shape the world.”

Meta’s position puts it at odds with AI companies that favour keeping their most advanced models closed. Zuckerberg has argued that concentrating superintelligence inside a handful of companies or governments could create a different kind of risk: too much technological and economic power in too few hands.

But the argument has a distinctly geopolitical edge, as Zuckerberg believes the US should be careful about putting restrictions on open AI development, including policies around model training and distillation. His concern is that excessive restrictions could weaken American AI companies while Chinese developers continue to advance.

The underlying argument is if the US closes down parts of its AI ecosystem in the name of safety or regulation, China could gain an advantage by building and distributing models more aggressively. “Open source is a positive and important force for both empowering people and preventing centralization," he says.

For Meta, open models are also a way to build a broader ecosystem around its technology. The company has previously used its Llama models to encourage developers and businesses to build applications on top of them. Zuckerberg’s latest argument extends that strategy to the future of superintelligence.

There is, however, a contradiction at the heart of the pitch. Meta is arguing that AI should distribute power even as it is spending enormous sums to build the computing infrastructure needed to develop increasingly powerful systems. Meta expects capital expenditure of $125 billion to $145 billion in 2026, much of it driven by AI infrastructure.

"The jobs AI could create"

Zuckerberg’s argument is also notable because it does not treat AI’s impact on employment simply as a story of machines replacing people.

He believes superintelligence could create new forms of work by lowering the cost of starting companies and allowing much smaller teams to build products that would previously have required large organisations.

“If AI mostly automates what people do today, then there is a risk that it could reduce the number of jobs and economic opportunities...But if AI mostly empowers people to do more and create more, then it could lead to an explosion of new opportunities.”

That argument comes at an awkward time for Meta itself. In May, the company began cutting about 8,000 jobs, or roughly 10% of its workforce, as it reorganised around AI. Another 7,000 employees were shifted into AI-focused teams. The restructuring was explicitly tied to the company’s effort to become more efficient while redirecting resources towards AI.

Zuckerberg’s broader prediction is that AI could create opportunities for entrepreneurs, researchers and small businesses by giving individuals capabilities that previously required large teams and significant capital.

"There are several reasons to be optimistic that there will be an abundance of jobs in the future. No matter how intelligent AI becomes, there will always be a finite amount of compute and therefore an opportunity cost for how we use it. If people can use AI to invent incredibly valuable new things, then it will make more sense to allocate it towards that rather than automating existing jobs. The more superintelligence serves as a tool of invention, the more likely that individual capability outpaces automation and the future is better for people."

He also expects AI infrastructure itself to create jobs. Meta has said the US will need hundreds of thousands of skilled workers, including electricians, welders, plumbers and fibre technicians to build the infrastructure required for the AI economy.

“America is going to need hundreds of thousands of skilled tradespeople to build the infrastructure for our country to lead in AI.”

That creates a more complicated picture of the AI jobs debate. The same technology that allows companies such as Meta to operate with fewer employees could also create demand for new kinds of workers, from AI engineers and researchers to the electricians and construction workers needed to build data centres.