Access Is Not Capability: Six Researchers Read Zuckerberg's Letter Against Kenya's Grid

On August 10, Mark Zuckerberg published a letter called "The Future for Everyone," laying out how superintelligence could improve lives, expand human capability, and distribute the benefits of AI more broadly ^[1]. He argued that wider distribution of AI access across countries and companies is "the path to a positive AI future," and described what Meta is doing to make that happen ^[2].

Rest of World put the letter to six AI observers from around the world and asked whether AI can be the great equalizer it describes ^[3]. Some asked who gets to decide what "everyone" means; others said access does not always equal power ^[4]. The most concrete version of that second objection is a project, not an argument.

Chinasa T. Okolo, founder and scientific director of the policy incubator Technecultura and the inaugural policy lead at Black in AI, said AI infrastructure investments in African countries have yet to yield significant tangible benefits ^[5][6]. Her example: Kenya is pausing its Microsoft and G42-backed geothermal data center because of grid constraints ^[7]. She added that forthcoming efforts such as Cassava AI's Nvidia-powered AI factories are likely to worsen grid stability concerns in countries including South Africa and Nigeria ^[8]. That is the distinction operators should sit with. A distribution argument assumes the binding constraint is who gets handed the model. A paused geothermal build says the binding constraint arrives earlier, in megawatts, and it can stop a project that already has two well-capitalised backers behind it.

Okolo also went after the economics. She said Meta has grossly overstated the benefits of its data centers, in the manifesto and in advertising campaigns aimed at smaller rural towns that once hosted manufacturing or mining plants that have largely gone bust ^[9]. Labor benefits concentrate during construction, and once a facility is operational it employs only a small fraction of permanent workers ^[10]. There is a marginal effect on home prices that does not translate into local wages ^[11]. She said data centers also carry negative health and environmental impacts that will be harder to manage in African countries, which have fewer protections and where governments often lack the ability to enforce compliance where environmental laws exist at all ^[12].

The other respondents drew the same line through demand rather than supply. Sagar Vishnoi of Future Shift Labs said Zuckerberg's "everyone" largely excludes speakers of low-resource and tribal languages, shared-device households, and communities with intermittent connectivity ^[13][14]. His prescription is unglamorous: consent-driven data sets, community-led language digitization, and compute access for local researchers and developers, rather than bigger models ^[15]. Trust, he said, will depend on accountable enforcement, independent oversight, and protections that work outside Silicon Valley's institutional framework ^[16]. Mark Mulobi of Digital Impact said the personal-tutor use case appeals to him as a Kenyan, but that not everyone is online, digitally proficient, or able to engage with AI in English ^[17][18]. Without reliable electricity, high-speed internet, affordable devices, and investment in African languages, he said, AI risks widening existing divides ^[19].

Juan Ortiz-Freuler of the Non-Aligned Technologies Movement added a different pressure: Meta is large enough to run afoul of many interest groups, and those contradictions sharpen as Washington expects U.S. multinationals to help uphold American hegemony against their own profit motives ^[20][21].

Watch whether the Kenyan project restarts and on what grid terms, and whether the Cassava AI factories land in South Africa and Nigeria without the stability problems Okolo predicts ^[7][8]. Watch, too, whether the construction-heavy employment pattern she describes shows up in the towns Meta is advertising to ^[9][10].