To me, ownership ought to mean something fairly concrete. Can I choose the model? Can I move my data elsewhere? Can I modify it? Can I keep using it if the vendor changes its mind? Can somebody else turn it off?
If the answer to most of those questions is no, what I have is access, not ownership.
This is where Porter’s criticism really bites. Zuckerberg says he wants to distribute intelligence. Porter asks whether having Zuckerberg distribute intelligence is the same thing as distributing power.
It isn’t necessarily.
The Internet offers a warning. It really did decentralize publishing and communication, but much of that activity eventually reconcentrated around a handful of enormous platforms. AI could follow the same path. The model may eventually sit on my laptop while the operating system, model store, updates and rules remain controlled by four or five companies.
We could achieve decentralization of computation without decentralization of power.
There is also a real danger going in the other direction. If powerful AI moves onto millions of privately owned machines, controlling malicious uses becomes much harder. A cloud provider can refuse a dangerous request. A model running independently on someone’s computer may not.
The tradeoff is real: distributing capability distributes both agency and risk. The policy question is not whether decentralization is perfectly safe. It is whether the risks of distributed capability are preferable to the risks of leaving powerful intelligence permanently under the control of a few institutions.
The problem is that both forms of concentration carry risks: concentrated capability can become concentrated control, while widely distributed capability can make harmful uses harder to contain.
Which brings me back to the giant data centers. I do not think they are going away. The International Energy Agency expects data-center electricity consumption to rise sharply through 2030.⁷ The largest centers may become busier than ever training frontier models, designing medicines, running scientific simulations and solving problems that genuinely require enormous computing power.
They may lose the email and gain the laboratory.
The more important change may happen at the other end of the wire. If powerful everyday intelligence moves onto the machines we own, we ought to decide what ownership actually means before somebody else decides for us.
Zuckerberg is right that AI becomes much more interesting if it empowers individuals rather than merely making giant corporations more powerful. Porter is right that handing everyone something called “personal superintelligence” does not mean the power behind it has actually become personal.
The PC revolution happened because computing moved from an institutional resource somebody else controlled to a machine ordinary people could own and use.
AI may be heading toward the same transition.
The most important question may not be how smart our personal AI becomes. It may be whether it actually belongs to us.
Bibliography
1. • Federal Trade Commission. “FTC Proposes Blanket Prohibition Preventing Facebook from Monetizing Youth Data,” May 3, 2023; and FTC staff report on social-media privacy and youth safeguards, September 19, 2024.
2. • Mark Zuckerberg. The Future Is for Everyone , Meta, August 10, 2026.
3. • OpenAI. “Building AI Infrastructure with the Effingham County Community,” July 22, 2026.
4. • Meta AI. “Introducing Llama 3.1: Our Most Capable Models to Date,” July 23, 2024.