The Dark Side of the AI Boom Is It’s Masking Weak Investment (www.bloomberg.com)

The AI boom has generated huge headlines and stock market gains, but that excitement is masking a broader weakness in real investment. Outside a handful of cloud and chip giants spending heavily on GPUs and data‑centre capacity, corporate capital spending, factory investment and broad R&D intensity have not accelerated commensurate with the hype. Much of the value creation is concentrated in market cap and software rollouts, not in widespread “capital deepening” (new machines, manufacturing lines or durable asset growth) that historically drives productivity gains. For the AI/ML community this matters because durable, inclusive progress depends on more than model releases and VC rounds. Technical constraints—concentrated compute (NVIDIA/TPU dominance), semiconductor supply chains, and uneven access to cloud resources—mean innovations can be gated by a few providers. Financial flows into high‑valuation startups and buybacks can crowd out long‑term investment in hardware, tooling, and open infrastructure. The implication: researchers and practitioners should prioritize compute efficiency, reproducibility, and deployment pathways that lower barriers; investors and policymakers should incentivize broader capital formation (manufacturing, R&D diffusion, workforce training) to turn AI’s promise into sustained productivity and competitive advantage.