Investors plowing money into AI companies are operating without reliable maps. That's the argument at the center of a March 12 analysis from The Economist, which contends that the tools used to price tech companies — revenue multiples, earnings forecasts, addressable market estimates — are poorly matched to a technology still actively rewriting what it can do.
The problem is sharpest at the infrastructure layer. Hyperscalers and AI labs have committed hundreds of billions of dollars to compute and model development. The returns, particularly from enterprise agent deployments and automation rollouts, remain thin and inconsistent. Productivity gains are real in isolated cases but haven't shown up in GDP data or produced the winner-take-most consolidation investors usually look for when betting on a platform shift.
The Economist also flags how fragmented the AI stack makes analysis genuinely hard. Foundation model providers, cloud infrastructure vendors, application-layer startups, and vertical-specific deployments each face different competitive pressures and carry different margin profiles. That's not how prior technology waves worked. With the internet and mobile, value concentrated in identifiable platform layers — you could buy Google or Apple and capture a large share of the upside. AI's gains may prove more diffuse, flowing to enterprises and end-users in ways that don't show up cleanly in any single company's income statement.
For investors tracking the agent ecosystem, that diffusion argument is the one to watch. Agent platforms are among the <a href="/news/2026-03-14-george-hotz-ai-agent-hype-toxic">most-hyped bets</a> in the market right now, and also among the hardest to underwrite. Business models around usage, pricing, and retention are still unsettled. Enterprise buyers haven't published enough deployment data to build <a href="/news/2026-03-14-perplexity-launches-personal-computer-ai-agent-platform-for-enterprise">credible productivity cases</a>. Until that changes, The Economist's read holds: pricing AI remains closer to guesswork than analysis, and the asset class will keep attracting capital that has no clear basis for the valuation it's paying.