AI isn't replacing jobs. AI spending is (www.fastcompany.com)

Major companies from Amazon to UPS and Target have recently cut thousands of jobs—and many leaders attribute those cuts to AI. But multiple data points suggest a different story: generative-AI pilots largely aren’t delivering productivity gains, revenues from AI remain tiny compared with the infrastructure bill, and firms are under financial strain from massive AI-related capital spending. Studies find 95% of generative-AI pilots fail (MIT Media Lab), 96% of businesses report no dramatic efficiency gains (Atlassian), and surveys show employees routinely spend ~2 hours fixing “AI slop.” The St. Louis Fed finds only a weak correlation between theoretical AI exposure and adoption across occupations. Meanwhile, Amazon’s planned CapEx rose from $54B in 2023 to $84B in 2024 (projected $118B in 2025); industry-wide AI infrastructure spending may approach $1T in 2025 while AI service revenues hover under $30B. For the AI/ML community this matters because the current wave is funding-heavy but revenue-light: sellers of chips and cloud (e.g., Nvidia) are reaping enormous valuations while many AI buyers and startups face deep losses (OpenAI projects large cumulative losses). The mismatch incentivizes cost-cutting—layoffs are an easy lever—rather than reflective productivity displacement by models. The lack of transparent revenue/usage reporting from key firms complicates assessment, risks misallocating talent and capital, and could slow sustainable, revenue-generating AI adoption unless projects focus more on measurable ROI and realistic deployment costs.