Techmeme
Dell Technologies added 1,000 customers for its AI Factory product line in the past quarter, bringing total clients to 5,000, up from 4,000 in February. The AI Factory bundles Nvidia chips, software, and services for enterprise AI workloads, with customers including Eli Lilly, Honeywell, and Samsung using the platform for drug discovery and AI-optimized semiconductor manufacturing. This signals sustained enterprise demand for turnkey AI infrastructure solutions among large organizations building internal capacity.
Azeem Azhar, Exponential View
A TrueFoundry survey of 200+ enterprise leaders managing live AI deployments reveals a fundamental mismatch between visible and actual AI costs. Token consumption has grown 17,000x over four years driven by agentic AI workloads, yet inference costs account for only 15-20% of total AI production spend—the remaining 80% hidden in orchestration overhead, tool call chains, embedding generation, retry loops, and engineering effort. Frontier models now process far more tokens per output than single-turn queries due to invisible work like context re-reading and tool calls, with runs on identical tasks varying by up to 30x in total tokens, making cost management nearly impossible. Safety and governance add another 20-40% to total spend, with evaluation-as-judge approaches costing 18x more than purpose-built guardrail models. The research found 95% of respondents run AI agents in production, 41% see costs only post-usage with no real-time visibility, and frontier models themselves cannot reliably predict their own token consumption (prediction accuracy as low as r=0.04-0.38), creating a forecasting problem that compounds as organizations accelerate agentic deployments with full knowledge of their exposure.