Agent Adoption vs. Data Challenges

Dave reviews surveys showing wide AI adoption but persistent data silos and planning failures blocking production deployments.

Play episode from 04:54

chevron_right

Transcript

chevron_right

Transcript

Episode notes

Recent analysis from Goldman Sachs indicates that $700 billion in AI investment during 2025 resulted in no measurable U.S. GDP growth, with most AI equipment imports negating domestic benefits and 80% of surveyed firms reporting no productivity or employment improvements. This pattern suggests that AI-related spending has primarily shifted margins from enterprise IT budgets to a small number of infrastructure vendors rather than delivering distributed value. Internal concerns are rising, with 90% of IT leaders questioning AI’s return on investment, and 80% citing fragmented data as a primary challenge to measuring outcomes.

Further context reveals that agentic AI initiatives face operational headwinds: Gartner expects 40% of such projects to be cancelled by 2027, and S&P Global found nearly half are abandoned before production, most often due to inadequate planning and data foundations. Margin erosion is widespread, attributed to AI implementation costs, and attempts to scale AI agents into production remain limited by inference costs and insufficient infrastructure. Despite increased adoption efforts, sustainable value delivery from AI platforms remains elusive for most organizations.

Enterprise AI access is becoming increasingly concentrated. OpenAI’s partnership with consulting firms such as BCG, McKinsey, Accenture, and Capgemini consolidates control of the enterprise distribution layer, narrowing competitive opportunities for smaller providers. Meanwhile, Amazon’s 13-hour AWS outage, linked to the misconfiguration of an internal AI tool, underscores the liability ambiguity in agentic systems—where vendors may attribute autonomous actions to user error, complicating risk assignment. Additional updates from vendors such as Anthropic, Cloudflare, and New Relic address incremental technical capabilities, with a distinct focus on cost, operational governance, and policy enforcement.

The prevailing themes for MSPs and IT leaders are increased scrutiny of AI value, heightened exposure to cost and accountability risk, and the emergence of managed service opportunities around data governance, cost instrumentation, and liability management. With enterprise market channels consolidating and risk shifting toward service providers, integrating robust contractual definitions for autonomy, incident attribution, and financial boundaries is essential to limit harm and clarify responsibility before incidents occur.

Four things to know today

00:00 Goldman: $700B AI Spend Delivered Near-Zero U.S. GDP Growth in 2025

03:49 OpenAI Enlists BCG, McKinsey, Accenture to Distribute Enterprise AI Agents

06:44 Report: Amazon's Own Engineers Prefer Claude Over Its Mandated Internal Tools

08:56 AI Inference Costs Are Falling — But Governance Gaps Are Growing

This is the Business of Tech.

Supported by:

CometBackup

Small Biz Thoughts Community