In partnership with
AI Mandate Scan
The enterprise AI ROI problem isn't going away, it's getting more expensive. A new Bain survey of 951 global companies, published June 1, found that 37% of enterprises targeted 11–20% in cost savings from AI and automation programs. Of those who measured actual outcomes, roughly 40% landed below 10%. The gap isn't a technology failure. Bain's own conclusion is that "the technology worked”, though the value didn't follow.
The most commonly cited barrier: data access and integration, at 41% of respondents, above budget constraints, skills gaps, and executive buy-in. And yet 90% of organizations surveyed are increasing their AI budgets, largely to fund the next wave of agentic AI, which are ironically funded by savings from the prior wave that never fully materialized. Bain called it "a circular bet with a structural leak." For government agencies now being pushed to scale faster under NSPM-11's 120-day procurement clock, that pattern deserves attention before the contracts are signed.
This week, three policy stories that may define how the government navigates that pressure.
What's Inside
The Great American AI Act — Congress's first serious attempt at comprehensive federal AI governance, and why the 3-year preemption clause is already generating heat
Government Equity in AI — The White House in preliminary talks to take stakes in AI companies, with Sam Altman's fingerprints all over the concept
NSPM-11 — Trump's new national security memorandum directs the military and IC to accelerate AI adoption — and has a clause aimed directly at Anthropic
Quick Hits — GSA's EOA playbook, the Tech Force hiring gap, the Anthropic/Pentagon appeals court divide, and CISA's new AI cybersecurity mandate
Let’s get into it.
