Enterprise AI spend meets skepticism
HR and enterprise leaders are increasingly questioning the ROI of rising AI spending — companies must show productivity gains to justify new costs or face scrutiny. That pressure is shifting how engineering leaders must frame AI investments in reviews: outcomes, not adoption. (businessinsider.com)
Business Insider convened a 14-person roundtable titled "Futureproofing Your Workforce in the Age of AI" in Toronto on March 19, 2026, where 1Password CPO Katya Laviolette warned that "tokens, seats" and enterprise AI tool costs were not budgeted and are pricey. (businessinsider.com) Boston Consulting Group reported companies plan to double AI spending in 2026, targeting roughly 1.7% of revenue on AI and placing CEOs as primary decision-makers in 72% of cases. (bcg.com) Gartner’s March 4, 2026 HR survey found 45% of managers said AI lived up to expectations in improving team work, underscoring a mixed evidence base for productivity gains. (gartner.com) IBM’s executive-level research estimates only about 25% of AI initiatives deliver expected ROI and only 16% scale enterprise-wide, framing AI ROI as a leadership and measurement problem. (ibm.com) For briefing structure, use an "Outcome → Evidence → Ask" three-part slide: state the single business outcome being chased with a numeric baseline and timebound target, show attribution instrumentation and experiment design, then close with cost-to-date, projected savings/revenue and payback period—baseline measurement and a 12–24 month realization window are recommended for realistic ROI claims. (producttalk.org) Quantify engineering outcomes with a combined set of metrics: DORA delivery metrics (deployment frequency, lead time, change-failure rate, MTTR) for delivery health plus AI unit-economics like cost-per-inference, token consumption by feature, and % of AI spend that is inference vs. training. (dora.dev) Lock governance by integrating FinOps for AI—apply token-level cost allocation, show CFO-auditable unit economics, and make AI cost controls part of release gates as FinOps priority for 2026 has risen across cloud and AI spend programs. (finops.org)
