Enterprise AI budgets keep climbing, yet the promised productivity boost remains elusive. In its latest quarter, Snowflake reported $1.04 billion in revenue, up 26 percent year over year, while NVIDIA’s data‑center business surged 69 percent year over year to $44.1 billion. Those numbers suggest wholesale adoption. In board meetings, however, it is still hard to name even one workflow that now runs faster, costs less or secures data better because of AI.
The root cause lies in spending priorities. Enterprises continue to pour billions into data‑lake migrations, multi‑year cloud contracts and sprawling vendor ecosystems under the assumption that progress starts with a capital‑intensive overhaul. Budgets swell, automation stalls and data scientists drown in governance checkpoints, while front-line teams are left wondering what changed. Analysts confirm the gap: roughly 85 percent of enterprise AI projects fail, and the share of companies abandoning their AI initiatives jumped to 42 percent last year. Architecture receives funding; outcomes do not, and legacy systems stay untouched.
The legacy trap
I have spent 15 years helping large enterprises deploy AI, and I have watched well‑funded programs collapse under yesterday’s playbook, replicating corporate stasis. One global bank I worked with illustrates the pattern. Determined to catalogue tens of thousands of “mission‑critical” data assets, it fielded an army of analysts to trace lineage, permissions and residency by hand. Months and millions of dollars later, barely a quarter of the estate was mapped, and fresh schema changes were already invalidating the work.
